biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Misc] Replace os environ to monkeypatch in test suite (#14516)

Browse Source 
Signed-off-by: sibi <85477603+t-sibiraj@users.noreply.github.com>
Signed-off-by: Aaron Pham <contact@aarnphm.xyz>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
Co-authored-by: Aaron Pham <contact@aarnphm.xyz>
This commit is contained in:
Sibi
2025-03-17 11:35:57 +08:00
committed by GitHub
parent 1e799b7ec1
commit a73e183e36
43 changed files with 1900 additions and 1658 deletions
Download Patch File Download Diff File Expand all files Collapse all files

60
tests/kernels/test_awq.py
View File

@@ -1,7 +1,5 @@
# SPDX-License-Identifier: Apache-2.0
import os
import pytest
import torch
@@ -11,36 +9,38 @@ from vllm import _custom_ops as ops # noqa: F401
@pytest.mark.skipif(not hasattr(torch.ops._C, "awq_dequantize"),
reason="AWQ is not supported on this GPU type.")
def test_awq_dequantize_opcheck():
os.environ["VLLM_USE_TRITON_AWQ"] = "0"
qweight = torch.randint(-2000000000,
2000000000, (8192, 256),
device='cuda',
dtype=torch.int32)
scales = torch.rand((64, 2048), device='cuda', dtype=torch.float16)
zeros = torch.empty((64, 256), device='cuda', dtype=torch.int32)
split_k_iters = 0
thx = 0
thy = 0
opcheck(torch.ops._C.awq_dequantize,
(qweight, scales, zeros, split_k_iters, thx, thy))
def test_awq_dequantize_opcheck(monkeypatch: pytest.MonkeyPatch):
with monkeypatch.context() as m:
m.setenv("VLLM_USE_TRITON_AWQ", "0")
qweight = torch.randint(-2000000000,
2000000000, (8192, 256),
device='cuda',
dtype=torch.int32)
scales = torch.rand((64, 2048), device='cuda', dtype=torch.float16)
zeros = torch.empty((64, 256), device='cuda', dtype=torch.int32)
split_k_iters = 0
thx = 0
thy = 0
opcheck(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():
os.environ["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.randint(-2000000000,
2000000000, (64, 256),
device='cuda',
dtype=torch.int32)
qzeros = torch.empty((64, 2048), device='cuda', dtype=torch.float16)
split_k_iters = 8
opcheck(torch.ops._C.awq_gemm,
(input, qweight, qzeros, scales, split_k_iters))
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.randint(-2000000000,
2000000000, (64, 256),
device='cuda',
dtype=torch.int32)
qzeros = torch.empty((64, 2048), device='cuda', dtype=torch.float16)
split_k_iters = 8
opcheck(torch.ops._C.awq_gemm,
(input, qweight, qzeros, scales, split_k_iters))