update variable naming and comments for rocm (#5299)

This commit is contained in:
Zhaoyi Li
2025-04-12 01:15:05 -05:00
committed by GitHub
parent 2eb55770f9
commit 3c9740d200
6 changed files with 14 additions and 10 deletions

View File

@@ -11,8 +11,8 @@ from vllm import _custom_ops as ops
from sglang.srt.utils import is_hip
is_hip_ = is_hip()
fp8_type_ = torch.float8_e4m3fnuz if is_hip_ else torch.float8_e4m3fn
_is_hip = is_hip()
fp8_type_ = torch.float8_e4m3fnuz if _is_hip else torch.float8_e4m3fn
def vllm_scaled_fp8_quant(

View File

@@ -8,8 +8,8 @@ from sgl_kernel import sgl_per_token_group_quant_fp8, sgl_per_token_group_quant_
from sglang.srt.utils import is_hip
is_hip_ = is_hip()
fp8_type_ = torch.float8_e4m3fnuz if is_hip_ else torch.float8_e4m3fn
_is_hip = is_hip()
fp8_type_ = torch.float8_e4m3fnuz if _is_hip else torch.float8_e4m3fn
@triton.jit

View File

@@ -9,8 +9,8 @@ from vllm import _custom_ops as ops
from sglang.srt.utils import is_hip
is_hip_ = is_hip()
fp8_type_ = torch.float8_e4m3fnuz if is_hip_ else torch.float8_e4m3fn
_is_hip = is_hip()
fp8_type_ = torch.float8_e4m3fnuz if _is_hip else torch.float8_e4m3fn
def vllm_per_token_quant_fp8(