Skip to content

[INTEL_HPU] Moe tensor align #2321

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
LeoZhao-Intel merged 2 commits into from
Jan 15, 2026
+409 −164
Merged

[INTEL_HPU] Moe tensor align #2321

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
09fbe8e
list to tensor 0x80 alignment padding
Dec 2, 2025
8667c8d
update gate_moe UT
yanfeich Jan 4, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
33 changes: 15 additions & 18 deletions backends/intel_hpu/custom_ops/llama_infer/fused_gate_moe.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -416,8 +416,7 @@ void FusedGateMoeKernel(
const phi::DenseTensor& gate_up_weights,
const phi::DenseTensor& down_weights,
const paddle::optional<phi::DenseTensor>& hidden_states_scales,
const paddle::optional<std::vector<phi::DenseTensor>>&
intermediate_hidden_states_scales,
const paddle::optional<phi::DenseTensor>& intermediate_hidden_states_scales,
const paddle::optional<phi::DenseTensor>& gate_up_weights_scales,
const paddle::optional<phi::DenseTensor>& down_weights_scales,
phi::DenseTensor* final_hidden_states,
Expand Down Expand Up @@ -468,9 +467,7 @@ void FusedGateMoeKernel(
ct.AddN(down_weights);

if (intermediate_hidden_states_scales) {
for (const auto& t : intermediate_hidden_states_scales.get()) {
ct.Add(t);
}
ct.AddN(intermediate_hidden_states_scales.get());
}
if (gate_up_weights_scales) {
ct.AddN(gate_up_weights_scales.get());
Expand Down Expand Up @@ -514,8 +511,7 @@ void CallFusedGateMoeKernel(
const phi::DenseTensor& gate_up_weights,
const phi::DenseTensor& down_weights,
const paddle::optional<phi::DenseTensor>& hidden_states_scales,
const paddle::optional<std::vector<phi::DenseTensor>>&
intermediate_hidden_states_scales,
const paddle::optional<phi::DenseTensor>& intermediate_hidden_states_scales,
const paddle::optional<phi::DenseTensor>& gate_up_weights_scales,
const paddle::optional<phi::DenseTensor>& down_weights_scales,
phi::DenseTensor* final_hidden_states,
Expand Down Expand Up @@ -634,7 +630,7 @@ std::vector<paddle::Tensor> FusedGateMoeForward(
*gate_up_weights_tensor,
*down_weights_tensor,
paddle::optional<phi::DenseTensor>(), /* hidden_states_scale */
paddle::optional<std::vector<phi::DenseTensor>>(), /* intermediate */
paddle::optional<phi::DenseTensor>(), /* intermediate */
paddle::optional<phi::DenseTensor>(), /* gate_up_weights_scales */
paddle::optional<phi::DenseTensor>(), /* down_weights_scales */
final_hidden_states.get(),
Expand All @@ -660,8 +656,7 @@ std::vector<paddle::Tensor> FusedGateMoeFP8Forward(
const paddle::Tensor& gate_up_weights,
const paddle::Tensor& down_weights,
const paddle::optional<paddle::Tensor>& hidden_states_scales,
const paddle::optional<std::vector<paddle::Tensor>>&
intermediate_hidden_states_scales,
const paddle::optional<paddle::Tensor>& intermediate_hidden_states_scales,
const paddle::Tensor& gate_up_weights_scales,
const paddle::Tensor& down_weights_scales,
const int top_k,
Expand Down Expand Up @@ -701,14 +696,16 @@ std::vector<paddle::Tensor> FusedGateMoeFP8Forward(
paddle::optional<phi::DenseTensor>(*hidden_states_scales_dt);
}

auto intermediate_hidden_states_scales_tensor =
paddle::optional<phi::DenseTensor>();
bool dynamic_scale = true;
std::vector<phi::DenseTensor> scales_vec;
if (intermediate_hidden_states_scales) {
dynamic_scale = false;
for (const auto& t : intermediate_hidden_states_scales.get()) {
scales_vec.push_back(
*static_cast<const phi::DenseTensor*>(t.impl().get()));
}
auto intermediate_hidden_states_scales_dt = static_cast<phi::DenseTensor*>(
intermediate_hidden_states_scales->impl().get());
intermediate_hidden_states_scales_tensor =
paddle::optional<phi::DenseTensor>(
*intermediate_hidden_states_scales_dt);
}
auto gate_up_weights_scales_tensor = paddle::optional<phi::DenseTensor>();
auto gate_up_weights_scales_dt =
Expand All @@ -735,7 +732,7 @@ std::vector<paddle::Tensor> FusedGateMoeFP8Forward(
*gate_up_weights_tensor,
*down_weights_tensor,
hidden_states_scales_tensor,
scales_vec,
intermediate_hidden_states_scales_tensor,
gate_up_weights_scales_tensor,
down_weights_scales_tensor,
final_hidden_states.get(),
Expand Down Expand Up @@ -816,7 +813,7 @@ std::vector<paddle::Tensor> FusedGateMoeBlockWiseFP8Forward(
*gate_up_weights_tensor,
*down_weights_tensor,
paddle::optional<phi::DenseTensor>(), /* hidden_states_scale */
paddle::optional<std::vector<phi::DenseTensor>>(), /* intermediate */
paddle::optional<phi::DenseTensor>(), /* intermediate */
gate_up_weights_scales_tensor,
down_weights_scales_tensor,
final_hidden_states.get(),
Expand Down Expand Up @@ -887,7 +884,7 @@ PD_BUILD_OP(fused_gate_moe_fp8)
"gate_up_weights",
"down_weights",
paddle::Optional("hidden_states_scales"),
paddle::Optional(paddle::Vec("intermediate_hidden_states_scales")),
paddle::Optional("intermediate_hidden_states_scales"),
"gate_up_weights_scales",
"down_weights_scales"})
.Outputs({"final_hidden_states"})
Expand Down
6 changes: 6 additions & 0 deletions backends/intel_hpu/kernels/funcs.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -352,6 +352,12 @@ class ConvertTensors {
if (addr_offset % 0x80 != 0) {
PADDLE_THROW("Tensor list offset is not algined.");
}
if (dims.size() == 2 && dims[0] == 1) {
// [list_size][1] is padded for 0x80 alignment as
// [list_size][1][padded_num].
// now addr_offset = 0x80;
dims.pop_back();
}

if (is_input) {
for (int64_t tensor_idx = 0; tensor_idx < num_list; tensor_idx++) {
Expand Down
1 change: 1 addition & 0 deletions backends/intel_hpu/tests/unittests/test_fused_fp8_qkv_rope.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,7 @@
import os

intel_hpus_module_id = os.environ.get("FLAGS_selected_intel_hpus", 4)
paddle.device.set_device(f"intel_hpu:{intel_hpus_module_id}")

paddle.seed(2025)

Expand Down
Loading
Loading