[Feature][ROCm] Add full graph capture support for TritonAttentionBackend (#19158)

Signed-off-by: charlifu <charlifu@amd.com>
This commit is contained in:
Charlie Fu
2025-06-17 17:03:06 -04:00
committed by GitHub
parent b447624ee3
commit a44b1c951d
5 changed files with 334 additions and 178 deletions
@@ -147,6 +147,7 @@ def test_lower_max_num_seqs(model, supported):
llm.generate(["Hello, my name is"] * 10)
@pytest.mark.skipif(not current_platform.is_cuda(), reason="Skip if not cuda")
def test_full_cudagraph_with_invalid_backend():
with temporary_environ({
"VLLM_USE_V1": "1",