Remove triton per group quant [ROCm] [Bugfix] (#49621)
Signed-off-by: Andy Friedrich <afriedri@amd.com>
This commit is contained in:
@@ -158,13 +158,6 @@ class TestSiluMulGroupFp8QuantModel(torch.nn.Module):
|
||||
input_dtype=dtype,
|
||||
)
|
||||
|
||||
if not current_platform.is_fp8_fnuz():
|
||||
kernel = self.w8a8_block_fp8_linear.kernel
|
||||
orig_quant = kernel.quant_fp8
|
||||
kernel.quant_fp8 = lambda *a, use_triton=False, **kw: orig_quant(
|
||||
*a, use_triton=True, **kw
|
||||
)
|
||||
|
||||
self.enable_silu_mul_custom_op = self.silu_and_mul.enabled()
|
||||
|
||||
def forward(self, x):
|
||||
@@ -175,9 +168,7 @@ class TestSiluMulGroupFp8QuantModel(torch.nn.Module):
|
||||
def ops_in_model_before(self):
|
||||
return [
|
||||
SILU_MUL_OP if self.enable_silu_mul_custom_op else torch.ops.aten.mul,
|
||||
rocm_aiter_ops.get_group_quant_op()
|
||||
if current_platform.is_fp8_fnuz()
|
||||
else torch.ops.vllm.triton_per_token_group_quant_fp8.default,
|
||||
rocm_aiter_ops.get_group_quant_op(),
|
||||
]
|
||||
|
||||
def ops_in_model_after(self):
|
||||
|
||||
Reference in New Issue
Block a user