[Core][Kernel] Share persistent workspaces for Marlin and Humming (#57421)
Signed-off-by: mgoin <mgoin64@gmail.com> Co-authored-by: ErinYin <liang.yin@daocloud.io>
This commit is contained in:
@@ -50,7 +50,7 @@ class MarlinNvFp4LinearKernel(NvFp4LinearKernel):
|
||||
weight=layer.weight,
|
||||
weight_scale=layer.weight_scale,
|
||||
weight_global_scale=layer.weight_global_scale,
|
||||
workspace=layer.workspace,
|
||||
workspace=None,
|
||||
size_n=layer.output_size_per_partition,
|
||||
size_k=layer.input_size_per_partition,
|
||||
bias=bias,
|
||||
|
||||
Reference in New Issue
Block a user