Skip to content

Commit 26524d0

Browse files
gabe-l-hartgabe-l-hart
committed
fix: Only enable ssm_scan_f32_group if simd_sum available (MTLGPUFamilyApple7)
Branch: GraniteFourPerf Signed-off-by: Gabe Goodhart <ghart@us.ibm.com>
1 parent 8d5a25d commit 26524d0

File tree

1 file changed

+1
-1
lines changed

1 file changed

+1
-1
lines changed

‎ggml/src/ggml-metal/ggml-metal.m

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1211,7 +1211,7 @@ @implementation GGMLMetalClass
12111211
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_NORM, norm, true);
12121212
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_SSM_CONV_F32, ssm_conv_f32, true);
12131213
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_SSM_SCAN_F32, ssm_scan_f32, true);
1214-
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_SSM_SCAN_F32_GROUP, ssm_scan_f32_group, true);
1214+
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_SSM_SCAN_F32_GROUP, ssm_scan_f32_group, has_simdgroup_mm);
12151215
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_RWKV_WKV6_F32, rwkv_wkv6_f32, true);
12161216
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_RWKV_WKV7_F32, rwkv_wkv7_f32, true);
12171217
GGML_METAL_ADD_KERNEL(GGML_METAL_KERNEL_TYPE_MUL_MV_F32_F32, mul_mv_f32_f32, has_simdgroup_reduction);

0 commit comments

Comments
 (0)