[Refactor] Remove dead quantization code and tests (#45454)

Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
Wentao Ye
2026-06-17 16:12:01 -04:00
committed by GitHub
parent 091386a99b
commit 2659f60a1a
5 changed files with 0 additions and 120 deletions
-20
View File
@@ -27,23 +27,3 @@ def test_awq_dequantize_opcheck(monkeypatch: pytest.MonkeyPatch):
torch.ops._C.awq_dequantize,
(qweight, scales, zeros, split_k_iters, thx, thy),
)
@pytest.mark.skip(reason="Not working; needs investigation.")
@pytest.mark.skipif(
not hasattr(torch.ops._C, "awq_gemm"),
reason="AWQ is not supported on this GPU type.",
)
def test_awq_gemm_opcheck(monkeypatch: pytest.MonkeyPatch):
with monkeypatch.context() as m:
m.setenv("VLLM_USE_TRITON_AWQ", "0")
input = torch.rand((2, 8192), device="cuda", dtype=torch.float16)
qweight = torch.randint(
-2000000000, 2000000000, (8192, 256), device="cuda", dtype=torch.int32
)
scales = torch.empty((64, 2048), device="cuda", dtype=torch.float16)
qzeros = torch.randint(
-2000000000, 2000000000, (64, 256), device="cuda", dtype=torch.int32
)
split_k_iters = 8
opcheck(torch.ops._C.awq_gemm, (input, qweight, scales, qzeros, split_k_iters))