Real time YOLO Benchmark
Model | Params (M) | FLOPS@640 (G) | mAP@640 | Input Size | Latency RK3588 (ms) |
---|---|---|---|---|---|
YOLOv5s | 7.2 | 16.5 | 37.4 | 416x416/640x640 | 17/21 |
YOLOv6s | 18.5 | 45.3 | 45.0 | 416x416/640x640 | 12/22 |
YOLOv6s_mbla | 11.6 | 29.8 | 47.0 | 416x416/640x640 | 28/50 |
YOLOv8s | 11.2 | 28.6 | 44.9 | 416x416/640x640 | 22/40 |
YOLOv10s | 7.2 | 21.6 | 46.3 | 416x416/640x640 | 32/71 |
DAMO-YOLO-S* | 16.3 | 39.8 | 47.7 | 416x416/640x640 | 12/23 |
Gold-YOLO-S | 21.5 | 46.0 | 46.4 | 416x416/640x640 | 28/33 |
NOTE: The Params, FLOPS and mAP are from the repos' README.
YOLOv10s
[*] ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Network Layer Information Table
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| ID OpType DataType Target InputShape OutputShape Cycles(DDR/NPU/Total) Time(us) MacUsage(%) WorkLoad(0/1/2) RW(KB) FullName
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| 1 InputOperator UINT8 CPU \ (1,3,640,640) 0/0/0 4 \ 0.0%/0.0%/0.0% 0 InputOperator:images
| 2 ConvExSwish UINT8 NPU (1,3,640,640),(32,3,3,3),(32) (1,32,320,320) 152549/921600/921600 2309 1.56/1.56/1.56 100.0%/0.0%/0.0% 1204 Conv:/model.0/conv/Conv
| 3 ConvExSwish INT8 NPU (1,32,320,320),(64,32,3,3),(64) (1,64,160,160) 166878/460800/460800 1349 14.23/14.23/14.23 100.0%/0.0%/0.0% 3218 Conv:/model.1/conv/Conv
| 4 ConvExSwish INT8 NPU (1,64,160,160),(64,64,1,1),(64) (1,64,160,160) 110981/204800/204800 1217 3.51/3.51/3.51 100.0%/0.0%/0.0% 1604 Conv:/model.2/cv1/conv/Conv
| 5 Split INT8 NPU (1,64,160,160),(2) (1,32,160,160),... 110825/0/110825 401 \ 100.0%/0.0%/0.0% 1600 Split:/model.2/Split
| 6 ConvExSwish INT8 NPU (1,32,160,160),(32,32,3,3),(32) (1,32,160,160) 55733/230400/230400 671 14.31/14.31/14.31 100.0%/0.0%/0.0% 809 Conv:/model.2/m.0/cv1/conv/Conv
| 7 ConvExSwish INT8 NPU (1,32,160,160),(32,32,3,3),(32) (1,32,160,160) 55733/230400/230400 674 14.24/14.24/14.24 100.0%/0.0%/0.0% 809 Conv:/model.2/m.0/cv2/conv/Conv
| 8 Add INT8 NPU (1,32,160,160),(1,32,160,160) (1,32,160,160) 83119/0/83119 177 \ 33.7%/33.1%/33.1% 1600 Add:/model.2/m.0/Add
| 9 Concat INT8 NPU (1,32,160,160),(1,32,160,160),... (1,96,160,160) 166237/0/166237 289 \ 33.7%/33.1%/33.1% 2400 Concat:/model.2/Concat
| 10 ConvExSwish INT8 NPU (1,96,160,160),(64,96,1,1),(64) (1,64,160,160) 138756/204800/204800 1278 5.01/5.01/5.01 100.0%/0.0%/0.0% 2406 Conv:/model.2/cv2/conv/Conv
| 11 ConvExSwish INT8 NPU (1,64,160,160),(128,64,3,3),(128) (1,128,80,80) 85647/460800/460800 811 23.67/23.67/23.67 100.0%/0.0%/0.0% 1673 Conv:/model.3/conv/Conv
| 12 ConvExSwish INT8 NPU (1,128,80,80),(128,128,1,1),(128) (1,128,80,80) 56001/102400/102400 677 6.30/6.30/6.30 100.0%/0.0%/0.0% 817 Conv:/model.4/cv1/conv/Conv
| 13 Split INT8 NPU (1,128,80,80),(2) (1,64,80,80),... 55413/0/55413 266 \ 100.0%/0.0%/0.0% 800 Split:/model.4/Split
| 14 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 434 22.12/22.12/22.12 100.0%/0.0%/0.0% 436 Conv:/model.4/m.0/cv1/conv/Conv
| 15 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 433 22.17/22.17/22.17 100.0%/0.0%/0.0% 436 Conv:/model.4/m.0/cv2/conv/Conv
| 16 Add INT8 NPU (1,64,80,80),(1,64,80,80) (1,64,80,80) 41560/0/41560 187 \ 33.7%/33.7%/32.5% 800 Add:/model.4/m.0/Add
| 17 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 435 22.07/22.07/22.07 100.0%/0.0%/0.0% 436 Conv:/model.4/m.1/cv1/conv/Conv
| 18 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 435 22.07/22.07/22.07 100.0%/0.0%/0.0% 436 Conv:/model.4/m.1/cv2/conv/Conv
| 19 Add INT8 NPU (1,64,80,80),(1,64,80,80) (1,64,80,80) 41560/0/41560 187 \ 33.7%/33.7%/32.5% 800 Add:/model.4/m.1/Add
| 20 Concat INT8 NPU (1,64,80,80),(1,64,80,80),... (1,256,80,80) 110825/0/110825 238 \ 33.7%/33.7%/32.5% 1600 Concat:/model.4/Concat
| 21 ConvExSwish INT8 NPU (1,256,80,80),(128,256,1,1),(128) (1,128,80,80) 84262/204800/204800 698 12.23/12.23/12.23 100.0%/0.0%/0.0% 1633 Conv:/model.4/cv2/conv/Conv
| 22 ConvExSwish INT8 NPU (1,128,80,80),(256,128,1,1),(256) (1,256,80,80) 84296/204800/204800 1182 7.22/7.22/7.22 100.0%/0.0%/0.0% 834 Conv:/model.5/cv1/conv/Conv
| 23 Conv INT8 NPU (1,256,80,80),(1,256,3,3),(256) (1,256,40,40) 69413/57600/69413 247 0.61/0.61/0.61 35.0%/32.5%/32.5% 1604 Conv:/model.5/cv2/conv/Conv
| 24 ConvExSwish INT8 NPU (1,256,40,40),(256,256,1,1),(256) (1,256,40,40) 29992/102400/102400 411 10.38/10.38/10.38 100.0%/0.0%/0.0% 466 Conv:/model.6/cv1/conv/Conv
| 25 Split INT8 NPU (1,256,40,40),(2) (1,128,40,40),... 27707/0/27707 202 \ 100.0%/0.0%/0.0% 400 Split:/model.6/Split
| 26 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 426 22.54/22.54/22.54 100.0%/0.0%/0.0% 345 Conv:/model.6/m.0/cv1/conv/Conv
| 27 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 426 22.54/22.54/22.54 100.0%/0.0%/0.0% 345 Conv:/model.6/m.0/cv2/conv/Conv
| 28 Add INT8 NPU (1,128,40,40),(1,128,40,40) (1,128,40,40) 20780/0/20780 167 \ 35.0%/32.5%/32.5% 400 Add:/model.6/m.0/Add
| 29 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 431 22.27/22.27/22.27 100.0%/0.0%/0.0% 345 Conv:/model.6/m.1/cv1/conv/Conv
| 30 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 424 22.64/22.64/22.64 100.0%/0.0%/0.0% 345 Conv:/model.6/m.1/cv2/conv/Conv
| 31 Add INT8 NPU (1,128,40,40),(1,128,40,40) (1,128,40,40) 20780/0/20780 167 \ 35.0%/32.5%/32.5% 400 Add:/model.6/m.1/Add
| 32 Concat INT8 NPU (1,128,40,40),(1,128,40,40),... (1,512,40,40) 55413/0/55413 217 \ 35.0%/32.5%/32.5% 800 Concat:/model.6/Concat
| 33 ConvExSwish INT8 NPU (1,512,40,40),(256,512,1,1),(256) (1,256,40,40) 46062/204800/204800 463 18.43/18.43/18.43 100.0%/0.0%/0.0% 930 Conv:/model.6/cv2/conv/Conv
| 34 ConvExSwish INT8 NPU (1,256,40,40),(512,256,1,1),(512) (1,512,40,40) 46131/204800/204800 664 12.85/12.85/12.85 100.0%/0.0%/0.0% 532 Conv:/model.7/cv1/conv/Conv
| 35 Conv INT8 NPU (1,512,40,40),(1,512,3,3),(512) (1,512,20,20) 34927/28800/34927 199 0.38/0.38/0.38 35.0%/35.0%/30.0% 808 Conv:/model.7/cv2/conv/Conv
| 36 ConvExSwish INT8 NPU (1,512,20,20),(512,512,1,1),(512) (1,512,20,20) 22858/102400/102400 276 15.46/15.46/15.46 100.0%/0.0%/0.0% 460 Conv:/model.8/cv1/conv/Conv
| 37 Split INT8 NPU (1,512,20,20),(2) (1,256,20,20),... 13854/0/13854 168 \ 100.0%/0.0%/0.0% 200 Split:/model.8/Split
| 38 ConvExSwish INT8 NPU (1,256,20,20),(1,256,3,3),(256) (1,256,20,20) 7074/14400/14400 205 0.18/0.18/0.18 100.0%/0.0%/0.0% 104 Conv:/model.8/m.0/cv1/cv1.0/conv/Conv
| 39 ConvExSwish INT8 NPU (1,256,20,20),(512,256,1,1),(512) (1,512,20,20) 14962/51200/51200 225 9.48/9.48/9.48 100.0%/0.0%/0.0% 232 Conv:/model.8/m.0/cv1/cv1.1/conv/Conv
| 40 ConvExSwish INT8 NPU (1,512,20,20),(1,512,7,7),(512) (1,512,20,20) 14841/156800/156800 344 1.19/1.19/1.19 100.0%/0.0%/0.0% 228 Conv:/model.8/m.0/cv1/cv1.2/conv/Conv
| 41 ConvExSwish INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 213 10.02/10.02/10.02 100.0%/0.0%/0.0% 330 Conv:/model.8/m.0/cv1/cv1.3/conv/Conv
| 42 ConvExSwish INT8 NPU (1,256,20,20),(1,256,3,3),(256) (1,256,20,20) 7074/14400/14400 205 0.18/0.18/0.18 100.0%/0.0%/0.0% 104 Conv:/model.8/m.0/cv1/cv1.4/conv/Conv
| 43 Add INT8 NPU (1,256,20,20),(1,256,20,20) (1,256,20,20) 10390/0/10390 155 \ 35.0%/35.0%/30.1% 200 Add:/model.8/m.0/Add
| 44 Concat INT8 NPU (1,256,20,20),(1,256,20,20),... (1,768,20,20) 20780/0/20780 171 \ 35.0%/35.0%/30.0% 300 Concat:/model.8/Concat
| 45 ConvExSwish INT8 NPU (1,768,20,20),(512,768,1,1),(512) (1,512,20,20) 30754/153600/153600 359 17.83/17.83/17.83 100.0%/0.0%/0.0% 688 Conv:/model.8/cv2/conv/Conv
| 46 ConvExSwish INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 213 10.02/10.02/10.02 100.0%/0.0%/0.0% 330 Conv:/model.9/cv1/conv/Conv
| 47 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 170 \ 100.0%/0.0%/0.0% 100 MaxPool:/model.9/m/MaxPool
| 48 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 127 \ 100.0%/0.0%/0.0% 100 MaxPool:/model.9/m_1/MaxPool
| 49 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 116 \ 100.0%/0.0%/0.0% 100 MaxPool:/model.9/m_2/MaxPool
| 50 Concat INT8 NPU (1,256,20,20),(1,256,20,20),... (1,1024,20,20) 27707/0/27707 128 \ 35.0%/35.0%/30.0% 400 Concat:/model.9/Concat
| 51 ConvExSwish INT8 NPU (1,1024,20,20),(512,1024,1,1),(512) (1,512,20,20) 38651/204800/204800 365 23.38/23.38/23.38 100.0%/0.0%/0.0% 916 Conv:/model.9/cv2/conv/Conv
| 52 ConvExSwish INT8 NPU (1,512,20,20),(512,512,1,1),(512) (1,512,20,20) 22858/102400/102400 222 19.22/19.22/19.22 100.0%/0.0%/0.0% 460 Conv:/model.10/cv1/conv/Conv
| 53 Split INT8 NPU (1,512,20,20),(2) (1,256,20,20),... 13854/0/13854 113 \ 100.0%/0.0%/0.0% 200 Split:/model.10/Split
| 54 Conv INT8 NPU (1,256,20,20),(512,256,1,1),(512) (1,512,20,20) 14962/51200/51200 124 17.20/17.20/17.20 35.0%/35.0%/30.0% 232 Conv:/model.10/attn/qkv/conv/Conv
| 55 Gather INT8 CPU (1,512,20,20),(512) (1,512,20,20) 0/0/0 1289 \ 0.0%/0.0%/0.0% 204 Gather:/model.10/attn/Reshape_output_0_shuffle_order_for_/model.10/attn/Split
| 56 Split INT8 NPU (1,512,20,20),(3) (1,128,20,20),... 13854/0/13854 122 \ 100.0%/0.0%/0.0% 200 Split:/model.10/attn/Split
| 57 Conv INT8 NPU (1,256,20,20),(1,256,3,3),(256) (1,256,20,20) 7074/14400/14400 109 0.34/0.34/0.34 35.0%/35.0%/30.1% 104 Conv:/model.10/attn/pe/conv/Conv
| 58 Reshape INT8 NPU (1,128,20,20),(4) (4,32,1,400) 3465/0/3465 0 \ 0.0%/0.0%/0.0% 50 Reshape:/model.10/attn/Transpose_output_0_rs_4exMatMul_/model.10/attn/MatMul_2reshape
| 59 Reshape INT8 NPU (1,128,20,20),(4) (4,32,1,400) 3465/0/3465 0 \ 0.0%/0.0%/0.0% 50 Reshape:/model.10/attn/Split_output_1_rs_4exMatMul_/model.10/attn/MatMul_2reshape
| 60 exMatMul INT8 NPU (4,32,1,400),(4,32,1,400) (4,400,1,400) 25109/0/25109 247 \ 100.0%/0.0%/0.0% 100 exMatMul:/model.10/attn/MatMul_output_0_4exMatMul_/model.10/attn/MatMul
| 61 Mul INT8 NPU (4,400,1,400),(1,1,1,1) (4,400,1,400) 43291/0/43291 244 \ 100.0%/0.0%/0.0% 625 Mul:/model.10/attn/Mul
| 62 Transpose INT8 NPU (4,400,1,400) (1,4,400,400) 43291/0/43291 496 \ 100.0%/0.0%/0.0% 625 Transpose:/model.10/attn/MatMul_output_0_rs_4exMatMul_/model.10/attn/MatMul_2reshape
| 63 exSoftmax13 INT8 CPU (1,4,400,400) (1,4,400,400) 0/0/0 8145 \ 0.0%/0.0%/0.0% 625 exSoftmax13:/model.10/attn/Softmax
| 64 Transpose INT8 NPU (1,4,400,400) (4,400,1,400) 108227/0/108227 2228 \ 100.0%/0.0%/0.0% 2500 Transpose:/model.10/attn/Transpose_1_output_0_rs_4exMatMul_/model.10/attn/MatMul_1
| 65 Reshape INT8 NPU (1,256,20,20),(4) (1,4,64,400) 17318/0/17318 633 \ 100.0%/0.0%/0.0% 100 Reshape:/model.10/attn/Split_output_2_rs/model.10/attn/Split
| 66 Transpose INT8 NPU (1,4,64,400) (4,400,1,64) 17317/0/17317 441 \ 100.0%/0.0%/0.0% 400 Transpose:/model.10/attn/Split_output_2_tp_4exMatMul_/model.10/attn/MatMul_1
| 67 exMatMul INT8 NPU (4,400,1,64),(4,400,1,400) (4,64,1,400) 28572/0/28572 454 \ 100.0%/0.0%/0.0% 725 exMatMul:/model.10/attn/MatMul_1_output_0_4exMatMul_/model.10/attn/MatMul_1
| 68 Reshape INT8 NPU (4,64,1,400),(4) (1,256,20,20) 6928/0/6928 0 \ 0.0%/0.0%/0.0% 100 Reshape:/model.10/attn/Reshape_1
| 69 Add INT8 NPU (1,256,20,20),(1,256,20,20) (1,256,20,20) 10390/0/10390 158 \ 35.0%/35.0%/30.1% 200 Add:/model.10/attn/Add
| 70 ConvAdd INT8 NPU (1,256,20,20),(256,256,1,1),(256),... (1,256,20,20) 12676/25600/25600 167 6.39/6.39/6.39 35.0%/35.0%/30.1% 266 Conv:/model.10/attn/proj/conv/Conv
| 71 ConvExSwish INT8 NPU (1,256,20,20),(512,256,1,1),(512) (1,512,20,20) 14962/51200/51200 268 7.96/7.96/7.96 100.0%/0.0%/0.0% 232 Conv:/model.10/ffn/ffn.0/conv/Conv
| 72 ConvAdd INT8 NPU (1,512,20,20),(256,512,1,1),(256),... (1,256,20,20) 18356/51200/51200 184 11.59/11.59/11.59 35.0%/35.0%/30.1% 430 Conv:/model.10/ffn/ffn.1/conv/Conv
| 73 Concat INT8 NPU (1,256,20,20),(1,256,20,20) (1,512,20,20) 13854/0/13854 161 \ 35.0%/35.0%/30.0% 200 Concat:/model.10/Concat
| 74 ConvExSwish INT8 NPU (1,512,20,20),(512,512,1,1),(512) (1,512,20,20) 22858/102400/102400 233 18.31/18.31/18.31 100.0%/0.0%/0.0% 460 Conv:/model.10/cv2/conv/Conv
| 75 Resize INT8 NPU (1,512,20,20),(0),(4) (1,512,40,40) 34634/0/34634 209 \ 100.0%/0.0%/0.0% 200 Resize:/model.11/Resize
| 76 Concat INT8 NPU (1,512,40,40),(1,256,40,40) (1,768,40,40) 83119/0/83119 200 \ 35.0%/32.5%/32.5% 1200 Concat:/model.12/Concat
| 77 ConvExSwish INT8 NPU (1,768,40,40),(256,768,1,1),(256) (1,256,40,40) 62131/307200/307200 632 20.25/20.25/20.25 100.0%/0.0%/0.0% 1394 Conv:/model.13/cv1/conv/Conv
| 78 Split INT8 NPU (1,256,40,40),(2) (1,128,40,40),... 27707/0/27707 200 \ 100.0%/0.0%/0.0% 400 Split:/model.13/Split
| 79 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 426 22.54/22.54/22.54 100.0%/0.0%/0.0% 345 Conv:/model.13/m.0/cv1/conv/Conv
| 80 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 426 22.54/22.54/22.54 100.0%/0.0%/0.0% 345 Conv:/model.13/m.0/cv2/conv/Conv
| 81 Concat INT8 NPU (1,128,40,40),(1,128,40,40),... (1,384,40,40) 41560/0/41560 198 \ 35.0%/32.5%/32.5% 600 Concat:/model.13/Concat
| 82 ConvExSwish INT8 NPU (1,384,40,40),(256,384,1,1),(256) (1,256,40,40) 38027/153600/153600 432 14.81/14.81/14.81 100.0%/0.0%/0.0% 698 Conv:/model.13/cv2/conv/Conv
| 83 Resize INT8 NPU (1,256,40,40),(0),(4) (1,256,80,80) 69266/0/69266 389 \ 100.0%/0.0%/0.0% 400 Resize:/model.14/Resize
| 84 Concat INT8 NPU (1,256,80,80),(1,128,80,80) (1,384,80,80) 166237/0/166237 355 \ 33.7%/33.7%/32.5% 2400 Concat:/model.15/Concat
| 85 ConvExSwish INT8 NPU (1,384,80,80),(128,384,1,1),(128) (1,128,80,80) 112522/307200/307200 857 14.94/14.94/14.94 100.0%/0.0%/0.0% 2449 Conv:/model.16/cv1/conv/Conv
| 86 Split INT8 NPU (1,128,80,80),(2) (1,64,80,80),... 55413/0/55413 266 \ 100.0%/0.0%/0.0% 800 Split:/model.16/Split
| 87 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 435 22.07/22.07/22.07 100.0%/0.0%/0.0% 436 Conv:/model.16/m.0/cv1/conv/Conv
| 88 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 434 22.12/22.12/22.12 100.0%/0.0%/0.0% 436 Conv:/model.16/m.0/cv2/conv/Conv
| 89 Concat INT8 NPU (1,64,80,80),(1,64,80,80),... (1,192,80,80) 83119/0/83119 248 \ 33.7%/33.7%/32.5% 1200 Concat:/model.16/Concat
| 90 ConvExSwish INT8 NPU (1,192,80,80),(128,192,1,1),(128) (1,128,80,80) 70132/153600/153600 715 8.95/8.95/8.95 100.0%/0.0%/0.0% 1225 Conv:/model.16/cv2/conv/Conv
| 91 ConvExSwish INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,40,40) 39655/230400/230400 487 19.71/19.71/19.71 100.0%/0.0%/0.0% 945 Conv:/model.17/conv/Conv
| 92 Concat INT8 NPU (1,128,40,40),(1,256,40,40) (1,384,40,40) 41560/0/41560 200 \ 35.0%/32.5%/32.5% 600 Concat:/model.18/Concat
| 93 ConvExSwish INT8 NPU (1,128,80,80),(64,128,3,3),(64) (1,64,80,80) 44071/460800/460800 741 25.91/25.91/25.91 100.0%/0.0%/0.0% 872 Conv:/model.23/one2one_cv2.0/one2one_cv2.0.0/conv/Conv
| 94 ConvExSwish INT8 NPU (1,128,80,80),(1,128,3,3),(128) (1,128,80,80) 55486/115200/115200 685 0.44/0.44/0.44 100.0%/0.0%/0.0% 802 Conv:/model.23/one2one_cv3.0/one2one_cv3.0.0/one2one_cv3.0.0.0/conv/Conv
| 95 ConvExSwish INT8 NPU (1,384,40,40),(256,384,1,1),(256) (1,256,40,40) 38027/153600/153600 431 14.85/14.85/14.85 100.0%/0.0%/0.0% 698 Conv:/model.19/cv1/conv/Conv
| 96 Split INT8 NPU (1,256,40,40),(2) (1,128,40,40),... 27707/0/27707 202 \ 100.0%/0.0%/0.0% 400 Split:/model.19/Split
| 97 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 436 22.02/22.02/22.02 100.0%/0.0%/0.0% 436 Conv:/model.23/one2one_cv2.0/one2one_cv2.0.1/conv/Conv
| 98 Conv INT8 NPU (1,64,80,80),(64,64,1,1),(64) (1,64,80,80) 27862/51200/51200 188 5.67/5.67/5.67 33.7%/33.7%/32.5% 404 Conv:/model.23/one2one_cv2.0/one2one_cv2.0.2/Conv
| 99 ConvExSwish INT8 NPU (1,128,80,80),(128,128,1,1),(128) (1,128,80,80) 56001/102400/102400 678 6.29/6.29/6.29 100.0%/0.0%/0.0% 817 Conv:/model.23/one2one_cv3.0/one2one_cv3.0.0/one2one_cv3.0.0.1/conv/Conv
| 100 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 426 22.54/22.54/22.54 100.0%/0.0%/0.0% 345 Conv:/model.19/m.0/cv1/conv/Conv
| 101 ConvExSwish INT8 NPU (1,128,80,80),(1,128,3,3),(128) (1,128,80,80) 55486/115200/115200 685 0.44/0.44/0.44 100.0%/0.0%/0.0% 802 Conv:/model.23/one2one_cv3.0/one2one_cv3.0.1/one2one_cv3.0.1.0/conv/Conv
| 102 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 425 22.59/22.59/22.59 100.0%/0.0%/0.0% 345 Conv:/model.19/m.0/cv2/conv/Conv
| 103 Concat INT8 NPU (1,128,40,40),(1,128,40,40),... (1,384,40,40) 41560/0/41560 197 \ 35.0%/32.5%/32.5% 600 Concat:/model.19/Concat
| 104 ConvExSwish INT8 NPU (1,128,80,80),(128,128,1,1),(128) (1,128,80,80) 56001/102400/102400 677 6.30/6.30/6.30 100.0%/0.0%/0.0% 817 Conv:/model.23/one2one_cv3.0/one2one_cv3.0.1/one2one_cv3.0.1.1/conv/Conv
| 105 Conv INT8 NPU (1,128,80,80),(80,128,1,1),(80) (1,80,80,80) 45395/76800/76800 172 15.50/15.50/15.50 33.7%/33.7%/32.5% 810 Conv:/model.23/one2one_cv3.0/one2one_cv3.0.2/Conv
| 106 Concat INT8 NPU (1,64,80,80),(1,80,80,80) (1,144,80,80) 62339/0/62339 165 \ 33.7%/33.7%/32.5% 900 Concat:/model.23/Concat
| 107 OutputOperator INT8 CPU (1,144,80,80) \ 0/0/0 40 \ 0.0%/0.0%/0.0% 900 OutputOperator:output0
| 108 ConvExSwish INT8 NPU (1,384,40,40),(256,384,1,1),(256) (1,256,40,40) 38027/153600/153600 387 16.54/16.54/16.54 100.0%/0.0%/0.0% 698 Conv:/model.19/cv2/conv/Conv
| 109 ConvExSwish INT8 NPU (1,256,40,40),(256,256,1,1),(256) (1,256,40,40) 29992/102400/102400 407 10.48/10.48/10.48 100.0%/0.0%/0.0% 466 Conv:/model.20/cv1/conv/Conv
| 110 Conv INT8 NPU (1,256,40,40),(1,256,3,3),(256) (1,256,20,20) 17464/14400/17464 173 0.22/0.22/0.22 35.0%/35.0%/30.1% 404 Conv:/model.20/cv2/conv/Conv
| 111 Concat INT8 NPU (1,256,20,20),(1,512,20,20) (1,768,20,20) 20780/0/20780 171 \ 35.0%/35.0%/30.0% 300 Concat:/model.21/Concat
| 112 ConvExSwish INT8 NPU (1,256,40,40),(64,256,3,3),(64) (1,64,40,40) 22321/230400/230400 443 21.67/21.67/21.67 100.0%/0.0%/0.0% 544 Conv:/model.23/one2one_cv2.1/one2one_cv2.1.0/conv/Conv
| 113 ConvExSwish INT8 NPU (1,256,40,40),(1,256,3,3),(256) (1,256,40,40) 27854/57600/57600 410 0.37/0.37/0.37 100.0%/0.0%/0.0% 404 Conv:/model.23/one2one_cv3.1/one2one_cv3.1.0/one2one_cv3.1.0.0/conv/Conv
| 114 ConvExSwish INT8 NPU (1,768,20,20),(512,768,1,1),(512) (1,512,20,20) 30754/153600/153600 359 17.83/17.83/17.83 100.0%/0.0%/0.0% 688 Conv:/model.22/cv1/conv/Conv
| 115 Split INT8 NPU (1,512,20,20),(2) (1,256,20,20),... 13854/0/13854 166 \ 100.0%/0.0%/0.0% 200 Split:/model.22/Split
| 116 ConvExSwish INT8 NPU (1,64,40,40),(64,64,3,3),(64) (1,64,40,40) 8191/57600/57600 212 11.32/11.32/11.32 100.0%/0.0%/0.0% 136 Conv:/model.23/one2one_cv2.1/one2one_cv2.1.1/conv/Conv
| 117 Conv INT8 NPU (1,64,40,40),(64,64,1,1),(64) (1,64,40,40) 7083/12800/12800 116 2.30/2.30/2.30 35.0%/32.5%/32.5% 104 Conv:/model.23/one2one_cv2.1/one2one_cv2.1.2/Conv
| 118 ConvExSwish INT8 NPU (1,256,40,40),(128,256,1,1),(128) (1,128,40,40) 21923/51200/51200 229 9.32/9.32/9.32 100.0%/0.0%/0.0% 433 Conv:/model.23/one2one_cv3.1/one2one_cv3.1.0/one2one_cv3.1.0.1/conv/Conv
| 119 ConvExSwish INT8 NPU (1,256,20,20),(1,256,3,3),(256) (1,256,20,20) 7074/14400/14400 151 0.25/0.25/0.25 100.0%/0.0%/0.0% 104 Conv:/model.22/m.0/cv1/cv1.0/conv/Conv
| 120 ConvExSwish INT8 NPU (1,128,40,40),(1,128,3,3),(128) (1,128,40,40) 13927/28800/28800 217 0.35/0.35/0.35 100.0%/0.0%/0.0% 202 Conv:/model.23/one2one_cv3.1/one2one_cv3.1.1/one2one_cv3.1.1.0/conv/Conv
| 121 ConvExSwish INT8 NPU (1,256,20,20),(512,256,1,1),(512) (1,512,20,20) 14962/51200/51200 266 8.02/8.02/8.02 100.0%/0.0%/0.0% 232 Conv:/model.22/m.0/cv1/cv1.1/conv/Conv
| 122 ConvExSwish INT8 NPU (1,128,40,40),(128,128,1,1),(128) (1,128,40,40) 14442/25600/25600 270 3.95/3.95/3.95 100.0%/0.0%/0.0% 217 Conv:/model.23/one2one_cv3.1/one2one_cv3.1.1/one2one_cv3.1.1.1/conv/Conv
| 123 Conv INT8 NPU (1,128,40,40),(80,128,1,1),(80) (1,80,40,40) 11628/19200/19200 166 4.02/4.02/4.02 35.0%/32.5%/32.5% 210 Conv:/model.23/one2one_cv3.1/one2one_cv3.1.2/Conv
| 124 Concat INT8 NPU (1,64,40,40),(1,80,40,40) (1,144,40,40) 15585/0/15585 163 \ 35.0%/32.5%/32.5% 225 Concat:/model.23/Concat_1
| 125 OutputOperator INT8 CPU (1,144,40,40) \ 0/0/0 14 \ 0.0%/0.0%/0.0% 225 OutputOperator:519
| 126 ConvExSwish INT8 NPU (1,512,20,20),(1,512,7,7),(512) (1,512,20,20) 14841/156800/156800 345 1.18/1.18/1.18 100.0%/0.0%/0.0% 228 Conv:/model.22/m.0/cv1/cv1.2/conv/Conv
| 127 ConvExSwish INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 213 10.02/10.02/10.02 100.0%/0.0%/0.0% 330 Conv:/model.22/m.0/cv1/cv1.3/conv/Conv
| 128 ConvExSwish INT8 NPU (1,256,20,20),(1,256,3,3),(256) (1,256,20,20) 7074/14400/14400 205 0.18/0.18/0.18 100.0%/0.0%/0.0% 104 Conv:/model.22/m.0/cv1/cv1.4/conv/Conv
| 129 Add INT8 NPU (1,256,20,20),(1,256,20,20) (1,256,20,20) 10390/0/10390 112 \ 35.0%/35.0%/30.1% 200 Add:/model.22/m.0/Add
| 130 Concat INT8 NPU (1,256,20,20),(1,256,20,20),... (1,768,20,20) 20780/0/20780 117 \ 35.0%/35.0%/30.0% 300 Concat:/model.22/Concat
| 131 ConvExSwish INT8 NPU (1,768,20,20),(512,768,1,1),(512) (1,512,20,20) 30754/153600/153600 305 20.98/20.98/20.98 100.0%/0.0%/0.0% 688 Conv:/model.22/cv2/conv/Conv
| 132 ConvExSwish INT8 NPU (1,512,20,20),(64,512,3,3),(64) (1,64,20,20) 17784/115200/115200 258 18.60/18.60/18.60 100.0%/0.0%/0.0% 488 Conv:/model.23/one2one_cv2.2/one2one_cv2.2.0/conv/Conv
| 133 ConvExSwish INT8 NPU (1,512,20,20),(1,512,3,3),(512) (1,512,20,20) 14148/28800/28800 220 0.34/0.34/0.34 100.0%/0.0%/0.0% 208 Conv:/model.23/one2one_cv3.2/one2one_cv3.2.0/one2one_cv3.2.0.0/conv/Conv
| 134 ConvExSwish INT8 NPU (1,64,20,20),(64,64,3,3),(64) (1,64,20,20) 2996/14400/14400 101 5.94/5.94/5.94 100.0%/0.0%/0.0% 61 Conv:/model.23/one2one_cv2.2/one2one_cv2.2.1/conv/Conv
| 135 Conv INT8 NPU (1,64,20,20),(64,64,1,1),(64) (1,64,20,20) 1888/3200/3200 98 0.68/0.68/0.68 34.9%/34.9%/30.2% 29 Conv:/model.23/one2one_cv2.2/one2one_cv2.2.2/Conv
| 136 ConvExSwish INT8 NPU (1,512,20,20),(128,512,1,1),(128) (1,128,20,20) 10910/25600/25600 127 8.40/8.40/8.40 100.0%/0.0%/0.0% 265 Conv:/model.23/one2one_cv3.2/one2one_cv3.2.0/one2one_cv3.2.0.1/conv/Conv
| 137 ConvExSwish INT8 NPU (1,128,20,20),(1,128,3,3),(128) (1,128,20,20) 3537/7200/7200 116 0.16/0.16/0.16 100.0%/0.0%/0.0% 52 Conv:/model.23/one2one_cv3.2/one2one_cv3.2.1/one2one_cv3.2.1.0/conv/Conv
| 138 ConvExSwish INT8 NPU (1,128,20,20),(128,128,1,1),(128) (1,128,20,20) 4053/6400/6400 181 1.47/1.47/1.47 100.0%/0.0%/0.0% 67 Conv:/model.23/one2one_cv3.2/one2one_cv3.2.1/one2one_cv3.2.1.1/conv/Conv
| 139 Conv INT8 NPU (1,128,20,20),(80,128,1,1),(80) (1,80,20,20) 3187/4800/4800 166 1.00/1.00/1.00 34.9%/34.9%/30.2% 60 Conv:/model.23/one2one_cv3.2/one2one_cv3.2.2/Conv
| 140 Concat INT8 NPU (1,64,20,20),(1,80,20,20) (1,144,20,20) 3897/0/3897 98 \ 34.9%/34.9%/30.1% 56 Concat:/model.23/Concat_2
| 141 OutputOperator INT8 CPU (1,144,20,20) \ 0/0/0 6 \ 0.0%/0.0%/0.0% 56 OutputOperator:540
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Total Operator Elapsed Per Frame Time(us): 58519
| Total Memory Read/Write Per Frame Size(KB): 82973.8
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| ---------------------------------------------------------------------------------------------------
| Operator Time Consuming Ranking Table
| ---------------------------------------------------------------------------------------------------
| OpType CallNumber CPUTime(us) GPUTime(us) NPUTime(us) TotalTime(us) TimeRatio(%)
| ---------------------------------------------------------------------------------------------------
| ConvExSwish 72 0 0 34426 34426 58.83%
| exSoftmax13 1 8145 0 0 8145 13.92%
| Concat 17 0 0 3316 3316 5.67%
| Transpose 3 0 0 3165 3165 5.41%
| Split 10 0 0 2106 2106 3.60%
| Conv 11 0 0 1758 1758 3.00%
| Add 8 0 0 1310 1310 2.24%
| Gather 1 1289 0 0 1289 2.20%
| exMatMul 2 0 0 701 701 1.20%
| Reshape 4 0 0 633 633 1.08%
| Resize 2 0 0 598 598 1.02%
| MaxPool 3 0 0 413 413 0.71%
| ConvAdd 2 0 0 351 351 0.60%
| Mul 1 0 0 244 244 0.42%
| OutputOperator 3 60 0 0 60 0.10%
| InputOperator 1 4 0 0 4 0.01%
| ---------------------------------------------------------------------------------------------------
YOLOv6s
[*] ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Network Layer Information Table
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| ID OpType DataType Target InputShape OutputShape Cycles(DDR/NPU/Total) Time(us) MacUsage(%) WorkLoad(0/1/2) RW(KB) FullName
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| 1 InputOperator UINT8 CPU \ (1,3,640,640) 0/0/0 4 \ 0.0%/0.0%/0.0% 0 InputOperator:images
| 2 ConvRelu UINT8 NPU (1,3,640,640),(32,3,3,3),(32) (1,32,320,320) 152549/921600/921600 601 5.99/5.99/5.99 33.4%/33.4%/33.1% 1204 Conv:/backbone/stem/rbr_reparam/Conv
| 3 ConvRelu INT8 NPU (1,32,320,320),(64,32,3,3),(64) (1,64,160,160) 166878/460800/460800 447 42.95/42.95/42.95 33.8%/33.1%/33.1% 3218 Conv:/backbone/ERBlock_2/ERBlock_2.0/rbr_reparam/Conv
| 4 ConvRelu INT8 NPU (1,64,160,160),(64,64,3,3),(64) (1,64,160,160) 112089/921600/921600 560 68.57/68.57/68.57 33.8%/33.1%/33.1% 1636 Conv:/backbone/ERBlock_2/ERBlock_2.1/conv1/rbr_reparam/Conv
| 5 ConvRelu INT8 NPU (1,64,160,160),(64,64,3,3),(64) (1,64,160,160) 112089/921600/921600 557 68.94/68.94/68.94 33.8%/33.1%/33.1% 1636 Conv:/backbone/ERBlock_2/ERBlock_2.1/block/block.0/rbr_reparam/Conv
| 6 ConvRelu INT8 NPU (1,64,160,160),(128,64,3,3),(128) (1,128,80,80) 85647/460800/460800 389 49.36/49.36/49.36 33.8%/33.8%/32.5% 1673 Conv:/backbone/ERBlock_3/ERBlock_3.0/rbr_reparam/Conv
| 7 ConvRelu INT8 NPU (1,64,160,160),(64,64,1,1),(64) (1,64,160,160) 110981/204800/204800 300 14.22/14.22/14.22 33.8%/33.1%/33.1% 1604 Conv:/neck/Bifusion1/cv2/block/conv/Conv
| 8 ConvRelu INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,80,80) 60434/921600/921600 565 67.96/67.96/67.96 33.8%/33.8%/32.5% 945 Conv:/backbone/ERBlock_3/ERBlock_3.1/conv1/rbr_reparam/Conv
| 9 ConvRelu INT8 NPU (1,64,160,160),(64,64,3,3),(64) (1,64,80,80) 70530/230400/230400 284 33.80/33.80/33.80 33.8%/33.8%/32.5% 1636 Conv:/neck/Bifusion1/downsample/block/conv/Conv
| 10 ConvRelu INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,80,80) 60434/921600/921600 565 67.96/67.96/67.96 33.8%/33.8%/32.5% 945 Conv:/backbone/ERBlock_3/ERBlock_3.1/block/block.0/rbr_reparam/Conv
| 11 ConvRelu INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,80,80) 60434/921600/921600 564 68.09/68.09/68.09 33.8%/33.8%/32.5% 945 Conv:/backbone/ERBlock_3/ERBlock_3.1/block/block.1/rbr_reparam/Conv
| 12 ConvRelu INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,80,80) 60434/921600/921600 563 68.21/68.21/68.21 33.8%/33.8%/32.5% 945 Conv:/backbone/ERBlock_3/ERBlock_3.1/block/block.2/rbr_reparam/Conv
| 13 ConvRelu INT8 NPU (1,128,80,80),(256,128,3,3),(256) (1,256,40,40) 51603/460800/460800 392 48.98/48.98/48.98 35.0%/32.5%/32.5% 1090 Conv:/backbone/ERBlock_4/ERBlock_4.0/rbr_reparam/Conv
| 14 ConvRelu INT8 NPU (1,128,80,80),(128,128,1,1),(128) (1,128,80,80) 56001/102400/102400 225 18.96/18.96/18.96 33.8%/33.8%/32.5% 817 Conv:/neck/Bifusion0/cv2/block/conv/Conv
| 15 ConvRelu INT8 NPU (1,128,80,80),(64,128,1,1),(64) (1,64,80,80) 41854/51200/51200 199 10.72/10.72/10.72 33.8%/33.8%/32.5% 808 Conv:/neck/Bifusion1/cv1/block/conv/Conv
| 16 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,40,40) 47724/921600/921600 617 62.24/62.24/62.24 37.5%/37.5%/25.0% 978 Conv:/backbone/ERBlock_4/ERBlock_4.1/conv1/rbr_reparam/Conv
| 17 ConvRelu INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,40,40) 39655/230400/230400 282 34.04/34.04/34.04 35.0%/32.5%/32.5% 945 Conv:/neck/Bifusion0/downsample/block/conv/Conv
| 18 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,40,40) 47724/921600/921600 622 61.74/61.74/61.74 37.5%/37.5%/25.0% 978 Conv:/backbone/ERBlock_4/ERBlock_4.1/block/block.0/rbr_reparam/Conv
| 19 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,40,40) 47724/921600/921600 616 62.34/62.34/62.34 37.5%/37.5%/25.0% 978 Conv:/backbone/ERBlock_4/ERBlock_4.1/block/block.1/rbr_reparam/Conv
| 20 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,40,40) 47724/921600/921600 617 62.24/62.24/62.24 37.5%/37.5%/25.0% 978 Conv:/backbone/ERBlock_4/ERBlock_4.1/block/block.2/rbr_reparam/Conv
| 21 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,40,40) 47724/921600/921600 615 62.44/62.44/62.44 37.5%/37.5%/25.0% 978 Conv:/backbone/ERBlock_4/ERBlock_4.1/block/block.3/rbr_reparam/Conv
| 22 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,40,40) 47724/921600/921600 617 62.24/62.24/62.24 37.5%/37.5%/25.0% 978 Conv:/backbone/ERBlock_4/ERBlock_4.1/block/block.4/rbr_reparam/Conv
| 23 ConvRelu INT8 NPU (1,256,40,40),(512,256,3,3),(512) (1,512,20,20) 60815/460800/460800 404 47.52/47.52/47.52 37.5%/31.2%/31.2% 1556 Conv:/backbone/ERBlock_5/ERBlock_5.0/rbr_reparam/Conv
| 24 ConvRelu INT8 NPU (1,256,40,40),(128,256,1,1),(128) (1,128,40,40) 21923/51200/51200 185 11.53/11.53/11.53 35.0%/32.5%/32.5% 433 Conv:/neck/Bifusion0/cv1/block/conv/Conv
| 25 ConvRelu INT8 NPU (1,512,20,20),(512,512,3,3),(512) (1,512,20,20) 93786/921600/921600 649 59.17/59.17/59.17 37.5%/31.2%/31.2% 2508 Conv:/backbone/ERBlock_5/ERBlock_5.1/conv1/rbr_reparam/Conv
| 26 ConvRelu INT8 NPU (1,512,20,20),(512,512,3,3),(512) (1,512,20,20) 93786/921600/921600 648 59.26/59.26/59.26 37.5%/31.2%/31.2% 2508 Conv:/backbone/ERBlock_5/ERBlock_5.1/block/block.0/rbr_reparam/Conv
| 27 ConvRelu INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 178 11.99/11.99/11.99 35.0%/35.0%/30.0% 330 Conv:/backbone/ERBlock_5/ERBlock_5.2/cspsppf/cv1/block/conv/Conv
| 28 ConvRelu INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 178 11.99/11.99/11.99 35.0%/35.0%/30.0% 330 Conv:/backbone/ERBlock_5/ERBlock_5.2/cspsppf/cv2/block/conv/Conv
| 29 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 234 41.03/41.03/41.03 37.5%/37.5%/25.0% 678 Conv:/backbone/ERBlock_5/ERBlock_5.2/cspsppf/cv3/block/conv/Conv
| 30 ConvRelu INT8 NPU (1,256,20,20),(256,256,1,1),(256) (1,256,20,20) 9213/25600/25600 166 6.43/6.43/6.43 35.0%/35.0%/30.1% 166 Conv:/backbone/ERBlock_5/ERBlock_5.2/cspsppf/cv4/block/conv/Conv
| 31 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 172 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/ERBlock_5/ERBlock_5.2/cspsppf/m/MaxPool
| 32 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 127 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/ERBlock_5/ERBlock_5.2/cspsppf/m_1/MaxPool
| 33 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 116 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/ERBlock_5/ERBlock_5.2/cspsppf/m_2/MaxPool
| 34 Concat INT8 NPU (1,256,20,20),(1,256,20,20),... (1,1024,20,20) 27707/0/27707 124 \ 35.0%/35.0%/30.0% 400 Concat:/backbone/ERBlock_5/ERBlock_5.2/cspsppf/Concat
| 35 ConvRelu INT8 NPU (1,1024,20,20),(256,1024,1,1),(256) (1,256,20,20) 26252/102400/102400 160 26.67/26.67/26.67 35.0%/35.0%/30.0% 658 Conv:/backbone/ERBlock_5/ERBlock_5.2/cspsppf/cv5/block/conv/Conv
| 36 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 215 44.65/44.65/44.65 37.4%/37.4%/25.1% 678 Conv:/backbone/ERBlock_5/ERBlock_5.2/cspsppf/cv6/block/conv/Conv
| 37 Concat INT8 NPU (1,256,20,20),(1,256,20,20) (1,512,20,20) 13854/0/13854 107 \ 35.0%/35.0%/30.0% 200 Concat:/backbone/ERBlock_5/ERBlock_5.2/cspsppf/Concat_1
| 38 ConvRelu INT8 NPU (1,512,20,20),(512,512,1,1),(512) (1,512,20,20) 22858/102400/102400 153 27.89/27.89/27.89 37.5%/31.2%/31.2% 460 Conv:/backbone/ERBlock_5/ERBlock_5.2/cspsppf/cv7/block/conv/Conv
| 39 ConvRelu INT8 NPU (1,512,20,20),(128,512,1,1),(128) (1,128,20,20) 10910/25600/25600 116 9.20/9.20/9.20 34.9%/34.9%/30.1% 265 Conv:/neck/reduce_layer0/block/conv/Conv
| 40 ConvTranspose INT8 NPU (1,128,20,20),(128,128,2,2),(128) (1,128,40,40) 10910/102400/102400 208 20.51/20.51/20.51 100.0%/0.0%/0.0% 115 ConvTranspose:/neck/Bifusion0/upsample/upsample_transpose/ConvTranspose
| 41 Concat INT8 NPU (1,128,40,40),(1,128,40,40),... (1,384,40,40) 41560/0/41560 143 \ 35.0%/32.5%/32.5% 600 Concat:/neck/Bifusion0/Concat
| 42 ConvRelu INT8 NPU (1,384,40,40),(128,384,1,1),(128) (1,128,40,40) 29404/76800/76800 199 16.08/16.08/16.08 35.0%/32.5%/32.5% 649 Conv:/neck/Bifusion0/cv3/block/conv/Conv
| 43 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 254 37.80/37.80/37.80 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_p4/conv1/rbr_reparam/Conv
| 44 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 209 45.93/45.93/45.93 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_p4/block/block.0/rbr_reparam/Conv
| 45 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 199 48.24/48.24/48.24 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_p4/block/block.1/rbr_reparam/Conv
| 46 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 251 38.25/38.25/38.25 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_p4/block/block.2/rbr_reparam/Conv
| 47 ConvRelu INT8 NPU (1,128,40,40),(64,128,1,1),(64) (1,64,40,40) 10685/12800/12800 163 3.27/3.27/3.27 35.0%/32.5%/32.5% 208 Conv:/neck/reduce_layer1/block/conv/Conv
| 48 ConvTranspose INT8 NPU (1,64,40,40),(64,64,2,2),(64) (1,64,80,80) 17888/102400/102400 266 16.04/16.04/16.04 100.0%/0.0%/0.0% 116 ConvTranspose:/neck/Bifusion1/upsample/upsample_transpose/ConvTranspose
| 49 Concat INT8 NPU (1,64,80,80),(1,64,80,80),... (1,192,80,80) 83119/0/83119 247 \ 33.7%/33.7%/32.5% 1200 Concat:/neck/Bifusion1/Concat
| 50 ConvRelu INT8 NPU (1,192,80,80),(64,192,1,1),(64) (1,64,80,80) 55846/76800/76800 226 14.16/14.16/14.16 33.8%/33.8%/32.5% 1212 Conv:/neck/Bifusion1/cv3/block/conv/Conv
| 51 ConvRelu INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 252 38.10/38.10/38.10 33.8%/33.8%/32.5% 436 Conv:/neck/Rep_p3/conv1/rbr_reparam/Conv
| 52 ConvRelu INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 252 38.10/38.10/38.10 33.8%/33.8%/32.5% 436 Conv:/neck/Rep_p3/block/block.0/rbr_reparam/Conv
| 53 ConvRelu INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 253 37.94/37.94/37.94 33.8%/33.8%/32.5% 436 Conv:/neck/Rep_p3/block/block.1/rbr_reparam/Conv
| 54 ConvRelu INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 251 38.25/38.25/38.25 33.8%/33.8%/32.5% 436 Conv:/neck/Rep_p3/block/block.2/rbr_reparam/Conv
| 55 ConvRelu INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,40,40) 18581/57600/57600 187 12.83/12.83/12.83 35.0%/32.5%/32.5% 436 Conv:/neck/downsample2/block/conv/Conv
| 56 Concat INT8 NPU (1,64,40,40),(1,64,40,40) (1,128,40,40) 13854/0/13854 161 \ 35.0%/32.5%/32.5% 200 Concat:/neck/Concat
| 57 ConvExSwish INT8 NPU (1,64,80,80),(64,64,1,1),(64) (1,64,80,80) 27862/51200/51200 402 2.65/2.65/2.65 100.0%/0.0%/0.0% 404 Conv:/detect/stems.0/block/conv/Conv
| 58 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 259 37.07/37.07/37.07 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_n3/conv1/rbr_reparam/Conv
| 59 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 391 24.55/24.55/24.55 100.0%/0.0%/0.0% 436 Conv:/detect/cls_convs.0/block/conv/Conv
| 60 Conv INT8 NPU (1,64,80,80),(80,64,1,1),(80) (1,80,80,80) 31369/76800/76800 148 9.01/9.01/9.01 33.7%/33.7%/32.5% 405 Conv:/detect/cls_preds.0/Conv
| 61 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 431 22.27/22.27/22.27 100.0%/0.0%/0.0% 436 Conv:/detect/reg_convs.0/block/conv/Conv
| 62 Conv INT8 NPU (1,64,80,80),(4,64,1,1),(4) (1,4,80,80) 17334/25600/25600 175 0.38/0.38/0.38 33.7%/33.7%/32.5% 400 Conv:/detect/reg_preds.0/Conv
| 63 Concat INT8 CPU (1,4,80,80),(1,80,80,80) (1,84,80,80) 0/0/0 981 \ 0.0%/0.0%/0.0% 600 Concat:/detect/Concat
| 64 OutputOperator INT8 CPU (1,84,80,80) \ 0/0/0 27 \ 0.0%/0.0%/0.0% 525 OutputOperator:outputs
| 65 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 253 37.94/37.94/37.94 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_n3/block/block.0/rbr_reparam/Conv
| 66 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 254 37.80/37.80/37.80 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_n3/block/block.1/rbr_reparam/Conv
| 67 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 253 37.94/37.94/37.94 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_n3/block/block.2/rbr_reparam/Conv
| 68 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,20,20) 13680/57600/57600 141 17.02/17.02/17.02 34.9%/34.9%/30.1% 345 Conv:/neck/downsample1/block/conv/Conv
| 69 Concat INT8 NPU (1,128,20,20),(1,128,20,20) (1,256,20,20) 6927/0/6927 98 \ 35.0%/35.0%/30.1% 100 Concat:/neck/Concat_1
| 70 ConvExSwish INT8 NPU (1,128,40,40),(128,128,1,1),(128) (1,128,40,40) 14442/25600/25600 216 4.94/4.94/4.94 100.0%/0.0%/0.0% 217 Conv:/detect/stems.1/block/conv/Conv
| 71 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 215 44.65/44.65/44.65 37.4%/37.4%/25.1% 678 Conv:/neck/Rep_n4/conv1/rbr_reparam/Conv
| 72 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 371 25.88/25.88/25.88 100.0%/0.0%/0.0% 345 Conv:/detect/cls_convs.1/block/conv/Conv
| 73 Conv INT8 NPU (1,128,40,40),(80,128,1,1),(80) (1,80,40,40) 11628/19200/19200 165 4.04/4.04/4.04 35.0%/32.5%/32.5% 210 Conv:/detect/cls_preds.1/Conv
| 74 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 426 22.54/22.54/22.54 100.0%/0.0%/0.0% 345 Conv:/detect/reg_convs.1/block/conv/Conv
| 75 Conv INT8 NPU (1,128,40,40),(4,128,1,1),(4) (1,4,40,40) 7819/6400/7819 160 0.21/0.21/0.21 34.9%/32.5%/32.5% 200 Conv:/detect/reg_preds.1/Conv
| 76 Concat INT8 CPU (1,4,40,40),(1,80,40,40) (1,84,40,40) 0/0/0 254 \ 0.0%/0.0%/0.0% 150 Concat:/detect/Concat_1
| 77 OutputOperator INT8 CPU (1,84,40,40) \ 0/0/0 8 \ 0.0%/0.0%/0.0% 131 OutputOperator:304
| 78 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 233 41.20/41.20/41.20 37.5%/37.5%/25.0% 678 Conv:/neck/Rep_n4/block/block.0/rbr_reparam/Conv
| 79 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 232 41.38/41.38/41.38 37.5%/37.5%/25.0% 678 Conv:/neck/Rep_n4/block/block.1/rbr_reparam/Conv
| 80 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 271 35.42/35.42/35.42 37.4%/37.4%/25.1% 678 Conv:/neck/Rep_n4/block/block.2/rbr_reparam/Conv
| 81 ConvExSwish INT8 NPU (1,256,20,20),(256,256,1,1),(256) (1,256,20,20) 9213/25600/25600 206 5.18/5.18/5.18 100.0%/0.0%/0.0% 166 Conv:/detect/stems.2/block/conv/Conv
| 82 ConvExSwish INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 383 25.07/25.07/25.07 100.0%/0.0%/0.0% 678 Conv:/detect/cls_convs.2/block/conv/Conv
| 83 Conv INT8 NPU (1,256,20,20),(80,256,1,1),(80) (1,80,20,20) 5265/9600/9600 105 3.17/3.17/3.17 34.9%/34.9%/30.2% 120 Conv:/detect/cls_preds.2/Conv
| 84 ConvExSwish INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 372 25.81/25.81/25.81 100.0%/0.0%/0.0% 678 Conv:/detect/reg_convs.2/block/conv/Conv
| 85 Conv INT8 NPU (1,256,20,20),(4,256,1,1),(4) (1,4,20,20) 3724/3200/3724 100 0.17/0.17/0.17 34.8%/34.8%/30.4% 101 Conv:/detect/reg_preds.2/Conv
| 86 Concat INT8 CPU (1,4,20,20),(1,80,20,20) (1,84,20,20) 0/0/0 69 \ 0.0%/0.0%/0.0% 37 Concat:/detect/Concat_2
| 87 OutputOperator INT8 CPU (1,84,20,20) \ 0/0/0 4 \ 0.0%/0.0%/0.0% 32 OutputOperator:305
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Total Operator Elapsed Per Frame Time(us): 25457
| Total Memory Read/Write Per Frame Size(KB): 56412.6
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| ---------------------------------------------------------------------------------------------------
| Operator Time Consuming Ranking Table
| ---------------------------------------------------------------------------------------------------
| OpType CallNumber CPUTime(us) GPUTime(us) NPUTime(us) TotalTime(us) TimeRatio(%)
| ---------------------------------------------------------------------------------------------------
| ConvRelu 54 0 0 18290 18290 71.85%
| ConvExSwish 9 0 0 3198 3198 12.56%
| Concat 9 1304 0 880 2184 8.58%
| Conv 6 0 0 853 853 3.35%
| ConvTranspose 2 0 0 474 474 1.86%
| MaxPool 3 0 0 415 415 1.63%
| OutputOperator 3 39 0 0 39 0.15%
| InputOperator 1 4 0 0 4 0.02%
| ---------------------------------------------------------------------------------------------------
DAMO-YOLO-S
[*] ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Network Layer Information Table
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| ID OpType DataType Target InputShape OutputShape Cycles(DDR/NPU/Total) Time(us) MacUsage(%) WorkLoad(0/1/2) RW(KB) FullName
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| 1 InputOperator UINT8 CPU \ (1,3,640,640) 0/0/0 5 \ 0.0%/0.0%/0.0% 0 InputOperator:images
| 2 Conv UINT8 NPU (1,3,640,640),(12,3,2,2),(12) (1,12,320,320) 97006/409600/409600 505 1.19/1.19/1.19 33.4%/33.4%/33.1% 1201 Conv:/backbone/block_list.0/Concat_focusConv
| 3 ConvRelu INT8 NPU (1,12,320,320),(32,12,3,3),(32) (1,32,320,320) 166402/921600/921600 576 25.00/25.00/25.00 33.4%/33.4%/33.1% 1604 Conv:/backbone/block_list.0/conv/conv/Conv
| 4 ConvRelu INT8 NPU (1,32,320,320),(24,32,1,1),(24) (1,24,320,320) 221684/409600/409600 407 7.86/7.86/7.86 33.4%/33.4%/33.1% 3201 Conv:/backbone/block_list.1/block_list.0/conv1/conv1/Conv
| 5 ConvRelu INT8 NPU (1,24,320,320),(128,24,3,3),(128) (1,128,160,160) 222931/921600/921600 632 45.57/45.57/45.57 33.8%/33.1%/33.1% 3237 Conv:/backbone/block_list.1/block_list.0/conv2/rbr_reparam/Conv
| 6 ConvRelu INT8 NPU (1,128,160,160),(88,128,1,1),(88) (1,88,160,160) 194350/307200/307200 411 28.55/28.55/28.55 33.8%/33.1%/33.1% 3211 Conv:/backbone/block_list.2/block_list.0/conv1/conv1/Conv
| 7 ConvRelu INT8 NPU (1,88,160,160),(128,88,3,3),(128) (1,128,80,80) 114600/691200/691200 532 49.62/49.62/49.62 33.8%/33.8%/32.5% 2509 Conv:/backbone/block_list.2/block_list.0/conv2/rbr_reparam/Conv
| 8 ConvRelu INT8 NPU (1,128,80,80),(88,128,1,1),(88) (1,88,80,80) 48893/76800/76800 214 13.71/13.71/13.71 33.8%/33.8%/32.5% 811 Conv:/backbone/block_list.2/block_list.1/conv1/conv1/Conv
| 9 ConvAddRelu INT8 NPU (1,88,80,80),(128,88,3,3),(128),... (1,128,80,80) 79967/691200/691200 445 59.33/59.33/59.33 33.8%/33.8%/32.5% 1509 Conv:/backbone/block_list.2/block_list.1/conv2/rbr_reparam/Conv
| 10 ConvRelu INT8 NPU (1,128,80,80),(88,128,1,1),(88) (1,88,80,80) 48893/76800/76800 215 13.64/13.64/13.64 33.8%/33.8%/32.5% 811 Conv:/backbone/block_list.2/block_list.2/conv1/conv1/Conv
| 11 ConvAddRelu INT8 NPU (1,88,80,80),(128,88,3,3),(128),... (1,128,80,80) 79967/691200/691200 443 59.59/59.59/59.59 33.8%/33.8%/32.5% 1509 Conv:/backbone/block_list.2/block_list.2/conv2/rbr_reparam/Conv
| 12 ConvRelu INT8 NPU (1,128,80,80),(88,128,1,1),(88) (1,88,80,80) 48893/76800/76800 214 13.71/13.71/13.71 33.8%/33.8%/32.5% 811 Conv:/backbone/block_list.2/block_list.3/conv1/conv1/Conv
| 13 ConvAddRelu INT8 NPU (1,88,80,80),(128,88,3,3),(128),... (1,128,80,80) 79967/691200/691200 458 57.64/57.64/57.64 33.8%/33.8%/32.5% 1509 Conv:/backbone/block_list.2/block_list.3/conv2/rbr_reparam/Conv
| 14 ConvRelu INT8 NPU (1,128,80,80),(88,128,1,1),(88) (1,88,80,80) 48893/76800/76800 214 13.71/13.71/13.71 33.8%/33.8%/32.5% 811 Conv:/backbone/block_list.2/block_list.4/conv1/conv1/Conv
| 15 ConvAddRelu INT8 NPU (1,88,80,80),(128,88,3,3),(128),... (1,128,80,80) 79967/691200/691200 443 59.59/59.59/59.59 33.8%/33.8%/32.5% 1509 Conv:/backbone/block_list.2/block_list.4/conv2/rbr_reparam/Conv
| 16 ConvRelu INT8 NPU (1,128,80,80),(128,128,1,1),(128) (1,128,80,80) 56001/102400/102400 227 18.80/18.80/18.80 33.8%/33.8%/32.5% 817 Conv:/backbone/block_list.3/block_list.0/conv1/conv1/Conv
| 17 ConvRelu INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,40,40) 39655/230400/230400 277 34.66/34.66/34.66 35.0%/32.5%/32.5% 945 Conv:/neck/bu_conv24/conv/Conv
| 18 ConvRelu INT8 NPU (1,128,80,80),(256,128,3,3),(256) (1,256,40,40) 51603/460800/460800 350 54.86/54.86/54.86 35.0%/32.5%/32.5% 1090 Conv:/backbone/block_list.3/block_list.0/conv2/rbr_reparam/Conv
| 19 ConvRelu INT8 NPU (1,256,40,40),(128,256,1,1),(128) (1,128,40,40) 21923/51200/51200 184 11.59/11.59/11.59 35.0%/32.5%/32.5% 433 Conv:/backbone/block_list.3/block_list.1/conv1/conv1/Conv
| 20 ConvAddRelu INT8 NPU (1,128,40,40),(256,128,3,3),(256),... (1,256,40,40) 44677/460800/460800 372 51.61/51.61/51.61 37.5%/37.5%/25.0% 890 Conv:/backbone/block_list.3/block_list.1/conv2/rbr_reparam/Conv
| 21 ConvRelu INT8 NPU (1,256,40,40),(128,256,1,1),(128) (1,128,40,40) 21923/51200/51200 190 11.23/11.23/11.23 35.0%/32.5%/32.5% 433 Conv:/backbone/block_list.3/block_list.2/conv1/conv1/Conv
| 22 ConvAddRelu INT8 NPU (1,128,40,40),(256,128,3,3),(256),... (1,256,40,40) 44677/460800/460800 313 61.34/61.34/61.34 37.5%/37.5%/25.0% 890 Conv:/backbone/block_list.3/block_list.2/conv2/rbr_reparam/Conv
| 23 ConvRelu INT8 NPU (1,256,40,40),(120,256,1,1),(120) (1,120,40,40) 21854/51200/51200 128 15.62/15.62/15.62 35.0%/32.5%/32.5% 431 Conv:/backbone/block_list.4/block_list.0/conv1/conv1/Conv
| 24 ConvAddRelu INT8 NPU (1,120,40,40),(256,120,3,3),(256),... (1,256,40,40) 44677/460800/460800 311 57.88/57.88/57.88 37.5%/37.5%/25.0% 890 Conv:/backbone/block_list.4/block_list.0/conv2/rbr_reparam/Conv
| 25 ConvRelu INT8 NPU (1,256,40,40),(120,256,1,1),(120) (1,120,40,40) 21854/51200/51200 183 10.93/10.93/10.93 35.0%/32.5%/32.5% 431 Conv:/backbone/block_list.4/block_list.1/conv1/conv1/Conv
| 26 ConvAddRelu INT8 NPU (1,120,40,40),(256,120,3,3),(256),... (1,256,40,40) 44677/460800/460800 373 48.26/48.26/48.26 37.5%/37.5%/25.0% 890 Conv:/backbone/block_list.4/block_list.1/conv2/rbr_reparam/Conv
| 27 ConvRelu INT8 NPU (1,256,40,40),(144,256,1,1),(144) (1,144,40,40) 22936/64000/64000 190 12.63/12.63/12.63 35.0%/32.5%/32.5% 437 Conv:/backbone/block_list.5/block_list.0/conv1/conv1/Conv
| 28 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,20,20) 37334/230400/230400 287 33.45/33.45/33.45 37.5%/37.5%/25.0% 978 Conv:/neck/bu_conv13/conv/Conv
| 29 ConvRelu INT8 NPU (1,144,40,40),(512,144,3,3),(512) (1,512,20,20) 37300/288000/288000 299 36.12/36.12/36.12 37.5%/31.3%/31.3% 877 Conv:/backbone/block_list.5/block_list.0/conv2/rbr_reparam/Conv
| 30 ConvExSwish INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 214 9.97/9.97/9.97 100.0%/0.0%/0.0% 330 Conv:/backbone/block_list.5/block_list.1/conv1/conv/Conv
| 31 MaxPool INT8 NPU (1,256,20,20) (1,256,21,21) 7282/0/7282 168 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/block_list.5/block_list.1/m.2/MaxPool_output_0010
| 32 MaxPool INT8 NPU (1,256,21,21) (1,256,20,20) 7282/0/7282 167 \ 100.0%/0.0%/0.0% 110 MaxPool:/backbone/block_list.5/block_list.1/m.2/MaxPool_output_0011
| 33 MaxPool INT8 NPU (1,256,20,20) (1,256,21,21) 7282/0/7282 168 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/block_list.5/block_list.1/m.2/MaxPool_output_010
| 34 MaxPool INT8 NPU (1,256,21,21) (1,256,20,20) 7282/0/7282 167 \ 100.0%/0.0%/0.0% 110 MaxPool:/backbone/block_list.5/block_list.1/m.2/MaxPool_output_011
| 35 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 121 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/block_list.5/block_list.1/m.1/MaxPool_output_0010
| 36 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 110 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/block_list.5/block_list.1/m.1/MaxPool_output_0011
| 37 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 109 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/block_list.5/block_list.1/m.1/MaxPool_output_010
| 38 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 110 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/block_list.5/block_list.1/m.1/MaxPool_output_011
| 39 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 115 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/block_list.5/block_list.1/m.0/MaxPool
| 40 Concat INT8 NPU (1,256,20,20),(1,256,20,20),... (1,1024,20,20) 27707/0/27707 124 \ 35.0%/35.0%/30.0% 400 Concat:/backbone/block_list.5/block_list.1/Concat
| 41 ConvExSwish INT8 NPU (1,1024,20,20),(512,1024,1,1),(512) (1,512,20,20) 38651/204800/204800 367 23.25/23.25/23.25 100.0%/0.0%/0.0% 916 Conv:/backbone/block_list.5/block_list.1/conv2/conv/Conv
| 42 Concat INT8 NPU (1,512,20,20),(1,256,20,20) (1,768,20,20) 20780/0/20780 116 \ 35.0%/35.0%/30.0% 300 Concat:/neck/Concat
| 43 ConvRelu INT8 NPU (1,768,20,20),(256,768,1,1),(256) (1,256,20,20) 20572/76800/76800 144 22.22/22.22/22.22 37.5%/37.5%/25.0% 494 Conv:/neck/merge_3/conv1/conv/Conv
| 44 ConvRelu INT8 NPU (1,768,20,20),(256,768,1,1),(256) (1,256,20,20) 20572/76800/76800 143 22.38/22.38/22.38 37.5%/37.5%/25.0% 494 Conv:/neck/merge_3/conv2/conv/Conv
| 45 ConvRelu INT8 NPU (1,256,20,20),(192,256,3,3),(192) (1,192,20,20) 21074/172800/172800 178 40.45/40.45/40.45 33.3%/33.3%/33.3% 533 Conv:/neck/merge_3/convs.0/conv2/rbr_reparam/Conv
| 46 ConvReluAdd INT8 NPU (1,192,20,20),(256,192,3,3),(256),... (1,256,20,20) 24555/172800/172800 210 34.29/34.29/34.29 37.5%/37.5%/25.0% 609 Conv:/neck/merge_3/convs.0/conv1/conv/Conv
| 47 ConvRelu INT8 NPU (1,256,20,20),(192,256,3,3),(192) (1,192,20,20) 21074/172800/172800 234 30.77/30.77/30.77 33.3%/33.3%/33.3% 533 Conv:/neck/merge_3/convs.1/conv2/rbr_reparam/Conv
| 48 ConvReluAdd INT8 NPU (1,192,20,20),(256,192,3,3),(256),... (1,256,20,20) 24555/172800/172800 215 33.49/33.49/33.49 37.5%/37.5%/25.0% 609 Conv:/neck/merge_3/convs.1/conv1/conv/Conv
| 49 ConvRelu INT8 NPU (1,256,20,20),(192,256,3,3),(192) (1,192,20,20) 21074/172800/172800 190 37.89/37.89/37.89 33.3%/33.3%/33.3% 533 Conv:/neck/merge_3/convs.2/conv2/rbr_reparam/Conv
| 50 ConvReluAdd INT8 NPU (1,192,20,20),(256,192,3,3),(256),... (1,256,20,20) 24555/172800/172800 186 38.71/38.71/38.71 37.4%/37.4%/25.1% 609 Conv:/neck/merge_3/convs.2/conv1/conv/Conv
| 51 Concat INT8 NPU (1,256,20,20),(1,256,20,20),... (1,1024,20,20) 27707/0/27707 178 \ 35.0%/35.0%/30.0% 400 Concat:/neck/merge_3/Concat
| 52 ConvRelu INT8 NPU (1,1024,20,20),(512,1024,1,1),(512) (1,512,20,20) 38651/204800/204800 286 29.84/29.84/29.84 37.5%/31.3%/31.3% 916 Conv:/neck/merge_3/conv3/conv/Conv
| 53 Resize INT8 NPU (1,512,20,20),(0),(4) (1,512,40,40) 34634/0/34634 255 \ 100.0%/0.0%/0.0% 200 Resize:/neck/upsample/Resize
| 54 Concat INT8 NPU (1,256,40,40),(1,128,40,40),... (1,896,40,40) 96972/0/96972 225 \ 35.0%/32.5%/32.5% 1400 Concat:/neck/Concat_1
| 55 ConvRelu INT8 NPU (1,896,40,40),(128,896,1,1),(128) (1,128,40,40) 59326/179200/179200 271 27.55/27.55/27.55 35.0%/32.5%/32.5% 1513 Conv:/neck/merge_4/conv1/conv/Conv
| 56 ConvRelu INT8 NPU (1,896,40,40),(128,896,1,1),(128) (1,128,40,40) 59326/179200/179200 276 27.05/27.05/27.05 35.0%/32.5%/32.5% 1513 Conv:/neck/merge_4/conv2/conv/Conv
| 57 ConvRelu INT8 NPU (1,128,40,40),(96,128,3,3),(96) (1,96,40,40) 15888/172800/172800 231 31.17/31.17/31.17 33.3%/33.3%/33.3% 308 Conv:/neck/merge_4/convs.0/conv2/rbr_reparam/Conv
| 58 ConvReluAdd INT8 NPU (1,96,40,40),(128,96,3,3),(128),... (1,128,40,40) 22823/172800/172800 184 39.13/39.13/39.13 35.0%/32.5%/32.5% 459 Conv:/neck/merge_4/convs.0/conv1/conv/Conv
| 59 ConvRelu INT8 NPU (1,128,40,40),(96,128,3,3),(96) (1,96,40,40) 15888/172800/172800 175 41.14/41.14/41.14 33.3%/33.3%/33.3% 308 Conv:/neck/merge_4/convs.1/conv2/rbr_reparam/Conv
| 60 ConvReluAdd INT8 NPU (1,96,40,40),(128,96,3,3),(128),... (1,128,40,40) 22823/172800/172800 172 41.86/41.86/41.86 35.0%/32.5%/32.5% 459 Conv:/neck/merge_4/convs.1/conv1/conv/Conv
| 61 ConvRelu INT8 NPU (1,128,40,40),(96,128,3,3),(96) (1,96,40,40) 15888/172800/172800 228 31.58/31.58/31.58 33.3%/33.3%/33.3% 308 Conv:/neck/merge_4/convs.2/conv2/rbr_reparam/Conv
| 62 ConvReluAdd INT8 NPU (1,96,40,40),(128,96,3,3),(128),... (1,128,40,40) 22823/172800/172800 233 30.90/30.90/30.90 35.0%/32.5%/32.5% 459 Conv:/neck/merge_4/convs.2/conv1/conv/Conv
| 63 Concat INT8 NPU (1,128,40,40),(1,128,40,40),... (1,512,40,40) 55413/0/55413 171 \ 35.0%/32.5%/32.5% 800 Concat:/neck/merge_4/Concat
| 64 ConvRelu INT8 NPU (1,512,40,40),(256,512,1,1),(256) (1,256,40,40) 46062/204800/204800 262 32.57/32.57/32.57 35.0%/32.5%/32.5% 930 Conv:/neck/merge_4/conv3/conv/Conv
| 65 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,20,20) 37334/230400/230400 305 31.48/31.48/31.48 37.4%/37.4%/25.1% 978 Conv:/neck/bu_conv46/conv/Conv
| 66 Resize INT8 NPU (1,256,40,40),(0),(4) (1,256,80,80) 69266/0/69266 364 \ 100.0%/0.0%/0.0% 400 Resize:/neck/upsample_1/Resize
| 67 Concat INT8 NPU (1,128,80,80),(1,256,80,80) (1,384,80,80) 166237/0/166237 346 \ 33.7%/33.7%/32.5% 2400 Concat:/neck/Concat_2
| 68 ConvRelu INT8 NPU (1,384,80,80),(64,384,1,1),(64) (1,64,80,80) 97820/153600/153600 293 21.84/21.84/21.84 33.8%/33.8%/32.5% 2424 Conv:/neck/merge_5/conv1/conv/Conv
| 69 ConvRelu INT8 NPU (1,384,80,80),(64,384,1,1),(64) (1,64,80,80) 97820/153600/153600 247 25.91/25.91/25.91 33.8%/33.8%/32.5% 2424 Conv:/neck/merge_5/conv2/conv/Conv
| 70 ConvRelu INT8 NPU (1,64,80,80),(48,64,3,3),(48) (1,48,80,80) 25196/230400/230400 198 36.36/36.36/36.36 33.8%/33.8%/32.5% 427 Conv:/neck/merge_5/convs.0/conv2/rbr_reparam/Conv
| 71 ConvReluAdd INT8 NPU (1,48,80,80),(64,48,3,3),(64),... (1,64,80,80) 39049/230400/230400 195 36.92/36.92/36.92 33.8%/33.8%/32.5% 727 Conv:/neck/merge_5/convs.0/conv1/conv/Conv
| 72 ConvRelu INT8 NPU (1,64,80,80),(48,64,3,3),(48) (1,48,80,80) 25196/230400/230400 196 36.73/36.73/36.73 33.8%/33.8%/32.5% 427 Conv:/neck/merge_5/convs.1/conv2/rbr_reparam/Conv
| 73 ConvReluAdd INT8 NPU (1,48,80,80),(64,48,3,3),(64),... (1,64,80,80) 39049/230400/230400 195 36.92/36.92/36.92 33.8%/33.8%/32.5% 727 Conv:/neck/merge_5/convs.1/conv1/conv/Conv
| 74 ConvRelu INT8 NPU (1,64,80,80),(48,64,3,3),(48) (1,48,80,80) 25196/230400/230400 197 36.55/36.55/36.55 33.8%/33.8%/32.5% 427 Conv:/neck/merge_5/convs.2/conv2/rbr_reparam/Conv
| 75 ConvReluAdd INT8 NPU (1,48,80,80),(64,48,3,3),(64),... (1,64,80,80) 39049/230400/230400 199 36.18/36.18/36.18 33.7%/33.7%/32.5% 727 Conv:/neck/merge_5/convs.2/conv1/conv/Conv
| 76 Concat INT8 NPU (1,64,80,80),(1,64,80,80),... (1,256,80,80) 110825/0/110825 224 \ 33.7%/33.7%/32.5% 1600 Concat:/neck/merge_5/Concat
| 77 ConvRelu INT8 NPU (1,256,80,80),(128,256,1,1),(128) (1,128,80,80) 84262/204800/204800 232 36.78/36.78/36.78 33.8%/33.8%/32.5% 1633 Conv:/neck/merge_5/conv3/conv/Conv
| 78 Conv INT8 NPU (1,128,80,80),(68,128,3,3),(68) (1,68,80,80) 47698/691200/691200 411 49.64/49.64/49.64 33.8%/33.8%/32.5% 877 Conv:/head/gfl_reg.0/Conv
| 79 Conv INT8 NPU (1,128,80,80),(80,128,3,3),(80) (1,80,80,80) 48166/691200/691200 414 57.97/57.97/57.97 33.7%/33.7%/32.5% 890 Conv:/head/gfl_cls.0/Conv
| 80 Concat INT8 CPU (1,68,80,80),(1,80,80,80) (1,148,80,80) 0/0/0 1607 \ 0.0%/0.0%/0.0% 1000 Concat:/head/Concat
| 81 OutputOperator INT8 CPU (1,148,80,80) \ 0/0/0 39 \ 0.0%/0.0%/0.0% 925 OutputOperator:output
| 82 ConvRelu INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,40,40) 39655/230400/230400 237 40.51/40.51/40.51 35.0%/32.5%/32.5% 945 Conv:/neck/bu_conv57/conv/Conv
| 83 Concat INT8 NPU (1,256,40,40),(1,128,40,40) (1,384,40,40) 41560/0/41560 142 \ 35.0%/32.5%/32.5% 600 Concat:/neck/Concat_3
| 84 ConvRelu INT8 NPU (1,384,40,40),(128,384,1,1),(128) (1,128,40,40) 29404/76800/76800 199 16.08/16.08/16.08 35.0%/32.5%/32.5% 649 Conv:/neck/merge_7/conv1/conv/Conv
| 85 ConvRelu INT8 NPU (1,384,40,40),(128,384,1,1),(128) (1,128,40,40) 29404/76800/76800 199 16.08/16.08/16.08 35.0%/32.5%/32.5% 649 Conv:/neck/merge_7/conv2/conv/Conv
| 86 ConvRelu INT8 NPU (1,128,40,40),(96,128,3,3),(96) (1,96,40,40) 15888/172800/172800 230 31.30/31.30/31.30 33.3%/33.3%/33.3% 308 Conv:/neck/merge_7/convs.0/conv2/rbr_reparam/Conv
| 87 ConvReluAdd INT8 NPU (1,96,40,40),(128,96,3,3),(128),... (1,128,40,40) 22823/172800/172800 229 31.44/31.44/31.44 35.0%/32.5%/32.5% 459 Conv:/neck/merge_7/convs.0/conv1/conv/Conv
| 88 ConvRelu INT8 NPU (1,128,40,40),(96,128,3,3),(96) (1,96,40,40) 15888/172800/172800 230 31.30/31.30/31.30 33.3%/33.3%/33.3% 308 Conv:/neck/merge_7/convs.1/conv2/rbr_reparam/Conv
| 89 ConvReluAdd INT8 NPU (1,96,40,40),(128,96,3,3),(128),... (1,128,40,40) 22823/172800/172800 184 39.13/39.13/39.13 35.0%/32.5%/32.5% 459 Conv:/neck/merge_7/convs.1/conv1/conv/Conv
| 90 ConvRelu INT8 NPU (1,128,40,40),(96,128,3,3),(96) (1,96,40,40) 15888/172800/172800 175 41.14/41.14/41.14 33.3%/33.3%/33.3% 308 Conv:/neck/merge_7/convs.2/conv2/rbr_reparam/Conv
| 91 ConvReluAdd INT8 NPU (1,96,40,40),(128,96,3,3),(128),... (1,128,40,40) 22823/172800/172800 235 30.64/30.64/30.64 35.0%/32.5%/32.5% 459 Conv:/neck/merge_7/convs.2/conv1/conv/Conv
| 92 Concat INT8 NPU (1,128,40,40),(1,128,40,40),... (1,512,40,40) 55413/0/55413 217 \ 35.0%/32.5%/32.5% 800 Concat:/neck/merge_7/Concat
| 93 ConvRelu INT8 NPU (1,512,40,40),(256,512,1,1),(256) (1,256,40,40) 46062/204800/204800 266 32.08/32.08/32.08 35.0%/32.5%/32.5% 930 Conv:/neck/merge_7/conv3/conv/Conv
| 94 Conv INT8 NPU (1,256,40,40),(68,256,3,3),(68) (1,68,40,40) 23507/345600/345600 307 33.22/33.22/33.22 35.0%/32.5%/32.5% 553 Conv:/head/gfl_reg.1/Conv
| 95 Conv INT8 NPU (1,256,40,40),(80,256,3,3),(80) (1,80,40,40) 24442/345600/345600 316 37.97/37.97/37.97 35.0%/32.5%/32.5% 580 Conv:/head/gfl_cls.1/Conv
| 96 Concat INT8 CPU (1,68,40,40),(1,80,40,40) (1,148,40,40) 0/0/0 416 \ 0.0%/0.0%/0.0% 250 Concat:/head/Concat_1
| 97 OutputOperator INT8 CPU (1,148,40,40) \ 0/0/0 13 \ 0.0%/0.0%/0.0% 231 OutputOperator:669
| 98 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,20,20) 37334/230400/230400 304 31.58/31.58/31.58 37.4%/37.4%/25.1% 978 Conv:/neck/bu_conv76/conv/Conv
| 99 Concat INT8 NPU (1,512,20,20),(1,256,20,20),... (1,1024,20,20) 27707/0/27707 180 \ 35.0%/35.0%/30.0% 400 Concat:/neck/Concat_4
| 100 ConvRelu INT8 NPU (1,1024,20,20),(256,1024,1,1),(256) (1,256,20,20) 26252/102400/102400 214 19.94/19.94/19.94 35.0%/35.0%/30.0% 658 Conv:/neck/merge_6/conv1/conv/Conv
| 101 ConvRelu INT8 NPU (1,1024,20,20),(256,1024,1,1),(256) (1,256,20,20) 26252/102400/102400 213 20.03/20.03/20.03 35.0%/35.0%/30.0% 658 Conv:/neck/merge_6/conv2/conv/Conv
| 102 ConvRelu INT8 NPU (1,256,20,20),(192,256,3,3),(192) (1,192,20,20) 21074/172800/172800 233 30.90/30.90/30.90 33.3%/33.3%/33.3% 533 Conv:/neck/merge_6/convs.0/conv2/rbr_reparam/Conv
| 103 ConvReluAdd INT8 NPU (1,192,20,20),(256,192,3,3),(256),... (1,256,20,20) 24555/172800/172800 214 33.64/33.64/33.64 37.5%/37.5%/25.0% 609 Conv:/neck/merge_6/convs.0/conv1/conv/Conv
| 104 ConvRelu INT8 NPU (1,256,20,20),(192,256,3,3),(192) (1,192,20,20) 21074/172800/172800 182 39.56/39.56/39.56 33.3%/33.3%/33.3% 533 Conv:/neck/merge_6/convs.1/conv2/rbr_reparam/Conv
| 105 ConvReluAdd INT8 NPU (1,192,20,20),(256,192,3,3),(256),... (1,256,20,20) 24555/172800/172800 158 45.57/45.57/45.57 37.5%/37.5%/25.0% 609 Conv:/neck/merge_6/convs.1/conv1/conv/Conv
| 106 ConvRelu INT8 NPU (1,256,20,20),(192,256,3,3),(192) (1,192,20,20) 21074/172800/172800 231 31.17/31.17/31.17 33.3%/33.3%/33.3% 533 Conv:/neck/merge_6/convs.2/conv2/rbr_reparam/Conv
| 107 ConvReluAdd INT8 NPU (1,192,20,20),(256,192,3,3),(256),... (1,256,20,20) 24555/172800/172800 197 36.55/36.55/36.55 37.4%/37.4%/25.1% 609 Conv:/neck/merge_6/convs.2/conv1/conv/Conv
| 108 Concat INT8 NPU (1,256,20,20),(1,256,20,20),... (1,1024,20,20) 27707/0/27707 177 \ 35.0%/35.0%/30.0% 400 Concat:/neck/merge_6/Concat
| 109 ConvRelu INT8 NPU (1,1024,20,20),(512,1024,1,1),(512) (1,512,20,20) 38651/204800/204800 281 30.37/30.37/30.37 37.5%/31.2%/31.2% 916 Conv:/neck/merge_6/conv3/conv/Conv
| 110 Conv INT8 NPU (1,512,20,20),(68,512,3,3),(68) (1,68,20,20) 18633/172800/172800 249 20.48/20.48/20.48 35.0%/35.0%/30.0% 506 Conv:/head/gfl_reg.2/Conv
| 111 Conv INT8 NPU (1,512,20,20),(80,512,3,3),(80) (1,80,20,20) 20503/172800/172800 251 23.90/23.90/23.90 34.9%/34.9%/30.2% 560 Conv:/head/gfl_cls.2/Conv
| 112 Concat INT8 CPU (1,68,20,20),(1,80,20,20) (1,148,20,20) 0/0/0 114 \ 0.0%/0.0%/0.0% 62 Concat:/head/Concat_2
| 113 OutputOperator INT8 CPU (1,148,20,20) \ 0/0/0 5 \ 0.0%/0.0%/0.0% 57 OutputOperator:670
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Total Operator Elapsed Per Frame Time(us): 28561
| Total Memory Read/Write Per Frame Size(KB): 88072.3
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| ---------------------------------------------------------------------------------------------------
| Operator Time Consuming Ranking Table
| ---------------------------------------------------------------------------------------------------
| OpType CallNumber CPUTime(us) GPUTime(us) NPUTime(us) TotalTime(us) TimeRatio(%)
| ---------------------------------------------------------------------------------------------------
| ConvRelu 52 0 0 13210 13210 46.25%
| Concat 14 2137 0 2100 4237 14.83%
| ConvAddRelu 8 0 0 3158 3158 11.06%
| ConvReluAdd 15 0 0 3006 3006 10.52%
| Conv 7 0 0 2453 2453 8.59%
| MaxPool 9 0 0 1235 1235 4.32%
| Resize 2 0 0 619 619 2.17%
| ConvExSwish 2 0 0 581 581 2.03%
| OutputOperator 3 57 0 0 57 0.20%
| InputOperator 1 5 0 0 5 0.02%
| ---------------------------------------------------------------------------------------------------
YOLOv8s
[*] ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Network Layer Information Table
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| ID OpType DataType Target InputShape OutputShape Cycles(DDR/NPU/Total) Time(us) MacUsage(%) WorkLoad(0/1/2) RW(KB) FullName
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| 1 InputOperator UINT8 CPU \ (1,3,640,640) 0/0/0 4 \ 0.0%/0.0%/0.0% 0 InputOperator:images
| 2 ConvExSwish UINT8 NPU (1,3,640,640),(32,3,3,3),(32) (1,32,320,320) 152549/921600/921600 2355 1.53/1.53/1.53 100.0%/0.0%/0.0% 1204 Conv:/model.0/conv/Conv
| 3 ConvExSwish INT8 NPU (1,32,320,320),(64,32,3,3),(64) (1,64,160,160) 166878/460800/460800 1335 14.38/14.38/14.38 100.0%/0.0%/0.0% 3218 Conv:/model.1/conv/Conv
| 4 ConvExSwish INT8 NPU (1,64,160,160),(64,64,1,1),(64) (1,64,160,160) 110981/204800/204800 1217 3.51/3.51/3.51 100.0%/0.0%/0.0% 1604 Conv:/model.2/cv1/conv/Conv
| 5 Split INT8 NPU (1,64,160,160),(2) (1,32,160,160),... 110825/0/110825 402 \ 100.0%/0.0%/0.0% 1600 Split:/model.2/Split
| 6 ConvExSwish INT8 NPU (1,32,160,160),(32,32,3,3),(32) (1,32,160,160) 55733/230400/230400 671 14.31/14.31/14.31 100.0%/0.0%/0.0% 809 Conv:/model.2/m.0/cv1/conv/Conv
| 7 ConvExSwish INT8 NPU (1,32,160,160),(32,32,3,3),(32) (1,32,160,160) 55733/230400/230400 669 14.35/14.35/14.35 100.0%/0.0%/0.0% 809 Conv:/model.2/m.0/cv2/conv/Conv
| 8 Add INT8 NPU (1,32,160,160),(1,32,160,160) (1,32,160,160) 83119/0/83119 231 \ 33.7%/33.1%/33.1% 1600 Add:/model.2/m.0/Add
| 9 Concat INT8 NPU (1,32,160,160),(1,32,160,160),... (1,96,160,160) 166237/0/166237 345 \ 33.7%/33.1%/33.1% 2400 Concat:/model.2/Concat
| 10 ConvExSwish INT8 NPU (1,96,160,160),(64,96,1,1),(64) (1,64,160,160) 138756/204800/204800 1283 4.99/4.99/4.99 100.0%/0.0%/0.0% 2406 Conv:/model.2/cv2/conv/Conv
| 11 ConvExSwish INT8 NPU (1,64,160,160),(128,64,3,3),(128) (1,128,80,80) 85647/460800/460800 811 23.67/23.67/23.67 100.0%/0.0%/0.0% 1673 Conv:/model.3/conv/Conv
| 12 ConvExSwish INT8 NPU (1,128,80,80),(128,128,1,1),(128) (1,128,80,80) 56001/102400/102400 677 6.30/6.30/6.30 100.0%/0.0%/0.0% 817 Conv:/model.4/cv1/conv/Conv
| 13 Split INT8 NPU (1,128,80,80),(2) (1,64,80,80),... 55413/0/55413 265 \ 100.0%/0.0%/0.0% 800 Split:/model.4/Split
| 14 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 434 22.12/22.12/22.12 100.0%/0.0%/0.0% 436 Conv:/model.4/m.0/cv1/conv/Conv
| 15 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 433 22.17/22.17/22.17 100.0%/0.0%/0.0% 436 Conv:/model.4/m.0/cv2/conv/Conv
| 16 Add INT8 NPU (1,64,80,80),(1,64,80,80) (1,64,80,80) 41560/0/41560 186 \ 33.7%/33.7%/32.5% 800 Add:/model.4/m.0/Add
| 17 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 434 22.12/22.12/22.12 100.0%/0.0%/0.0% 436 Conv:/model.4/m.1/cv1/conv/Conv
| 18 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 439 21.87/21.87/21.87 100.0%/0.0%/0.0% 436 Conv:/model.4/m.1/cv2/conv/Conv
| 19 Add INT8 NPU (1,64,80,80),(1,64,80,80) (1,64,80,80) 41560/0/41560 186 \ 33.7%/33.7%/32.5% 800 Add:/model.4/m.1/Add
| 20 Concat INT8 NPU (1,64,80,80),(1,64,80,80),... (1,256,80,80) 110825/0/110825 279 \ 33.7%/33.7%/32.5% 1600 Concat:/model.4/Concat
| 21 ConvExSwish INT8 NPU (1,256,80,80),(128,256,1,1),(128) (1,128,80,80) 84262/204800/204800 752 11.35/11.35/11.35 100.0%/0.0%/0.0% 1633 Conv:/model.4/cv2/conv/Conv
| 22 ConvExSwish INT8 NPU (1,128,80,80),(256,128,3,3),(256) (1,256,40,40) 51603/460800/460800 793 24.21/24.21/24.21 100.0%/0.0%/0.0% 1090 Conv:/model.5/conv/Conv
| 23 ConvExSwish INT8 NPU (1,256,40,40),(256,256,1,1),(256) (1,256,40,40) 29992/102400/102400 410 10.41/10.41/10.41 100.0%/0.0%/0.0% 466 Conv:/model.6/cv1/conv/Conv
| 24 Split INT8 NPU (1,256,40,40),(2) (1,128,40,40),... 27707/0/27707 199 \ 100.0%/0.0%/0.0% 400 Split:/model.6/Split
| 25 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 425 22.59/22.59/22.59 100.0%/0.0%/0.0% 345 Conv:/model.6/m.0/cv1/conv/Conv
| 26 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 424 22.64/22.64/22.64 100.0%/0.0%/0.0% 345 Conv:/model.6/m.0/cv2/conv/Conv
| 27 Add INT8 NPU (1,128,40,40),(1,128,40,40) (1,128,40,40) 20780/0/20780 166 \ 35.0%/32.5%/32.5% 400 Add:/model.6/m.0/Add
| 28 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 426 22.54/22.54/22.54 100.0%/0.0%/0.0% 345 Conv:/model.6/m.1/cv1/conv/Conv
| 29 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 425 22.59/22.59/22.59 100.0%/0.0%/0.0% 345 Conv:/model.6/m.1/cv2/conv/Conv
| 30 Add INT8 NPU (1,128,40,40),(1,128,40,40) (1,128,40,40) 20780/0/20780 166 \ 35.0%/32.5%/32.5% 400 Add:/model.6/m.1/Add
| 31 Concat INT8 NPU (1,128,40,40),(1,128,40,40),... (1,512,40,40) 55413/0/55413 214 \ 35.0%/32.5%/32.5% 800 Concat:/model.6/Concat
| 32 ConvExSwish INT8 NPU (1,512,40,40),(256,512,1,1),(256) (1,256,40,40) 46062/204800/204800 417 20.46/20.46/20.46 100.0%/0.0%/0.0% 930 Conv:/model.6/cv2/conv/Conv
| 33 ConvExSwish INT8 NPU (1,256,40,40),(512,256,3,3),(512) (1,512,20,20) 60815/460800/460800 772 24.87/24.87/24.87 100.0%/0.0%/0.0% 1556 Conv:/model.7/conv/Conv
| 34 ConvExSwish INT8 NPU (1,512,20,20),(512,512,1,1),(512) (1,512,20,20) 22858/102400/102400 283 15.08/15.08/15.08 100.0%/0.0%/0.0% 460 Conv:/model.8/cv1/conv/Conv
| 35 Split INT8 NPU (1,512,20,20),(2) (1,256,20,20),... 13854/0/13854 166 \ 100.0%/0.0%/0.0% 200 Split:/model.8/Split
| 36 ConvExSwish INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 426 22.54/22.54/22.54 100.0%/0.0%/0.0% 678 Conv:/model.8/m.0/cv1/conv/Conv
| 37 ConvExSwish INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 426 22.54/22.54/22.54 100.0%/0.0%/0.0% 678 Conv:/model.8/m.0/cv2/conv/Conv
| 38 Add INT8 NPU (1,256,20,20),(1,256,20,20) (1,256,20,20) 10390/0/10390 155 \ 35.0%/35.0%/30.1% 200 Add:/model.8/m.0/Add
| 39 Concat INT8 NPU (1,256,20,20),(1,256,20,20),... (1,768,20,20) 20780/0/20780 170 \ 35.0%/35.0%/30.0% 300 Concat:/model.8/Concat
| 40 ConvExSwish INT8 NPU (1,768,20,20),(512,768,1,1),(512) (1,512,20,20) 30754/153600/153600 359 17.83/17.83/17.83 100.0%/0.0%/0.0% 688 Conv:/model.8/cv2/conv/Conv
| 41 ConvExSwish INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 213 10.02/10.02/10.02 100.0%/0.0%/0.0% 330 Conv:/model.9/cv1/conv/Conv
| 42 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 171 \ 100.0%/0.0%/0.0% 100 MaxPool:/model.9/m/MaxPool
| 43 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 171 \ 100.0%/0.0%/0.0% 100 MaxPool:/model.9/m_1/MaxPool
| 44 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 170 \ 100.0%/0.0%/0.0% 100 MaxPool:/model.9/m_2/MaxPool
| 45 Concat INT8 NPU (1,256,20,20),(1,256,20,20),... (1,1024,20,20) 27707/0/27707 136 \ 35.0%/35.0%/30.0% 400 Concat:/model.9/Concat
| 46 ConvExSwish INT8 NPU (1,1024,20,20),(512,1024,1,1),(512) (1,512,20,20) 38651/204800/204800 366 23.32/23.32/23.32 100.0%/0.0%/0.0% 916 Conv:/model.9/cv2/conv/Conv
| 47 Resize INT8 NPU (1,512,20,20),(0),(4) (1,512,40,40) 34634/0/34634 260 \ 100.0%/0.0%/0.0% 200 Resize:/model.10/Resize
| 48 Concat INT8 NPU (1,512,40,40),(1,256,40,40) (1,768,40,40) 83119/0/83119 209 \ 35.0%/32.5%/32.5% 1200 Concat:/model.11/Concat
| 49 ConvExSwish INT8 NPU (1,768,40,40),(256,768,1,1),(256) (1,256,40,40) 62131/307200/307200 628 20.38/20.38/20.38 100.0%/0.0%/0.0% 1394 Conv:/model.12/cv1/conv/Conv
| 50 Split INT8 NPU (1,256,40,40),(2) (1,128,40,40),... 27707/0/27707 201 \ 100.0%/0.0%/0.0% 400 Split:/model.12/Split
| 51 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 425 22.59/22.59/22.59 100.0%/0.0%/0.0% 345 Conv:/model.12/m.0/cv1/conv/Conv
| 52 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 425 22.59/22.59/22.59 100.0%/0.0%/0.0% 345 Conv:/model.12/m.0/cv2/conv/Conv
| 53 Concat INT8 NPU (1,128,40,40),(1,128,40,40),... (1,384,40,40) 41560/0/41560 196 \ 35.0%/32.5%/32.5% 600 Concat:/model.12/Concat
| 54 ConvExSwish INT8 NPU (1,384,40,40),(256,384,1,1),(256) (1,256,40,40) 38027/153600/153600 431 14.85/14.85/14.85 100.0%/0.0%/0.0% 698 Conv:/model.12/cv2/conv/Conv
| 55 Resize INT8 NPU (1,256,40,40),(0),(4) (1,256,80,80) 69266/0/69266 388 \ 100.0%/0.0%/0.0% 400 Resize:/model.13/Resize
| 56 Concat INT8 NPU (1,256,80,80),(1,128,80,80) (1,384,80,80) 166237/0/166237 352 \ 33.7%/33.7%/32.5% 2400 Concat:/model.14/Concat
| 57 ConvExSwish INT8 NPU (1,384,80,80),(128,384,1,1),(128) (1,128,80,80) 112522/307200/307200 857 14.94/14.94/14.94 100.0%/0.0%/0.0% 2449 Conv:/model.15/cv1/conv/Conv
| 58 Split INT8 NPU (1,128,80,80),(2) (1,64,80,80),... 55413/0/55413 267 \ 100.0%/0.0%/0.0% 800 Split:/model.15/Split
| 59 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 435 22.07/22.07/22.07 100.0%/0.0%/0.0% 436 Conv:/model.15/m.0/cv1/conv/Conv
| 60 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 433 22.17/22.17/22.17 100.0%/0.0%/0.0% 436 Conv:/model.15/m.0/cv2/conv/Conv
| 61 Concat INT8 NPU (1,64,80,80),(1,64,80,80),... (1,192,80,80) 83119/0/83119 245 \ 33.7%/33.7%/32.5% 1200 Concat:/model.15/Concat
| 62 ConvExSwish INT8 NPU (1,192,80,80),(128,192,1,1),(128) (1,128,80,80) 70132/153600/153600 716 8.94/8.94/8.94 100.0%/0.0%/0.0% 1225 Conv:/model.15/cv2/conv/Conv
| 63 ConvExSwish INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,40,40) 39655/230400/230400 511 18.79/18.79/18.79 100.0%/0.0%/0.0% 945 Conv:/model.16/conv/Conv
| 64 Concat INT8 NPU (1,128,40,40),(1,256,40,40) (1,384,40,40) 41560/0/41560 221 \ 35.0%/32.5%/32.5% 600 Concat:/model.17/Concat
| 65 ConvExSwish INT8 NPU (1,128,80,80),(64,128,3,3),(64) (1,64,80,80) 44071/460800/460800 761 25.23/25.23/25.23 100.0%/0.0%/0.0% 872 Conv:/model.22/cv2.0/cv2.0.0/conv/Conv
| 66 ConvExSwish INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,80,80) 60434/921600/921600 1335 28.76/28.76/28.76 100.0%/0.0%/0.0% 945 Conv:/model.22/cv3.0/cv3.0.0/conv/Conv
| 67 ConvExSwish INT8 NPU (1,384,40,40),(256,384,1,1),(256) (1,256,40,40) 38027/153600/153600 450 14.22/14.22/14.22 100.0%/0.0%/0.0% 698 Conv:/model.18/cv1/conv/Conv
| 68 Split INT8 NPU (1,256,40,40),(2) (1,128,40,40),... 27707/0/27707 220 \ 100.0%/0.0%/0.0% 400 Split:/model.18/Split
| 69 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 454 21.15/21.15/21.15 100.0%/0.0%/0.0% 436 Conv:/model.22/cv2.0/cv2.0.1/conv/Conv
| 70 Conv INT8 NPU (1,64,80,80),(64,64,1,1),(64) (1,64,80,80) 27862/51200/51200 207 5.15/5.15/5.15 33.7%/33.7%/32.5% 404 Conv:/model.22/cv2.0/cv2.0.2/Conv
| 71 ConvExSwish INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,80,80) 60434/921600/921600 1334 28.79/28.79/28.79 100.0%/0.0%/0.0% 945 Conv:/model.22/cv3.0/cv3.0.1/conv/Conv
| 72 Conv INT8 NPU (1,128,80,80),(80,128,1,1),(80) (1,80,80,80) 45395/76800/76800 189 14.11/14.11/14.11 33.7%/33.7%/32.5% 810 Conv:/model.22/cv3.0/cv3.0.2/Conv
| 73 Concat INT8 NPU (1,64,80,80),(1,80,80,80) (1,144,80,80) 62339/0/62339 186 \ 33.7%/33.7%/32.5% 900 Concat:/model.22/Concat
| 74 OutputOperator INT8 CPU (1,144,80,80) \ 0/0/0 87 \ 0.0%/0.0%/0.0% 900 OutputOperator:output0
| 75 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 392 24.49/24.49/24.49 100.0%/0.0%/0.0% 345 Conv:/model.18/m.0/cv1/conv/Conv
| 76 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 391 24.55/24.55/24.55 100.0%/0.0%/0.0% 345 Conv:/model.18/m.0/cv2/conv/Conv
| 77 Concat INT8 NPU (1,128,40,40),(1,128,40,40),... (1,384,40,40) 41560/0/41560 163 \ 35.0%/32.5%/32.5% 600 Concat:/model.18/Concat
| 78 ConvExSwish INT8 NPU (1,384,40,40),(256,384,1,1),(256) (1,256,40,40) 38027/153600/153600 397 16.12/16.12/16.12 100.0%/0.0%/0.0% 698 Conv:/model.18/cv2/conv/Conv
| 79 ConvExSwish INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,20,20) 37334/230400/230400 439 21.87/21.87/21.87 100.0%/0.0%/0.0% 978 Conv:/model.19/conv/Conv
| 80 Concat INT8 NPU (1,256,20,20),(1,512,20,20) (1,768,20,20) 20780/0/20780 134 \ 35.0%/35.0%/30.0% 300 Concat:/model.20/Concat
| 81 ConvExSwish INT8 NPU (1,256,40,40),(64,256,3,3),(64) (1,64,40,40) 22321/230400/230400 464 20.69/20.69/20.69 100.0%/0.0%/0.0% 544 Conv:/model.22/cv2.1/cv2.1.0/conv/Conv
| 82 ConvExSwish INT8 NPU (1,256,40,40),(128,256,3,3),(128) (1,128,40,40) 30789/460800/460800 783 24.52/24.52/24.52 100.0%/0.0%/0.0% 689 Conv:/model.22/cv3.1/cv3.1.0/conv/Conv
| 83 ConvExSwish INT8 NPU (1,768,20,20),(512,768,1,1),(512) (1,512,20,20) 30754/153600/153600 379 16.89/16.89/16.89 100.0%/0.0%/0.0% 688 Conv:/model.21/cv1/conv/Conv
| 84 Split INT8 NPU (1,512,20,20),(2) (1,256,20,20),... 13854/0/13854 185 \ 100.0%/0.0%/0.0% 200 Split:/model.21/Split
| 85 ConvExSwish INT8 NPU (1,64,40,40),(64,64,3,3),(64) (1,64,40,40) 8191/57600/57600 231 10.39/10.39/10.39 100.0%/0.0%/0.0% 136 Conv:/model.22/cv2.1/cv2.1.1/conv/Conv
| 86 Conv INT8 NPU (1,64,40,40),(64,64,1,1),(64) (1,64,40,40) 7083/12800/12800 179 1.49/1.49/1.49 35.0%/32.5%/32.5% 104 Conv:/model.22/cv2.1/cv2.1.2/Conv
| 87 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 444 21.62/21.62/21.62 100.0%/0.0%/0.0% 345 Conv:/model.22/cv3.1/cv3.1.1/conv/Conv
| 88 Conv INT8 NPU (1,128,40,40),(80,128,1,1),(80) (1,80,40,40) 11628/19200/19200 185 3.60/3.60/3.60 35.0%/32.5%/32.5% 210 Conv:/model.22/cv3.1/cv3.1.2/Conv
| 89 Concat INT8 NPU (1,64,40,40),(1,80,40,40) (1,144,40,40) 15585/0/15585 182 \ 35.0%/32.5%/32.5% 225 Concat:/model.22/Concat_1
| 90 OutputOperator INT8 CPU (1,144,40,40) \ 0/0/0 29 \ 0.0%/0.0%/0.0% 225 OutputOperator:output1
| 91 ConvExSwish INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 444 21.62/21.62/21.62 100.0%/0.0%/0.0% 678 Conv:/model.21/m.0/cv1/conv/Conv
| 92 ConvExSwish INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 447 21.48/21.48/21.48 100.0%/0.0%/0.0% 678 Conv:/model.21/m.0/cv2/conv/Conv
| 93 Concat INT8 NPU (1,256,20,20),(1,256,20,20),... (1,768,20,20) 20780/0/20780 143 \ 35.0%/35.0%/30.0% 300 Concat:/model.21/Concat
| 94 ConvExSwish INT8 NPU (1,768,20,20),(512,768,1,1),(512) (1,512,20,20) 30754/153600/153600 326 19.63/19.63/19.63 100.0%/0.0%/0.0% 688 Conv:/model.21/cv2/conv/Conv
| 95 ConvExSwish INT8 NPU (1,512,20,20),(64,512,3,3),(64) (1,64,20,20) 17784/115200/115200 332 14.46/14.46/14.46 100.0%/0.0%/0.0% 488 Conv:/model.22/cv2.2/cv2.2.0/conv/Conv
| 96 ConvExSwish INT8 NPU (1,512,20,20),(128,512,3,3),(128) (1,128,20,20) 28642/230400/230400 504 19.05/19.05/19.05 100.0%/0.0%/0.0% 777 Conv:/model.22/cv3.2/cv3.2.0/conv/Conv
| 97 ConvExSwish INT8 NPU (1,64,20,20),(64,64,3,3),(64) (1,64,20,20) 2996/14400/14400 176 3.41/3.41/3.41 100.0%/0.0%/0.0% 61 Conv:/model.22/cv2.2/cv2.2.1/conv/Conv
| 98 Conv INT8 NPU (1,64,20,20),(64,64,1,1),(64) (1,64,20,20) 1888/3200/3200 172 0.39/0.39/0.39 34.9%/34.9%/30.2% 29 Conv:/model.22/cv2.2/cv2.2.2/Conv
| 99 ConvExSwish INT8 NPU (1,128,20,20),(128,128,3,3),(128) (1,128,20,20) 8485/57600/57600 230 10.43/10.43/10.43 100.0%/0.0%/0.0% 195 Conv:/model.22/cv3.2/cv3.2.1/conv/Conv
| 100 Conv INT8 NPU (1,128,20,20),(80,128,1,1),(80) (1,80,20,20) 3187/4800/4800 173 0.96/0.96/0.96 34.9%/34.9%/30.2% 60 Conv:/model.22/cv3.2/cv3.2.2/Conv
| 101 Concat INT8 NPU (1,64,20,20),(1,80,20,20) (1,144,20,20) 3897/0/3897 167 \ 34.9%/34.9%/30.1% 56 Concat:/model.22/Concat_2
| 102 OutputOperator INT8 CPU (1,144,20,20) \ 0/0/0 13 \ 0.0%/0.0%/0.0% 56 OutputOperator:output2
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Total Operator Elapsed Per Frame Time(us): 41984
| Total Memory Read/Write Per Frame Size(KB): 72183.2
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| ---------------------------------------------------------------------------------------------------
| Operator Time Consuming Ranking Table
| ---------------------------------------------------------------------------------------------------
| OpType CallNumber CPUTime(us) GPUTime(us) NPUTime(us) TotalTime(us) TimeRatio(%)
| ---------------------------------------------------------------------------------------------------
| ConvExSwish 57 0 0 33249 33249 79.19%
| Concat 16 0 0 3342 3342 7.96%
| Split 8 0 0 1905 1905 4.54%
| Conv 6 0 0 1105 1105 2.63%
| Add 6 0 0 1090 1090 2.60%
| Resize 2 0 0 648 648 1.54%
| MaxPool 3 0 0 512 512 1.22%
| OutputOperator 3 129 0 0 129 0.31%
| InputOperator 1 4 0 0 4 0.01%
| ---------------------------------------------------------------------------------------------------
YOLOv5s
[*] ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Network Layer Information Table |
---|
ID OpType DataType Target InputShape OutputShape Cycles(DDR/NPU/Total) Time(us) MacUsage(%) WorkLoad(0/1/2) RW(KB) FullName |
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
1 InputOperator UINT8 CPU \ (1,3,640,640) 0/0/0 4 \ 0.0%/0.0%/0.0% 0 InputOperator:images |
2 Conv UINT8 NPU (1,3,640,640),(12,3,2,2),(12) (1,12,320,320) 97006/409600/409600 414 1.45/1.45/1.45 33.4%/33.4%/33.1% 1201 Conv:Conv_0 |
3 ConvRelu INT8 NPU (1,12,320,320),(32,12,3,3),(32) (1,32,320,320) 166402/921600/921600 517 27.85/27.85/27.85 33.4%/33.4%/33.1% 1604 Conv:Conv_1 |
4 ConvRelu INT8 NPU (1,32,320,320),(64,32,3,3),(64) (1,64,160,160) 166878/460800/460800 447 42.95/42.95/42.95 33.8%/33.1%/33.1% 3218 Conv:Conv_3 |
5 ConvRelu INT8 NPU (1,64,160,160),(32,64,1,1),(32) (1,32,160,160) 83197/102400/102400 249 8.57/8.57/8.57 33.8%/33.1%/33.1% 1602 Conv:Conv_5 |
6 ConvRelu INT8 NPU (1,32,160,160),(32,32,1,1),(32) (1,32,160,160) 55456/102400/102400 179 5.96/5.96/5.96 33.8%/33.1%/33.1% 801 Conv:Conv_7 |
7 ConvReluAdd INT8 NPU (1,32,160,160),(32,32,3,3),(32),... (1,32,160,160) 83439/230400/230400 262 36.64/36.64/36.64 33.8%/33.1%/33.1% 1609 Conv:Conv_9 |
8 ConvRelu INT8 NPU (1,64,160,160),(32,64,1,1),(32) (1,32,160,160) 83197/102400/102400 251 8.50/8.50/8.50 33.7%/33.1%/33.1% 1602 Conv:Conv_12 |
9 Concat INT8 NPU (1,32,160,160),(1,32,160,160) (1,64,160,160) 110825/0/110825 234 \ 33.7%/33.1%/33.1% 1600 Concat:Concat_14 |
10 ConvRelu INT8 NPU (1,64,160,160),(64,64,1,1),(64) (1,64,160,160) 110981/204800/204800 298 14.32/14.32/14.32 33.8%/33.1%/33.1% 1604 Conv:Conv_15 |
11 ConvRelu INT8 NPU (1,64,160,160),(128,64,3,3),(128) (1,128,80,80) 85647/460800/460800 387 49.61/49.61/49.61 33.8%/33.8%/32.5% 1673 Conv:Conv_17 |
12 ConvRelu INT8 NPU (1,128,80,80),(64,128,1,1),(64) (1,64,80,80) 41854/51200/51200 198 10.77/10.77/10.77 33.8%/33.8%/32.5% 808 Conv:Conv_19 |
13 ConvRelu INT8 NPU (1,64,80,80),(64,64,1,1),(64) (1,64,80,80) 27862/51200/51200 185 5.77/5.77/5.77 33.8%/33.8%/32.5% 404 Conv:Conv_21 |
14 ConvReluAdd INT8 NPU (1,64,80,80),(64,64,3,3),(64),... (1,64,80,80) 42824/230400/230400 252 38.10/38.10/38.10 33.8%/33.8%/32.5% 836 Conv:Conv_23 |
15 ConvRelu INT8 NPU (1,64,80,80),(64,64,1,1),(64) (1,64,80,80) 27862/51200/51200 143 7.46/7.46/7.46 33.8%/33.8%/32.5% 404 Conv:Conv_26 |
16 ConvReluAdd INT8 NPU (1,64,80,80),(64,64,3,3),(64),... (1,64,80,80) 42824/230400/230400 252 38.10/38.10/38.10 33.8%/33.8%/32.5% 836 Conv:Conv_28 |
17 ConvRelu INT8 NPU (1,128,80,80),(64,128,1,1),(64) (1,64,80,80) 41854/51200/51200 203 10.51/10.51/10.51 33.7%/33.7%/32.5% 808 Conv:Conv_31 |
18 Concat INT8 NPU (1,64,80,80),(1,64,80,80) (1,128,80,80) 55413/0/55413 164 \ 33.7%/33.7%/32.5% 800 Concat:Concat_33 |
19 ConvRelu INT8 NPU (1,128,80,80),(128,128,1,1),(128) (1,128,80,80) 56001/102400/102400 168 25.40/25.40/25.40 33.8%/33.8%/32.5% 817 Conv:Conv_34 |
20 ConvRelu INT8 NPU (1,128,80,80),(256,128,3,3),(256) (1,256,40,40) 51603/460800/460800 390 49.23/49.23/49.23 35.0%/32.5%/32.5% 1090 Conv:Conv_36 |
21 ConvRelu INT8 NPU (1,256,40,40),(128,256,1,1),(128) (1,128,40,40) 21923/51200/51200 184 11.59/11.59/11.59 35.0%/32.5%/32.5% 433 Conv:Conv_38 |
22 ConvRelu INT8 NPU (1,128,40,40),(128,128,1,1),(128) (1,128,40,40) 14442/25600/25600 168 6.35/6.35/6.35 35.0%/32.5%/32.5% 217 Conv:Conv_40 |
23 ConvReluAdd INT8 NPU (1,128,40,40),(128,128,3,3),(128),... (1,128,40,40) 25802/230400/230400 252 38.10/38.10/38.10 35.0%/32.5%/32.5% 545 Conv:Conv_42 |
24 ConvRelu INT8 NPU (1,128,40,40),(128,128,1,1),(128) (1,128,40,40) 14442/25600/25600 168 6.35/6.35/6.35 35.0%/32.5%/32.5% 217 Conv:Conv_45 |
25 ConvReluAdd INT8 NPU (1,128,40,40),(128,128,3,3),(128),... (1,128,40,40) 25802/230400/230400 251 38.25/38.25/38.25 35.0%/32.5%/32.5% 545 Conv:Conv_47 |
26 ConvRelu INT8 NPU (1,128,40,40),(128,128,1,1),(128) (1,128,40,40) 14442/25600/25600 174 6.13/6.13/6.13 35.0%/32.5%/32.5% 217 Conv:Conv_50 |
27 ConvReluAdd INT8 NPU (1,128,40,40),(128,128,3,3),(128),... (1,128,40,40) 25802/230400/230400 143 67.13/67.13/67.13 35.0%/32.5%/32.5% 545 Conv:Conv_52 |
28 ConvRelu INT8 NPU (1,256,40,40),(128,256,1,1),(128) (1,128,40,40) 21923/51200/51200 73 29.22/29.22/29.22 35.0%/32.5%/32.5% 433 Conv:Conv_55 |
29 Concat INT8 NPU (1,128,40,40),(1,128,40,40) (1,256,40,40) 27707/0/27707 66 \ 35.0%/32.5%/32.5% 400 Concat:Concat_57 |
30 ConvRelu INT8 NPU (1,256,40,40),(256,256,1,1),(256) (1,256,40,40) 29992/102400/102400 93 45.88/45.88/45.88 35.0%/32.5%/32.5% 466 Conv:Conv_58 |
31 ConvRelu INT8 NPU (1,256,40,40),(512,256,3,3),(512) (1,512,20,20) 60815/460800/460800 301 63.79/63.79/63.79 37.5%/31.2%/31.2% 1556 Conv:Conv_60 |
32 ConvRelu INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 66 32.32/32.32/32.32 35.0%/35.0%/30.0% 330 Conv:Conv_62 |
33 ConvRelu INT8 NPU (1,256,20,20),(256,256,1,1),(256) (1,256,20,20) 9213/25600/25600 105 10.16/10.16/10.16 35.0%/35.0%/30.0% 166 Conv:Conv_64 |
34 ConvReluAdd INT8 NPU (1,256,20,20),(256,256,3,3),(256),... (1,256,20,20) 30408/230400/230400 182 52.75/52.75/52.75 37.5%/37.5%/25.0% 778 Conv:Conv_66 |
35 ConvRelu INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 180 11.85/11.85/11.85 35.0%/35.0%/30.1% 330 Conv:Conv_69 |
36 Concat INT8 NPU (1,256,20,20),(1,256,20,20) (1,512,20,20) 13854/0/13854 117 \ 35.0%/35.0%/30.0% 200 Concat:Concat_71 |
37 ConvRelu INT8 NPU (1,512,20,20),(512,512,1,1),(512) (1,512,20,20) 22858/102400/102400 154 27.71/27.71/27.71 37.5%/31.2%/31.2% 460 Conv:Conv_72 |
38 ConvRelu INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 181 11.79/11.79/11.79 35.0%/35.0%/30.1% 330 Conv:Conv_74 |
39 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 165 \ 100.0%/0.0%/0.0% 100 MaxPool:MaxPool_76 |
40 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 121 \ 100.0%/0.0%/0.0% 100 MaxPool:MaxPool_77 |
41 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 161 \ 100.0%/0.0%/0.0% 100 MaxPool:MaxPool_78 |
42 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 120 \ 100.0%/0.0%/0.0% 100 MaxPool:MaxPool_79 |
43 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 109 \ 100.0%/0.0%/0.0% 100 MaxPool:MaxPool_80 |
44 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 109 \ 100.0%/0.0%/0.0% 100 MaxPool:MaxPool_81 |
45 Concat INT8 NPU (1,256,20,20),(1,256,20,20),... (1,1024,20,20) 27707/0/27707 123 \ 35.0%/35.0%/30.0% 400 Concat:Concat_82 |
46 ConvRelu INT8 NPU (1,1024,20,20),(512,1024,1,1),(512) (1,512,20,20) 38651/204800/204800 226 37.76/37.76/37.76 37.5%/31.2%/31.2% 916 Conv:Conv_83 |
47 ConvRelu INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 128 16.67/16.67/16.67 35.0%/35.0%/30.1% 330 Conv:Conv_85 |
48 ConvTranspose INT8 NPU (1,256,20,20),(256,1,4,4),(256) (1,256,40,40) 17525/102400/102400 211 1.26/1.26/1.26 100.0%/0.0%/0.0% 106 ConvTranspose:Resize_88_2deconv |
49 Concat INT8 NPU (1,256,40,40),(1,256,40,40) (1,512,40,40) 55413/0/55413 161 \ 35.0%/32.5%/32.5% 800 Concat:Concat_89 |
50 ConvRelu INT8 NPU (1,512,40,40),(128,512,1,1),(128) (1,128,40,40) 36884/102400/102400 162 26.34/26.34/26.34 35.0%/32.5%/32.5% 865 Conv:Conv_90 |
51 ConvRelu INT8 NPU (1,128,40,40),(128,128,1,1),(128) (1,128,40,40) 14442/25600/25600 167 6.39/6.39/6.39 35.0%/32.5%/32.5% 217 Conv:Conv_92 |
52 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 259 37.07/37.07/37.07 35.0%/32.5%/32.5% 345 Conv:Conv_94 |
53 ConvRelu INT8 NPU (1,512,40,40),(128,512,1,1),(128) (1,128,40,40) 36884/102400/102400 220 19.39/19.39/19.39 35.0%/32.5%/32.5% 865 Conv:Conv_96 |
54 Concat INT8 NPU (1,128,40,40),(1,128,40,40) (1,256,40,40) 27707/0/27707 179 \ 35.0%/32.5%/32.5% 400 Concat:Concat_98 |
55 ConvRelu INT8 NPU (1,256,40,40),(256,256,1,1),(256) (1,256,40,40) 29992/102400/102400 205 20.81/20.81/20.81 35.0%/32.5%/32.5% 466 Conv:Conv_99 |
56 ConvRelu INT8 NPU (1,256,40,40),(128,256,1,1),(128) (1,128,40,40) 21923/51200/51200 185 11.53/11.53/11.53 35.0%/32.5%/32.5% 433 Conv:Conv_101 |
57 ConvTranspose INT8 NPU (1,128,40,40),(128,1,4,4),(128) (1,128,80,80) 34737/204800/204800 353 1.51/1.51/1.51 100.0%/0.0%/0.0% 203 ConvTranspose:Resize_104_2deconv |
58 Concat INT8 NPU (1,128,80,80),(1,128,80,80) (1,256,80,80) 110825/0/110825 226 \ 33.7%/33.7%/32.5% 1600 Concat:Concat_105 |
59 ConvRelu INT8 NPU (1,256,80,80),(64,256,1,1),(64) (1,64,80,80) 69837/102400/102400 191 22.34/22.34/22.34 33.8%/33.8%/32.5% 1616 Conv:Conv_106 |
60 ConvRelu INT8 NPU (1,64,80,80),(64,64,1,1),(64) (1,64,80,80) 27862/51200/51200 185 5.77/5.77/5.77 33.8%/33.8%/32.5% 404 Conv:Conv_108 |
61 ConvRelu INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 252 38.10/38.10/38.10 33.8%/33.8%/32.5% 436 Conv:Conv_110 |
62 ConvRelu INT8 NPU (1,256,80,80),(64,256,1,1),(64) (1,64,80,80) 69837/102400/102400 248 17.20/17.20/17.20 33.7%/33.7%/32.5% 1616 Conv:Conv_112 |
63 Concat INT8 NPU (1,64,80,80),(1,64,80,80) (1,128,80,80) 55413/0/55413 209 \ 33.7%/33.7%/32.5% 800 Concat:Concat_114 |
64 ConvRelu INT8 NPU (1,128,80,80),(128,128,1,1),(128) (1,128,80,80) 56001/102400/102400 223 19.13/19.13/19.13 33.8%/33.8%/32.5% 817 Conv:Conv_115 |
65 ConvRelu INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,40,40) 39655/230400/230400 288 33.33/33.33/33.33 35.0%/32.5%/32.5% 945 Conv:Conv_117 |
66 Concat INT8 NPU (1,128,40,40),(1,128,40,40) (1,256,40,40) 27707/0/27707 137 \ 35.0%/32.5%/32.5% 400 Concat:Concat_119 |
67 ConvRelu INT8 NPU (1,256,40,40),(128,256,1,1),(128) (1,128,40,40) 21923/51200/51200 181 11.79/11.79/11.79 35.0%/32.5%/32.5% 433 Conv:Conv_120 |
68 ConvRelu INT8 NPU (1,128,40,40),(128,128,1,1),(128) (1,128,40,40) 14442/25600/25600 170 6.27/6.27/6.27 35.0%/32.5%/32.5% 217 Conv:Conv_122 |
69 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 252 38.10/38.10/38.10 35.0%/32.5%/32.5% 345 Conv:Conv_124 |
70 ConvRelu INT8 NPU (1,256,40,40),(128,256,1,1),(128) (1,128,40,40) 21923/51200/51200 185 11.53/11.53/11.53 35.0%/32.5%/32.5% 433 Conv:Conv_126 |
71 Concat INT8 NPU (1,128,40,40),(1,128,40,40) (1,256,40,40) 27707/0/27707 177 \ 35.0%/32.5%/32.5% 400 Concat:Concat_128 |
72 ConvRelu INT8 NPU (1,256,40,40),(256,256,1,1),(256) (1,256,40,40) 29992/102400/102400 205 20.81/20.81/20.81 35.0%/32.5%/32.5% 466 Conv:Conv_129 |
73 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,20,20) 37334/230400/230400 259 37.07/37.07/37.07 37.4%/37.4%/25.1% 978 Conv:Conv_131 |
74 Concat INT8 NPU (1,256,20,20),(1,256,20,20) (1,512,20,20) 13854/0/13854 107 \ 35.0%/35.0%/30.0% 200 Concat:Concat_133 |
75 ConvRelu INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 123 17.34/17.34/17.34 35.0%/35.0%/30.0% 330 Conv:Conv_134 |
76 ConvRelu INT8 NPU (1,256,20,20),(256,256,1,1),(256) (1,256,20,20) 9213/25600/25600 109 9.79/9.79/9.79 35.0%/35.0%/30.0% 166 Conv:Conv_136 |
77 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 178 53.93/53.93/53.93 37.5%/37.5%/25.0% 678 Conv:Conv_138 |
78 ConvRelu INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 127 16.80/16.80/16.80 35.0%/35.0%/30.1% 330 Conv:Conv_140 |
79 Concat INT8 NPU (1,256,20,20),(1,256,20,20) (1,512,20,20) 13854/0/13854 105 \ 35.0%/35.0%/30.0% 200 Concat:Concat_142 |
80 ConvRelu INT8 NPU (1,512,20,20),(512,512,1,1),(512) (1,512,20,20) 22858/102400/102400 161 26.50/26.50/26.50 37.5%/31.3%/31.3% 460 Conv:Conv_143 |
81 ConvSigmoid INT8 NPU (1,512,20,20),(255,512,1,1),(255) (1,255,20,20) 14893/51200/51200 158 13.45/13.45/13.45 100.0%/0.0%/0.0% 330 Conv:Conv_149 |
82 OutputOperator INT8 CPU (1,255,20,20) \ 0/0/0 7 \ 0.0%/0.0%/0.0% 100 OutputOperator:285 |
83 ConvSigmoid INT8 NPU (1,256,40,40),(255,256,1,1),(255) (1,255,40,40) 29992/102400/102400 356 11.94/11.94/11.94 100.0%/0.0%/0.0% 466 Conv:Conv_147 |
84 OutputOperator INT8 CPU (1,255,40,40) \ 0/0/0 18 \ 0.0%/0.0%/0.0% 400 OutputOperator:283 |
85 ConvSigmoid INT8 NPU (1,128,80,80),(255,128,1,1),(255) (1,255,80,80) 84296/204800/204800 1180 7.20/7.20/7.20 100.0%/0.0%/0.0% 834 Conv:Conv_145 |
86 OutputOperator INT8 CPU (1,255,80,80) \ 0/0/0 73 \ 0.0%/0.0%/0.0% 1600 OutputOperator:output |
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
Total Operator Elapsed Per Frame Time(us): 17509 |
Total Memory Read/Write Per Frame Size(KB): 56438.8 |
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
--------------------------------------------------------------------------------------------------- |
Operator Time Consuming Ranking Table |
--------------------------------------------------------------------------------------------------- |
OpType CallNumber CPUTime(us) GPUTime(us) NPUTime(us) TotalTime(us) TimeRatio(%) |
--------------------------------------------------------------------------------------------------- |
ConvRelu 50 0 0 10351 10351 59.12% |
Concat 13 0 0 2005 2005 11.45% |
ConvSigmoid 3 0 0 1694 1694 9.68% |
ConvReluAdd 7 0 0 1594 1594 9.10% |
MaxPool 6 0 0 785 785 4.48% |
ConvTranspose 2 0 0 564 564 3.22% |
Conv 1 0 0 414 414 2.36% |
OutputOperator 3 98 0 0 98 0.56% |
InputOperator 1 4 0 0 4 0.02% |
--------------------------------------------------------------------------------------------------- |
Gold-YOLO-s
[*] ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Network Layer Information Table
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| ID OpType DataType Target InputShape OutputShape Cycles(DDR/NPU/Total) Time(us) MacUsage(%) WorkLoad(0/1/2) RW(KB) FullName
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| 1 InputOperator UINT8 CPU \ (1,3,640,640) 0/0/0 4 \ 0.0%/0.0%/0.0% 0 InputOperator:images
| 2 ConvRelu UINT8 NPU (1,3,640,640),(32,3,3,3),(32) (1,32,320,320) 152549/921600/921600 620 5.81/5.81/5.81 33.4%/33.4%/33.1% 1204 Conv:/backbone/stem/rbr_reparam/Conv
| 3 ConvRelu INT8 NPU (1,32,320,320),(64,32,3,3),(64) (1,64,160,160) 166878/460800/460800 384 50.00/50.00/50.00 33.8%/33.1%/33.1% 3218 Conv:/backbone/ERBlock_2/ERBlock_2.0/rbr_reparam/Conv
| 4 ConvRelu INT8 NPU (1,64,160,160),(64,64,3,3),(64) (1,64,160,160) 112089/921600/921600 567 67.72/67.72/67.72 33.8%/33.1%/33.1% 1636 Conv:/backbone/ERBlock_2/ERBlock_2.1/conv1/rbr_reparam/Conv
| 5 ConvRelu INT8 NPU (1,64,160,160),(64,64,3,3),(64) (1,64,160,160) 112089/921600/921600 568 67.61/67.61/67.61 33.7%/33.1%/33.1% 1636 Conv:/backbone/ERBlock_2/ERBlock_2.1/block/block.0/rbr_reparam/Conv
| 6 AveragePool INT8 NPU (1,64,160,160) (1,64,40,40) 58876/0/58876 635 \ 100.0%/0.0%/0.0% 1600 AveragePool:/neck/low_FAM/AveragePool
| 7 AveragePool INT8 NPU (1,64,160,160) (1,64,80,80) 69266/0/69266 635 \ 100.0%/0.0%/0.0% 1600 AveragePool:/neck/LAF_p3/AveragePool
| 8 ConvRelu INT8 NPU (1,64,160,160),(128,64,3,3),(128) (1,128,80,80) 85647/460800/460800 388 49.48/49.48/49.48 33.8%/33.8%/32.5% 1673 Conv:/backbone/ERBlock_3/ERBlock_3.0/rbr_reparam/Conv
| 9 ConvRelu INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,80,80) 60434/921600/921600 564 68.09/68.09/68.09 33.8%/33.8%/32.5% 945 Conv:/backbone/ERBlock_3/ERBlock_3.1/conv1/rbr_reparam/Conv
| 10 ConvRelu INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,80,80) 60434/921600/921600 562 68.33/68.33/68.33 33.8%/33.8%/32.5% 945 Conv:/backbone/ERBlock_3/ERBlock_3.1/block/block.0/rbr_reparam/Conv
| 11 ConvRelu INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,80,80) 60434/921600/921600 564 68.09/68.09/68.09 33.8%/33.8%/32.5% 945 Conv:/backbone/ERBlock_3/ERBlock_3.1/block/block.1/rbr_reparam/Conv
| 12 ConvRelu INT8 NPU (1,128,80,80),(128,128,3,3),(128) (1,128,80,80) 60434/921600/921600 564 68.09/68.09/68.09 33.7%/33.7%/32.5% 945 Conv:/backbone/ERBlock_3/ERBlock_3.1/block/block.2/rbr_reparam/Conv
| 13 AveragePool INT8 NPU (1,128,80,80) (1,128,40,40) 34633/0/34633 383 \ 100.0%/0.0%/0.0% 800 AveragePool:/neck/low_FAM/AveragePool_1
| 14 ConvRelu INT8 NPU (1,128,80,80),(256,128,3,3),(256) (1,256,40,40) 51603/460800/460800 393 48.85/48.85/48.85 35.0%/32.5%/32.5% 1090 Conv:/backbone/ERBlock_4/ERBlock_4.0/rbr_reparam/Conv
| 15 ConvRelu INT8 NPU (1,128,80,80),(64,128,1,1),(64) (1,64,80,80) 41854/51200/51200 199 10.72/10.72/10.72 33.8%/33.8%/32.5% 808 Conv:/neck/LAF_p3/cv1/conv/Conv
| 16 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,40,40) 47724/921600/921600 619 62.04/62.04/62.04 37.5%/37.5%/25.0% 978 Conv:/backbone/ERBlock_4/ERBlock_4.1/conv1/rbr_reparam/Conv
| 17 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,40,40) 47724/921600/921600 616 62.34/62.34/62.34 37.5%/37.5%/25.0% 978 Conv:/backbone/ERBlock_4/ERBlock_4.1/block/block.0/rbr_reparam/Conv
| 18 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,40,40) 47724/921600/921600 620 61.94/61.94/61.94 37.5%/37.5%/25.0% 978 Conv:/backbone/ERBlock_4/ERBlock_4.1/block/block.1/rbr_reparam/Conv
| 19 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,40,40) 47724/921600/921600 559 68.69/68.69/68.69 37.5%/37.5%/25.0% 978 Conv:/backbone/ERBlock_4/ERBlock_4.1/block/block.2/rbr_reparam/Conv
| 20 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,40,40) 47724/921600/921600 554 69.31/69.31/69.31 37.5%/37.5%/25.0% 978 Conv:/backbone/ERBlock_4/ERBlock_4.1/block/block.3/rbr_reparam/Conv
| 21 ConvRelu INT8 NPU (1,256,40,40),(256,256,3,3),(256) (1,256,40,40) 47724/921600/921600 621 61.84/61.84/61.84 37.5%/37.5%/25.0% 978 Conv:/backbone/ERBlock_4/ERBlock_4.1/block/block.4/rbr_reparam/Conv
| 22 ConvRelu INT8 NPU (1,256,40,40),(512,256,3,3),(512) (1,512,20,20) 60815/460800/460800 406 47.29/47.29/47.29 37.5%/31.2%/31.2% 1556 Conv:/backbone/ERBlock_5/ERBlock_5.0/rbr_reparam/Conv
| 23 ConvRelu INT8 NPU (1,256,40,40),(128,256,1,1),(128) (1,128,40,40) 21923/51200/51200 184 11.59/11.59/11.59 35.0%/32.5%/32.5% 433 Conv:/neck/LAF_p4/cv1/conv/Conv
| 24 ConvRelu INT8 NPU (1,512,20,20),(512,512,3,3),(512) (1,512,20,20) 93786/921600/921600 646 59.44/59.44/59.44 37.5%/31.2%/31.2% 2508 Conv:/backbone/ERBlock_5/ERBlock_5.1/conv1/rbr_reparam/Conv
| 25 ConvRelu INT8 NPU (1,512,20,20),(512,512,3,3),(512) (1,512,20,20) 93786/921600/921600 650 59.08/59.08/59.08 37.5%/31.2%/31.2% 2508 Conv:/backbone/ERBlock_5/ERBlock_5.1/block/block.0/rbr_reparam/Conv
| 26 ConvRelu INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 178 11.99/11.99/11.99 35.0%/35.0%/30.0% 330 Conv:/backbone/ERBlock_5/ERBlock_5.2/cv1/conv/Conv
| 27 ConvRelu INT8 NPU (1,512,20,20),(256,512,1,1),(256) (1,256,20,20) 14893/51200/51200 179 11.92/11.92/11.92 35.0%/35.0%/30.0% 330 Conv:/backbone/ERBlock_5/ERBlock_5.2/cv2/conv/Conv
| 28 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 232 41.38/41.38/41.38 37.5%/37.5%/25.0% 678 Conv:/backbone/ERBlock_5/ERBlock_5.2/cv3/conv/Conv
| 29 ConvRelu INT8 NPU (1,256,20,20),(256,256,1,1),(256) (1,256,20,20) 9213/25600/25600 170 6.27/6.27/6.27 35.0%/35.0%/30.1% 166 Conv:/backbone/ERBlock_5/ERBlock_5.2/cv4/conv/Conv
| 30 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 157 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/ERBlock_5/ERBlock_5.2/m/MaxPool
| 31 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 119 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/ERBlock_5/ERBlock_5.2/m_1/MaxPool
| 32 MaxPool INT8 NPU (1,256,20,20) (1,256,20,20) 6927/0/6927 116 \ 100.0%/0.0%/0.0% 100 MaxPool:/backbone/ERBlock_5/ERBlock_5.2/m_2/MaxPool
| 33 Concat INT8 NPU (1,256,20,20),(1,256,20,20),... (1,1024,20,20) 27707/0/27707 124 \ 35.0%/35.0%/30.0% 400 Concat:/backbone/ERBlock_5/ERBlock_5.2/Concat
| 34 ConvRelu INT8 NPU (1,1024,20,20),(256,1024,1,1),(256) (1,256,20,20) 26252/102400/102400 159 26.83/26.83/26.83 35.0%/35.0%/30.0% 658 Conv:/backbone/ERBlock_5/ERBlock_5.2/cv5/conv/Conv
| 35 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 215 44.65/44.65/44.65 37.4%/37.4%/25.1% 678 Conv:/backbone/ERBlock_5/ERBlock_5.2/cv6/conv/Conv
| 36 Concat INT8 NPU (1,256,20,20),(1,256,20,20) (1,512,20,20) 13854/0/13854 106 \ 35.0%/35.0%/30.0% 200 Concat:/backbone/ERBlock_5/ERBlock_5.2/Concat_1
| 37 ConvRelu INT8 NPU (1,512,20,20),(512,512,1,1),(512) (1,512,20,20) 22858/102400/102400 160 26.67/26.67/26.67 37.5%/31.3%/31.3% 460 Conv:/backbone/ERBlock_5/ERBlock_5.2/cv7/conv/Conv
| 38 AveragePool INT8 NPU (1,512,20,20) (1,512,10,10) 8659/0/8659 193 \ 100.0%/0.0%/0.0% 200 AveragePool:/neck/high_FAM/AveragePool_2
| 39 ConvTranspose INT8 NPU (1,512,20,20),(512,1,4,4),(512) (1,512,40,40) 35049/204800/204800 356 1.50/1.50/1.50 100.0%/0.0%/0.0% 212 ConvTranspose:/neck/low_FAM/Resize_2deconv
| 40 Concat INT8 NPU (1,64,40,40),(1,128,40,40),... (1,960,40,40) 103898/0/103898 278 \ 35.0%/32.5%/32.5% 1500 Concat:/neck/low_FAM/Concat_1
| 41 ConvRelu INT8 NPU (1,512,20,20),(128,512,1,1),(128) (1,128,20,20) 10910/25600/25600 170 6.27/6.27/6.27 34.9%/34.9%/30.1% 265 Conv:/neck/reduce_layer_c5/conv/Conv
| 42 ConvTranspose INT8 NPU (1,128,20,20),(128,1,4,4),(128) (1,128,40,40) 8763/51200/51200 201 0.66/0.66/0.66 100.0%/0.0%/0.0% 53 ConvTranspose:/neck/LAF_p4/Resize_2deconv
| 43 Concat INT8 NPU (1,128,40,40),(1,128,40,40),... (1,384,40,40) 41560/0/41560 198 \ 35.0%/32.5%/32.5% 600 Concat:/neck/LAF_p4/Concat_1
| 44 ConvExSwish INT8 NPU (1,960,40,40),(128,960,1,1),(128) (1,128,40,40) 63066/192000/192000 525 15.24/15.24/15.24 100.0%/0.0%/0.0% 1621 Conv:/neck/low_IFM/low_IFM.0/conv/Conv
| 45 ConvRelu INT8 NPU (1,384,40,40),(128,384,1,1),(128) (1,128,40,40) 29404/76800/76800 202 15.84/15.84/15.84 35.0%/32.5%/32.5% 649 Conv:/neck/LAF_p4/cv_fuse/conv/Conv
| 46 Conv INT8 NPU (1,128,40,40),(128,128,1,1),(128) (1,128,40,40) 14442/25600/25600 169 6.31/6.31/6.31 35.0%/32.5%/32.5% 217 Conv:/neck/Inject_p4/local_embedding/conv/Conv
| 47 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 210 45.71/45.71/45.71 35.0%/32.5%/32.5% 345 Conv:/neck/low_IFM/low_IFM.1/rbr_reparam/Conv
| 48 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 199 48.24/48.24/48.24 35.0%/32.5%/32.5% 345 Conv:/neck/low_IFM/low_IFM.2/rbr_reparam/Conv
| 49 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 258 37.21/37.21/37.21 35.0%/32.5%/32.5% 345 Conv:/neck/low_IFM/low_IFM.3/rbr_reparam/Conv
| 50 ConvExSwish INT8 NPU (1,128,40,40),(192,128,1,1),(192) (1,192,40,40) 18200/38400/38400 334 4.79/4.79/4.79 100.0%/0.0%/0.0% 225 Conv:/neck/low_IFM/low_IFM.4/conv/Conv
| 51 Split INT8 NPU (1,192,40,40),(2) (1,128,40,40),... 20781/0/20781 183 \ 100.0%/0.0%/0.0% 300 Split:/neck/Split
| 52 Conv INT8 NPU (1,128,40,40),(128,128,1,1),(128) (1,128,40,40) 14442/25600/25600 171 6.24/6.24/6.24 35.0%/32.5%/32.5% 217 Conv:/neck/Inject_p4/global_act/conv/Conv
| 53 Add INT8 NPU (1,128,40,40),(1) (1,128,40,40) 13854/0/13854 161 \ 35.0%/32.5%/32.5% 200 Add:/neck/Inject_p4/act/Add
| 54 Clip INT8 NPU (1,128,40,40),(1),(1) (1,128,40,40) 13854/0/13854 160 \ 35.0%/32.5%/32.5% 200 Clip:/neck/Inject_p4/act/relu/Clip
| 55 Mul INT8 NPU (1,128,40,40),(1) (1,128,40,40) 13854/0/13854 167 \ 100.0%/0.0%/0.0% 200 Mul:/neck/Inject_p4/act/Div_2mul
| 56 Conv INT8 NPU (1,64,40,40),(64,64,1,1),(64) (1,64,40,40) 7083/12800/12800 116 2.30/2.30/2.30 35.0%/32.5%/32.5% 104 Conv:/neck/Inject_p3/global_act/conv/Conv
| 57 Conv INT8 NPU (1,64,40,40),(64,64,1,1),(64) (1,64,40,40) 7083/12800/12800 106 2.52/2.52/2.52 35.0%/32.5%/32.5% 104 Conv:/neck/Inject_p3/global_embedding/conv/Conv
| 58 ConvTranspose INT8 NPU (1,64,40,40),(64,1,4,4),(64) (1,64,80,80) 17369/102400/102400 211 1.26/1.26/1.26 100.0%/0.0%/0.0% 101 ConvTranspose:/neck/Inject_p3/Resize_1_2deconv
| 59 Mul INT8 NPU (1,128,40,40),(1,128,40,40) (1,128,40,40) 20780/0/20780 121 \ 100.0%/0.0%/0.0% 400 Mul:/neck/Inject_p4/Mul
| 60 ConvAdd INT8 NPU (1,128,40,40),(128,128,1,1),(128),... (1,128,40,40) 21369/25600/25600 117 9.12/9.12/9.12 35.0%/32.5%/32.5% 417 Conv:/neck/Inject_p4/global_embedding/conv/Conv
| 61 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 199 48.24/48.24/48.24 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_p4/conv1/rbr_reparam/Conv
| 62 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 197 48.73/48.73/48.73 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_p4/block/block.0/rbr_reparam/Conv
| 63 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 251 38.25/38.25/38.25 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_p4/block/block.1/rbr_reparam/Conv
| 64 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 260 36.92/36.92/36.92 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_p4/block/block.2/rbr_reparam/Conv
| 65 AveragePool INT8 NPU (1,128,40,40) (1,128,10,10) 7360/0/7360 195 \ 100.0%/0.0%/0.0% 200 AveragePool:/neck/high_FAM/AveragePool_1
| 66 ConvRelu INT8 NPU (1,128,40,40),(64,128,1,1),(64) (1,64,40,40) 10685/12800/12800 120 4.44/4.44/4.44 35.0%/32.5%/32.5% 208 Conv:/neck/reduce_layer_p4/conv/Conv
| 67 ConvTranspose INT8 NPU (1,64,40,40),(64,1,4,4),(64) (1,64,80,80) 17369/102400/102400 211 1.26/1.26/1.26 100.0%/0.0%/0.0% 101 ConvTranspose:/neck/LAF_p3/Resize_2deconv
| 68 Concat INT8 NPU (1,64,80,80),(1,64,80,80),... (1,192,80,80) 83119/0/83119 243 \ 33.7%/33.7%/32.5% 1200 Concat:/neck/LAF_p3/Concat_1
| 69 ConvRelu INT8 NPU (1,192,80,80),(64,192,1,1),(64) (1,64,80,80) 55846/76800/76800 178 17.98/17.98/17.98 33.8%/33.8%/32.5% 1212 Conv:/neck/LAF_p3/cv_fuse/conv/Conv
| 70 Conv INT8 NPU (1,64,80,80),(64,64,1,1),(64) (1,64,80,80) 27862/51200/51200 187 5.70/5.70/5.70 33.7%/33.7%/32.5% 404 Conv:/neck/Inject_p3/local_embedding/conv/Conv
| 71 Add INT8 NPU (1,64,40,40),(1) (1,64,40,40) 6927/0/6927 108 \ 35.0%/32.5%/32.5% 100 Add:/neck/Inject_p3/act/Add
| 72 Clip INT8 NPU (1,64,40,40),(1),(1) (1,64,40,40) 6927/0/6927 96 \ 35.0%/32.5%/32.5% 100 Clip:/neck/Inject_p3/act/relu/Clip
| 73 Mul INT8 NPU (1,64,40,40),(1) (1,64,40,40) 6927/0/6927 95 \ 100.0%/0.0%/0.0% 100 Mul:/neck/Inject_p3/act/Div_2mul
| 74 ConvTranspose INT8 NPU (1,64,40,40),(64,1,4,4),(64) (1,64,80,80) 17369/102400/102400 210 1.27/1.27/1.27 100.0%/0.0%/0.0% 101 ConvTranspose:/neck/Inject_p3/Resize_2deconv
| 75 Mul INT8 NPU (1,64,80,80),(1,64,80,80) (1,64,80,80) 41560/0/41560 220 \ 100.0%/0.0%/0.0% 800 Mul:/neck/Inject_p3/Mul
| 76 Add INT8 NPU (1,64,80,80),(1,64,80,80) (1,64,80,80) 41560/0/41560 143 \ 33.7%/33.7%/32.5% 800 Add:/neck/Inject_p3/Add
| 77 ConvRelu INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 197 48.73/48.73/48.73 33.8%/33.8%/32.5% 436 Conv:/neck/Rep_p3/conv1/rbr_reparam/Conv
| 78 ConvRelu INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 251 38.25/38.25/38.25 33.8%/33.8%/32.5% 436 Conv:/neck/Rep_p3/block/block.0/rbr_reparam/Conv
| 79 ConvRelu INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 251 38.25/38.25/38.25 33.8%/33.8%/32.5% 436 Conv:/neck/Rep_p3/block/block.1/rbr_reparam/Conv
| 80 ConvRelu INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 257 37.35/37.35/37.35 33.7%/33.7%/32.5% 436 Conv:/neck/Rep_p3/block/block.2/rbr_reparam/Conv
| 81 AveragePool INT8 NPU (1,64,80,80) (1,64,10,10) 14070/0/14070 258 \ 100.0%/0.0%/0.0% 400 AveragePool:/neck/high_FAM/AveragePool
| 82 Concat INT8 NPU (1,64,10,10),(1,128,10,10),... (1,704,10,10) 4762/0/4762 107 \ 39.8%/30.1%/30.1% 68 Concat:/neck/high_FAM/Concat
| 83 Conv INT8 NPU (1,704,10,10),(32,704,1,1),(32) (1,32,10,10) 3260/2464/3260 100 0.92/0.92/0.92 40.0%/30.0%/30.0% 91 Conv:/neck/high_IFM/transformer_blocks.0/attn/to_q/c/Conv
| 84 Conv INT8 NPU (1,704,10,10),(32,704,1,1),(32) (1,32,10,10) 3260/2464/3260 100 0.92/0.92/0.92 40.0%/30.0%/30.0% 91 Conv:/neck/high_IFM/transformer_blocks.0/attn/to_k/c/Conv
| 85 Conv INT8 NPU (1,704,10,10),(64,704,1,1),(64) (1,64,10,10) 4139/4928/4928 101 1.82/1.82/1.82 38.5%/30.7%/30.7% 113 Conv:/neck/high_IFM/transformer_blocks.0/attn/to_v/c/Conv
| 86 AveragePool INT8 NPU (1,64,80,80) (1,64,40,40) 17317/0/17317 204 \ 100.0%/0.0%/0.0% 400 AveragePool:/neck/LAF_n4/AveragePool
| 87 Concat INT8 NPU (1,64,40,40),(1,64,40,40) (1,128,40,40) 13854/0/13854 105 \ 35.0%/32.5%/32.5% 200 Concat:/neck/LAF_n4/Concat
| 88 Conv INT8 NPU (1,128,40,40),(128,128,1,1),(128) (1,128,40,40) 14442/25600/25600 116 9.20/9.20/9.20 35.0%/32.5%/32.5% 217 Conv:/neck/Inject_n4/local_embedding/conv/Conv
| 89 ConvExSwish INT8 NPU (1,64,80,80),(64,64,1,1),(64) (1,64,80,80) 27862/51200/51200 348 3.07/3.07/3.07 100.0%/0.0%/0.0% 404 Conv:/detect/stems.0/conv/Conv
| 90 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 379 25.33/25.33/25.33 100.0%/0.0%/0.0% 436 Conv:/detect/cls_convs.0/conv/Conv
| 91 Conv INT8 NPU (1,64,80,80),(80,64,1,1),(80) (1,80,80,80) 31369/76800/76800 199 6.70/6.70/6.70 33.7%/33.7%/32.5% 405 Conv:/detect/cls_preds.0/Conv
| 92 ConvExSwish INT8 NPU (1,64,80,80),(64,64,3,3),(64) (1,64,80,80) 28971/230400/230400 436 22.02/22.02/22.02 100.0%/0.0%/0.0% 436 Conv:/detect/reg_convs.0/conv/Conv
| 93 Conv INT8 NPU (1,64,80,80),(4,64,1,1),(4) (1,4,80,80) 17334/25600/25600 176 0.38/0.38/0.38 33.7%/33.7%/32.5% 400 Conv:/detect/reg_preds_lrtb.0/Conv
| 94 Concat INT8 CPU (1,4,80,80),(1,80,80,80) (1,84,80,80) 0/0/0 959 \ 0.0%/0.0%/0.0% 600 Concat:/detect/Concat
| 95 OutputOperator INT8 CPU (1,84,80,80) \ 0/0/0 27 \ 0.0%/0.0%/0.0% 525 OutputOperator:output0
| 96 Reshape INT8 CPU (1,32,10,10),(4) (4,8,1,100) 0/0/0 47 \ 0.0%/0.0%/0.0% 3 Reshape:/neck/high_IFM/transformer_blocks.0/attn/Transpose_output_0_Q_tp_for_/neck/high_IFM/transformer_blocks.0/attn/MatMul_1_exsdpa_rs_2reshape
| 97 exDataConvert INT8 CPU (4,8,1,100) (4,8,1,100) 0/0/0 24 \ 0.0%/0.0%/0.0% 3 exDataConvert:/neck/high_IFM/transformer_blocks.0/attn/Transpose_output_0_Q_tp_for_/neck/high_IFM/transformer_blocks.0/attn/MatMul_1_exsd__1__cvt_int8_float16
| 98 Reshape INT8 CPU (1,32,10,10),(4) (4,8,1,100) 0/0/0 36 \ 0.0%/0.0%/0.0% 3 Reshape:/neck/high_IFM/transformer_blocks.0/attn/Reshape_1_output_0_K_rs_for_/neck/high_IFM/transformer_blocks.0/attn/MatMul_1_exsdpa
| 99 exDataConvert INT8 CPU (4,8,1,100) (4,8,1,100) 0/0/0 24 \ 0.0%/0.0%/0.0% 3 exDataConvert:/neck/high_IFM/transformer_blocks.0/attn/Reshape_1_output_0_K_rs_for_/neck/high_IFM/transformer_blocks.0/attn/MatMul_1_exsdpa__cvt_int8_float16
| 100 Reshape INT8 NPU (1,64,10,10),(4) (4,16,1,100) 434/0/434 1 \ 0.0%/0.0%/0.0% 6 Reshape:/neck/high_IFM/transformer_blocks.0/attn/Transpose_1_output_0_V_tp_for_/neck/high_IFM/transformer_blocks.0/attn/MatMul_1_exsdpa_rs_2reshape
| 101 exDataConvert INT8 NPU (4,16,1,100) (4,16,1,100) 650/0/650 137 \ 100.0%/0.0%/0.0% 6 exDataConvert:/neck/high_IFM/transformer_blocks.0/attn/Transpose_1_output_0_V_tp_for_/neck/high_IFM/transformer_blocks.0/attn/MatMul_1_ex__1__cvt_int8_float16
| 102 exSDPAttention FLOAT16 NPU (4,8,1,100),(4,8,1,100),(4,16,1,100) (4,16,1,100) 1299/0/1299 334 \ 100.0%/0.0%/0.0% 25 exSDPAttention:/neck/high_IFM/transformer_blocks.0/attn/MatMul_1_exsdpa
| 103 exDataConvert FLOAT16 NPU (4,16,1,100) (4,16,1,100) 650/0/650 134 \ 100.0%/0.0%/0.0% 12 exDataConvert:/neck/high_IFM/transformer_blocks.0/attn/MatMul_1_output_0_for_/neck/high_IFM/transformer_blocks.0/attn/MatMul_1_exsdpa__cvt_float16_int8
| 104 Reshape INT8 NPU (4,16,1,100),(4) (1,64,10,10) 434/0/434 1 \ 0.0%/0.0%/0.0% 6 Reshape:/neck/high_IFM/transformer_blocks.0/attn/Reshape_3
| 105 Clip INT8 NPU (1,64,10,10),(1),(1) (1,64,10,10) 434/0/434 144 \ 38.5%/30.7%/30.7% 6 Clip:/neck/high_IFM/transformer_blocks.0/attn/proj/proj.0/Clip
| 106 ConvAdd INT8 NPU (1,64,10,10),(704,64,1,1),(704),... (1,704,10,10) 6693/9856/9856 158 1.16/1.16/1.16 40.0%/30.0%/30.0% 124 Conv:/neck/high_IFM/transformer_blocks.0/attn/proj/proj.1/c/Conv
| 107 Conv INT8 NPU (1,704,10,10),(704,704,1,1),(704) (1,704,10,10) 21715/54208/54208 140 14.40/14.40/14.40 36.4%/31.8%/31.8% 558 Conv:/neck/high_IFM/transformer_blocks.0/mlp/fc1/c/Conv
| 108 ConvClip INT8 NPU (1,704,10,10),(1,704,3,3),(704) (1,704,10,10) 5167/11088/11088 155 0.17/0.17/0.17 40.0%/30.0%/30.0% 80 Conv:/neck/high_IFM/transformer_blocks.0/mlp/dwconv/Conv
| 109 ConvAdd INT8 NPU (1,704,10,10),(704,704,1,1),(704),... (1,704,10,10) 24096/54208/54208 139 14.51/14.51/14.51 36.4%/31.8%/31.8% 627 Conv:/neck/high_IFM/transformer_blocks.0/mlp/fc2/c/Conv
| 110 Conv INT8 NPU (1,704,10,10),(32,704,1,1),(32) (1,32,10,10) 3260/2464/3260 99 0.93/0.93/0.93 40.0%/30.0%/30.0% 91 Conv:/neck/high_IFM/transformer_blocks.1/attn/to_q/c/Conv
| 111 Conv INT8 NPU (1,704,10,10),(32,704,1,1),(32) (1,32,10,10) 3260/2464/3260 98 0.94/0.94/0.94 40.0%/30.0%/30.0% 91 Conv:/neck/high_IFM/transformer_blocks.1/attn/to_k/c/Conv
| 112 Conv INT8 NPU (1,704,10,10),(64,704,1,1),(64) (1,64,10,10) 4139/4928/4928 101 1.82/1.82/1.82 38.5%/30.7%/30.7% 113 Conv:/neck/high_IFM/transformer_blocks.1/attn/to_v/c/Conv
| 113 Reshape INT8 CPU (1,32,10,10),(4) (4,8,1,100) 0/0/0 42 \ 0.0%/0.0%/0.0% 3 Reshape:/neck/high_IFM/transformer_blocks.1/attn/Transpose_output_0_Q_tp_for_/neck/high_IFM/transformer_blocks.1/attn/MatMul_1_exsdpa_rs_2reshape
| 114 exDataConvert INT8 CPU (4,8,1,100) (4,8,1,100) 0/0/0 25 \ 0.0%/0.0%/0.0% 3 exDataConvert:/neck/high_IFM/transformer_blocks.1/attn/Transpose_output_0_Q_tp_for_/neck/high_IFM/transformer_blocks.1/attn/MatMul_1_exsd__1__cvt_int8_float16
| 115 Reshape INT8 CPU (1,32,10,10),(4) (4,8,1,100) 0/0/0 43 \ 0.0%/0.0%/0.0% 3 Reshape:/neck/high_IFM/transformer_blocks.1/attn/Reshape_1_output_0_K_rs_for_/neck/high_IFM/transformer_blocks.1/attn/MatMul_1_exsdpa
| 116 exDataConvert INT8 CPU (4,8,1,100) (4,8,1,100) 0/0/0 25 \ 0.0%/0.0%/0.0% 3 exDataConvert:/neck/high_IFM/transformer_blocks.1/attn/Reshape_1_output_0_K_rs_for_/neck/high_IFM/transformer_blocks.1/attn/MatMul_1_exsdpa__cvt_int8_float16
| 117 Reshape INT8 NPU (1,64,10,10),(4) (4,16,1,100) 434/0/434 1 \ 0.0%/0.0%/0.0% 6 Reshape:/neck/high_IFM/transformer_blocks.1/attn/Transpose_1_output_0_V_tp_for_/neck/high_IFM/transformer_blocks.1/attn/MatMul_1_exsdpa_rs_2reshape
| 118 exDataConvert INT8 NPU (4,16,1,100) (4,16,1,100) 650/0/650 85 \ 100.0%/0.0%/0.0% 6 exDataConvert:/neck/high_IFM/transformer_blocks.1/attn/Transpose_1_output_0_V_tp_for_/neck/high_IFM/transformer_blocks.1/attn/MatMul_1_ex__1__cvt_int8_float16
| 119 exSDPAttention FLOAT16 NPU (4,8,1,100),(4,8,1,100),(4,16,1,100) (4,16,1,100) 1299/0/1299 279 \ 100.0%/0.0%/0.0% 25 exSDPAttention:/neck/high_IFM/transformer_blocks.1/attn/MatMul_1_exsdpa
| 120 exDataConvert FLOAT16 NPU (4,16,1,100) (4,16,1,100) 650/0/650 80 \ 100.0%/0.0%/0.0% 12 exDataConvert:/neck/high_IFM/transformer_blocks.1/attn/MatMul_1_output_0_for_/neck/high_IFM/transformer_blocks.1/attn/MatMul_1_exsdpa__cvt_float16_int8
| 121 Reshape INT8 NPU (4,16,1,100),(4) (1,64,10,10) 434/0/434 1 \ 0.0%/0.0%/0.0% 6 Reshape:/neck/high_IFM/transformer_blocks.1/attn/Reshape_3
| 122 Clip INT8 NPU (1,64,10,10),(1),(1) (1,64,10,10) 434/0/434 90 \ 38.5%/30.7%/30.7% 6 Clip:/neck/high_IFM/transformer_blocks.1/attn/proj/proj.0/Clip
| 123 ConvAdd INT8 NPU (1,64,10,10),(704,64,1,1),(704),... (1,704,10,10) 6693/9856/9856 102 1.80/1.80/1.80 40.0%/30.0%/30.0% 124 Conv:/neck/high_IFM/transformer_blocks.1/attn/proj/proj.1/c/Conv
| 124 Conv INT8 NPU (1,704,10,10),(704,704,1,1),(704) (1,704,10,10) 21715/54208/54208 126 16.01/16.01/16.01 36.4%/31.8%/31.8% 558 Conv:/neck/high_IFM/transformer_blocks.1/mlp/fc1/c/Conv
| 125 ConvClip INT8 NPU (1,704,10,10),(1,704,3,3),(704) (1,704,10,10) 5167/11088/11088 102 0.25/0.25/0.25 40.0%/30.0%/30.0% 80 Conv:/neck/high_IFM/transformer_blocks.1/mlp/dwconv/Conv
| 126 ConvAdd INT8 NPU (1,704,10,10),(704,704,1,1),(704),... (1,704,10,10) 24096/54208/54208 127 15.88/15.88/15.88 36.4%/31.8%/31.8% 627 Conv:/neck/high_IFM/transformer_blocks.1/mlp/fc2/c/Conv
| 127 Conv INT8 NPU (1,704,10,10),(384,704,1,1),(384) (1,384,10,10) 12927/29568/29568 118 9.32/9.32/9.32 33.3%/33.3%/33.3% 335 Conv:/neck/conv_1x1_n/Conv
| 128 Split INT8 NPU (1,384,10,10),(2) (1,128,10,10),... 2598/0/2598 86 \ 100.0%/0.0%/0.0% 37 Split:/neck/Split_1
| 129 Conv INT8 NPU (1,128,10,10),(128,128,1,1),(128) (1,128,10,10) 1455/1792/1792 97 0.69/0.69/0.69 40.0%/30.0%/30.0% 29 Conv:/neck/Inject_n4/global_act/conv/Conv
| 130 Conv INT8 NPU (1,128,10,10),(128,128,1,1),(128) (1,128,10,10) 1455/1792/1792 98 0.68/0.68/0.68 40.0%/30.0%/30.0% 29 Conv:/neck/Inject_n4/global_embedding/conv/Conv
| 131 Conv INT8 NPU (1,256,10,10),(256,256,1,1),(256) (1,256,10,10) 4018/7168/7168 102 2.61/2.61/2.61 40.0%/30.0%/30.0% 91 Conv:/neck/Inject_n5/global_act/conv/Conv
| 132 Conv INT8 NPU (1,256,10,10),(256,256,1,1),(256) (1,256,10,10) 4018/7168/7168 103 2.59/2.59/2.59 39.6%/30.2%/30.2% 91 Conv:/neck/Inject_n5/global_embedding/conv/Conv
| 133 ConvTranspose INT8 NPU (1,128,10,10),(128,1,8,8),(128) (1,128,40,40) 7672/204800/204800 331 1.61/1.61/1.61 100.0%/0.0%/0.0% 21 ConvTranspose:/neck/Inject_n4/Resize_1_2deconv
| 134 ConvTranspose INT8 NPU (1,256,10,10),(256,1,4,4),(256) (1,256,20,20) 4537/25600/25600 112 0.60/0.60/0.60 100.0%/0.0%/0.0% 31 ConvTranspose:/neck/Inject_n5/Resize_1_2deconv
| 135 Add INT8 NPU (1,128,10,10),(1) (1,128,10,10) 866/0/866 91 \ 39.2%/30.4%/30.4% 12 Add:/neck/Inject_n4/act/Add
| 136 Clip INT8 NPU (1,128,10,10),(1),(1) (1,128,10,10) 867/0/867 92 \ 39.2%/30.4%/30.4% 12 Clip:/neck/Inject_n4/act/relu/Clip
| 137 Mul INT8 NPU (1,128,10,10),(1) (1,128,10,10) 866/0/866 80 \ 100.0%/0.0%/0.0% 12 Mul:/neck/Inject_n4/act/Div_2mul
| 138 ConvTranspose INT8 NPU (1,128,10,10),(128,1,8,8),(128) (1,128,40,40) 7672/204800/204800 330 1.62/1.62/1.62 100.0%/0.0%/0.0% 21 ConvTranspose:/neck/Inject_n4/Resize_2deconv
| 139 Mul INT8 NPU (1,128,40,40),(1,128,40,40) (1,128,40,40) 20780/0/20780 121 \ 100.0%/0.0%/0.0% 400 Mul:/neck/Inject_n4/Mul
| 140 Add INT8 NPU (1,128,40,40),(1,128,40,40) (1,128,40,40) 20780/0/20780 110 \ 35.0%/32.5%/32.5% 400 Add:/neck/Inject_n4/Add
| 141 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 199 48.24/48.24/48.24 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_n4/conv1/rbr_reparam/Conv
| 142 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 198 48.48/48.48/48.48 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_n4/block/block.0/rbr_reparam/Conv
| 143 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 251 38.25/38.25/38.25 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_n4/block/block.1/rbr_reparam/Conv
| 144 ConvRelu INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 261 36.78/36.78/36.78 35.0%/32.5%/32.5% 345 Conv:/neck/Rep_n4/block/block.2/rbr_reparam/Conv
| 145 AveragePool INT8 NPU (1,128,40,40) (1,128,20,20) 8659/0/8659 194 \ 100.0%/0.0%/0.0% 200 AveragePool:/neck/LAF_n5/AveragePool
| 146 Concat INT8 NPU (1,128,20,20),(1,128,20,20) (1,256,20,20) 6927/0/6927 154 \ 35.0%/35.0%/30.1% 100 Concat:/neck/LAF_n5/Concat
| 147 Conv INT8 NPU (1,256,20,20),(256,256,1,1),(256) (1,256,20,20) 9213/25600/25600 167 6.39/6.39/6.39 35.0%/35.0%/30.1% 166 Conv:/neck/Inject_n5/local_embedding/conv/Conv
| 148 ConvExSwish INT8 NPU (1,128,40,40),(128,128,1,1),(128) (1,128,40,40) 14442/25600/25600 271 3.94/3.94/3.94 100.0%/0.0%/0.0% 217 Conv:/detect/stems.1/conv/Conv
| 149 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 425 22.59/22.59/22.59 100.0%/0.0%/0.0% 345 Conv:/detect/cls_convs.1/conv/Conv
| 150 Conv INT8 NPU (1,128,40,40),(80,128,1,1),(80) (1,80,40,40) 11628/19200/19200 167 3.99/3.99/3.99 35.0%/32.5%/32.5% 210 Conv:/detect/cls_preds.1/Conv
| 151 ConvExSwish INT8 NPU (1,128,40,40),(128,128,3,3),(128) (1,128,40,40) 18875/230400/230400 426 22.54/22.54/22.54 100.0%/0.0%/0.0% 345 Conv:/detect/reg_convs.1/conv/Conv
| 152 Conv INT8 NPU (1,128,40,40),(4,128,1,1),(4) (1,4,40,40) 7819/6400/7819 159 0.21/0.21/0.21 34.9%/32.5%/32.5% 200 Conv:/detect/reg_preds_lrtb.1/Conv
| 153 Concat INT8 CPU (1,4,40,40),(1,80,40,40) (1,84,40,40) 0/0/0 250 \ 0.0%/0.0%/0.0% 150 Concat:/detect/Concat_1
| 154 OutputOperator INT8 CPU (1,84,40,40) \ 0/0/0 9 \ 0.0%/0.0%/0.0% 131 OutputOperator:output1
| 155 Add INT8 NPU (1,256,10,10),(1) (1,256,10,10) 1732/0/1732 147 \ 39.6%/30.2%/30.2% 25 Add:/neck/Inject_n5/act/Add
| 156 Clip INT8 NPU (1,256,10,10),(1),(1) (1,256,10,10) 1732/0/1732 145 \ 39.6%/30.2%/30.2% 25 Clip:/neck/Inject_n5/act/relu/Clip
| 157 Mul INT8 NPU (1,256,10,10),(1) (1,256,10,10) 1732/0/1732 137 \ 100.0%/0.0%/0.0% 25 Mul:/neck/Inject_n5/act/Div_2mul
| 158 ConvTranspose INT8 NPU (1,256,10,10),(256,1,4,4),(256) (1,256,20,20) 4537/25600/25600 166 0.40/0.40/0.40 100.0%/0.0%/0.0% 31 ConvTranspose:/neck/Inject_n5/Resize_2deconv
| 159 Mul INT8 NPU (1,256,20,20),(1,256,20,20) (1,256,20,20) 10390/0/10390 109 \ 100.0%/0.0%/0.0% 200 Mul:/neck/Inject_n5/Mul
| 160 Add INT8 NPU (1,256,20,20),(1,256,20,20) (1,256,20,20) 10390/0/10390 100 \ 35.0%/35.0%/30.1% 200 Add:/neck/Inject_n5/Add
| 161 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 179 53.63/53.63/53.63 37.5%/37.5%/25.0% 678 Conv:/neck/Rep_n5/conv1/rbr_reparam/Conv
| 162 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 180 53.33/53.33/53.33 37.5%/37.5%/25.0% 678 Conv:/neck/Rep_n5/block/block.0/rbr_reparam/Conv
| 163 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 178 53.93/53.93/53.93 37.5%/37.5%/25.0% 678 Conv:/neck/Rep_n5/block/block.1/rbr_reparam/Conv
| 164 ConvRelu INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 269 35.69/35.69/35.69 37.4%/37.4%/25.1% 678 Conv:/neck/Rep_n5/block/block.2/rbr_reparam/Conv
| 165 ConvExSwish INT8 NPU (1,256,20,20),(256,256,1,1),(256) (1,256,20,20) 9213/25600/25600 206 5.18/5.18/5.18 100.0%/0.0%/0.0% 166 Conv:/detect/stems.2/conv/Conv
| 166 ConvExSwish INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 427 22.48/22.48/22.48 100.0%/0.0%/0.0% 678 Conv:/detect/cls_convs.2/conv/Conv
| 167 Conv INT8 NPU (1,256,20,20),(80,256,1,1),(80) (1,80,20,20) 5265/9600/9600 159 2.10/2.10/2.10 34.9%/34.9%/30.2% 120 Conv:/detect/cls_preds.2/Conv
| 168 ConvExSwish INT8 NPU (1,256,20,20),(256,256,3,3),(256) (1,256,20,20) 26945/230400/230400 428 22.43/22.43/22.43 100.0%/0.0%/0.0% 678 Conv:/detect/reg_convs.2/conv/Conv
| 169 Conv INT8 NPU (1,256,20,20),(4,256,1,1),(4) (1,4,20,20) 3724/3200/3724 157 0.11/0.11/0.11 34.8%/34.8%/30.4% 101 Conv:/detect/reg_preds_lrtb.2/Conv
| 170 Concat INT8 CPU (1,4,20,20),(1,80,20,20) (1,84,20,20) 0/0/0 70 \ 0.0%/0.0%/0.0% 37 Concat:/detect/Concat_2
| 171 OutputOperator INT8 CPU (1,84,20,20) \ 0/0/0 4 \ 0.0%/0.0%/0.0% 32 OutputOperator:output2
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| Total Operator Elapsed Per Frame Time(us): 37673
| Total Memory Read/Write Per Frame Size(KB): 71416.8
| ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| ---------------------------------------------------------------------------------------------------
| Operator Time Consuming Ranking Table
| ---------------------------------------------------------------------------------------------------
| OpType CallNumber CPUTime(us) GPUTime(us) NPUTime(us) TotalTime(us) TimeRatio(%)
| ---------------------------------------------------------------------------------------------------
| ConvRelu 51 0 0 17056 17056 45.27%
| ConvExSwish 11 0 0 4205 4205 11.16%
| Conv 26 0 0 3432 3432 9.11%
| AveragePool 8 0 0 2697 2697 7.16%
| Concat 11 1279 0 1315 2594 6.89%
| ConvTranspose 9 0 0 2128 2128 5.65%
| Mul 8 0 0 1050 1050 2.79%
| Add 7 0 0 860 860 2.28%
| Clip 6 0 0 727 727 1.93%
| ConvAdd 5 0 0 643 643 1.71%
| exSDPAttention 2 0 0 613 613 1.63%
| exDataConvert 8 98 0 436 534 1.42%
| MaxPool 3 0 0 392 392 1.04%
| Split 2 0 0 269 269 0.71%
| ConvClip 2 0 0 257 257 0.68%
| Reshape 8 168 0 4 172 0.46%
| OutputOperator 3 40 0 0 40 0.11%
| InputOperator 1 4 0 0 4 0.01%
| ---------------------------------------------------------------------------------------------------