[cpu][perf] Accelerate unquantized-linear for AArch64 through oneDNN/ACL and weight prepack (#25948)

Signed-off-by: Fadi Arafeh <fadi.arafeh@arm.com>
Co-authored-by: Li, Jiang <jiang1.li@intel.com>
This commit is contained in:
Fadi Arafeh
2025-10-04 12:16:38 +08:00
committed by GitHub
co-authored by Li, Jiang <jiang1.li@intel.com>
parent 2f7dbc9b42
commit 9705fba7b7
8 changed files with 111 additions and 16 deletions
+3 -2
View File
@@ -165,8 +165,9 @@ def dispatch_cpu_unquantized_gemm(
if remove_weight:
layer.weight = torch.nn.Parameter(torch.empty(0),
requires_grad=False)
elif (ops._supports_onednn
and current_platform.get_cpu_architecture() == CpuArchEnum.X86):
elif ops._supports_onednn and (current_platform.get_cpu_architecture()
== CpuArchEnum.X86
or ops.is_onednn_acl_supported()):
origin_weight = layer.weight
if remove_weight:
layer.weight = torch.nn.Parameter(torch.empty(0),