[Perf] Optimize Fp8BlockScaledMMLinearKernel input_scale tensor using new_empty() (#43677)
Signed-off-by: Xin Yang <xyangx@amazon.com>
This commit is contained in:
@@ -126,7 +126,7 @@ class Fp8BlockScaledMMLinearKernel(
|
||||
# always Tensors. Subclasses with apply_input_quant=False must not
|
||||
# use As in apply_block_scaled_mm.
|
||||
input_scale = (
|
||||
input_scale if input_scale is not None else input_2d.new_ones(1)
|
||||
input_scale if input_scale is not None else input_2d.new_empty(1)
|
||||
)
|
||||
|
||||
output = self.apply_block_scaled_mm(
|
||||
|
||||
Reference in New Issue
Block a user