Skip to content

Commit 47154cf

Browse files
authored
Fixed GPU stats merging (#851)
* Added handling for n_gpu_samples and gpu_mem_samples. * Include files with any GPU samples.
1 parent bdf9218 commit 47154cf

File tree

2 files changed

+7
-0
lines changed

2 files changed

+7
-0
lines changed

scalene/scalene_json.py

+2
Original file line numberDiff line numberDiff line change
@@ -305,6 +305,7 @@ def output_profiles(
305305
not stats.total_cpu_samples
306306
and not stats.total_memory_malloc_samples
307307
and not stats.total_memory_free_samples
308+
and not stats.total_gpu_samples
308309
):
309310
# Nothing to output.
310311
return {}
@@ -315,6 +316,7 @@ def output_profiles(
315316
+ list(stats.cpu_samples_c.keys())
316317
+ list(stats.memory_free_samples.keys())
317318
+ list(stats.memory_malloc_samples.keys())
319+
+ list(stats.gpu_samples.keys())
318320
)
319321
)
320322
if not all_instrumented_files:

scalene/scalene_statistics.py

+5
Original file line numberDiff line numberDiff line change
@@ -362,6 +362,8 @@ def build_function_stats(self, filename: Filename) -> ScaleneStatistics:
362362
"function_map",
363363
"firstline_map",
364364
"gpu_samples",
365+
"n_gpu_samples",
366+
"gpu_mem_samples",
365367
"total_gpu_samples",
366368
"memory_malloc_count",
367369
"memory_free_count",
@@ -458,6 +460,9 @@ def merge_stats(self, the_dir_name: pathlib.Path) -> None:
458460
self.increment_per_line_samples(
459461
self.n_gpu_samples, x.n_gpu_samples
460462
)
463+
self.increment_per_line_samples(
464+
self.gpu_mem_samples, x.gpu_mem_samples
465+
)
461466
self.increment_per_line_samples(
462467
self.memcpy_samples, x.memcpy_samples
463468
)

0 commit comments

Comments
 (0)