EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

[0.11.0]Chery pick pta upgrade change (#3940)

Browse Source 
This PR cherry-pick two commit from main to upgrade torch-npu to 2.7.1
official release

---------

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
This commit is contained in:
wangxiyuan
2025-10-31 22:14:26 +08:00
committed by GitHub
parent 3d81ea03ed
commit 8a7154001e
16 changed files with 93 additions and 171 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
tests/ut/torchair/quantization/test_torchair_w8a8_dynamic.py
View File

@@ -23,9 +23,9 @@ class TestAscendW8A8FusedMoEMethod(TestBase):
@patch("torch_npu.npu_swiglu")
@patch("torch_npu.npu_dynamic_quant")
@patch("torch_npu.npu_moe_finalize_routing")
@patch("torch_npu.npu_moe_init_routing")
@patch("torch_npu.npu_moe_init_routing_quant")
def test_torchair_fused_experts_with_all2all(
self, mock_moe_init_routing, mock_moe_finalize_routing,
self, mock_npu_moe_init_routing_quant, mock_moe_finalize_routing,
mock_dynamic_quant, mock_swiglu, mock_grouped_matmul,
mock_moe_re_routing, mock_all_to_all_single):
@@ -38,11 +38,10 @@ class TestAscendW8A8FusedMoEMethod(TestBase):
placeholder_ones = torch.ones(self.num_tokens, dtype=torch.int32)
mock_all_to_all_single.side_effect = lambda output, input, *args, **kwargs: output.copy_(
input)
mock_moe_init_routing.return_value = (
placeholder_int8,
placeholder_ones,
placeholder_ones,
)
mock_npu_moe_init_routing_quant.return_value = (
placeholder_int8, placeholder_ones, placeholder_ones,
torch.bincount(placeholder_ones, minlength=len(expert_map)),
torch.randn(self.num_tokens))
mock_moe_re_routing.return_value = (placeholder_int8, self.placeholder,
torch.randint(0,
100,