Skip to content

[https://nvbugs/6212252][fix] Select CUTLASS MoE backend on non-Blackwell SMs in TestQwen3_5_35B_A3B::test_fp8 #15081

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
xxi-nv merged 2 commits into from
Jun 9, 2026
+8 −2
Merged

[https://nvbugs/6212252][fix] Select CUTLASS MoE backend on non-Blackwell SMs in TestQwen3_5_35B_A3B::test_fp8 #15081

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
ba0e59c
[https://nvbugs/6212252][fix] Select CUTLASS MoE backend on non-Black…
xxi-nv Jun 8, 2026
8e82f0a
Merge branch 'main' into fix/bug-6212252-qwen35-deepgemm-sm90
xxi-nv Jun 8, 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
9 changes: 8 additions & 1 deletion tests/integration/defs/accuracy/test_llm_api_pytorch.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6253,7 +6253,14 @@ def test_fp8(self, enable_block_reuse, mocker):

kv_cache_config = KvCacheConfig(free_gpu_memory_fraction=0.8,
enable_block_reuse=enable_block_reuse)
moe_config = MoeConfig(backend='DEEPGEMM')
# DeepGEMM MoE kernels only support datacenter Blackwell (SM100/SM103).
# Fall back to the CUTLASS MoE backend (which supports FP8 block scales)
# on other architectures such as Hopper (SM90) and consumer Blackwell
# (SM120/SM121); otherwise the unsupported kernel trips a scale-factor
# dtype assertion at warmup.
moe_backend = "DEEPGEMM" if get_sm_version() in (100,
103) else "CUTLASS"
moe_config = MoeConfig(backend=moe_backend)
cuda_graph_config = CudaGraphConfig(enable_padding=True,
max_batch_size=128)
with LLM(model_dir,
Expand Down
1 change: 0 additions & 1 deletion tests/integration/test_lists/waives.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -130,7 +130,6 @@ accuracy/test_llm_api_pytorch.py::TestQwen3_235B_A22B::test_fp8[throughput_laten
accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B_Instruct_2507::test_skip_softmax_attention_4gpus[target_sparsity_0.5-fp8kv=False] SKIP (https://nvbugs/6260915)
accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B_Instruct_2507::test_skip_softmax_attention_4gpus[target_sparsity_0.5-fp8kv=True] SKIP (https://nvbugs/6248783)
accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B_Instruct_2507::test_skip_softmax_attention_4gpus[target_sparsity_0.9-fp8kv=False] SKIP (https://nvbugs/6260915)
accuracy/test_llm_api_pytorch.py::TestQwen3_5_35B_A3B::test_fp8[enable_block_reuse=False] SKIP (https://nvbugs/6212252)
accuracy/test_llm_api_pytorch.py::TestQwen3_5_397B_A17B::test_nvfp4[tep4_cutedsl] SKIP (https://nvbugs/6255417)
accuracy/test_llm_api_pytorch.py::TestQwen3_5_9B::test_bf16[mtp_off] SKIP (https://nvbugs/6212250)
accuracy/test_llm_api_pytorch.py::TestQwen3_5_9B::test_bf16[mtp_on] SKIP (https://nvbugs/6212250)
Expand Down
Loading