kanshan/TensorRT-LLMs
1
0
Fork 0
mirror of https://github.com/NVIDIA/TensorRT-LLM.git synced 2026-02-04 02:02:01 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

[None][fix] AutoDeploy: skip mxfp4_moe test unless on Hopper (#10729)

Browse Source 
Signed-off-by: Fridah-nv <201670829+Fridah-nv@users.noreply.github.com>
This commit is contained in:
Frida Hou 2026-01-16 13:24:37 -08:00 committed by GitHub
parent 0b748d5bba
commit 069ad68d3c
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 5 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
tests/unittest/_torch/auto_deploy/unit/multigpu/custom_ops/test_mxfp4_moe_ep.py
View File

@ -4,6 +4,7 @@ import pytest
import torch
import torch.distributed as dist
from _dist_test_utils import get_device_counts
from utils.util import getSMVersion
from tensorrt_llm._torch.auto_deploy.custom_ops.fused_moe.mxfp4_moe import (
IS_TRITON_KERNELS_AVAILABLE,
@ -109,6 +110,10 @@ def _run_mxfp4_mlp_ep_dtype_test(num_experts: int, topk: int, rank: int, world_s
torch.testing.assert_close(part_out, ref_out, rtol=5e-2, atol=5e-2, equal_nan=True)
@pytest.mark.skipif(
getSMVersion() != 90,
reason="triton_mxfp4_moe is only supported in Hopper architecture",
)
@pytest.mark.skipif(
not IS_TRITON_KERNELS_AVAILABLE,
reason="triton_kernels unavailable",