mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-11-18 11:46:58 +00:00
kleidiai : fix work size and threads sync for fp16 (#16246)
This commit is contained in:
@@ -513,9 +513,9 @@ function(ggml_add_cpu_backend_variant_impl tag_name)
|
||||
|
||||
# Fetch KleidiAI sources:
|
||||
include(FetchContent)
|
||||
set(KLEIDIAI_COMMIT_TAG "v1.13.0")
|
||||
set(KLEIDIAI_COMMIT_TAG "v1.14.0")
|
||||
set(KLEIDIAI_DOWNLOAD_URL "https://github.com/ARM-software/kleidiai/archive/refs/tags/${KLEIDIAI_COMMIT_TAG}.tar.gz")
|
||||
set(KLEIDIAI_ARCHIVE_MD5 "d82a8de939d9814621a5ba23907bdac1")
|
||||
set(KLEIDIAI_ARCHIVE_MD5 "45e110675d93f99f82c23a1afcca76bc")
|
||||
|
||||
if (POLICY CMP0135)
|
||||
cmake_policy(SET CMP0135 NEW)
|
||||
@@ -592,6 +592,7 @@ function(ggml_add_cpu_backend_variant_impl tag_name)
|
||||
${KLEIDIAI_SRC}/kai/ukernels/matmul/matmul_clamp_f32_qsi8d32p_qsi4c32p/kai_matmul_clamp_f32_qsi8d32p1vlx4_qsi4c32p4vlx4_1vlx4vl_sme2_mopa.c
|
||||
${KLEIDIAI_SRC}/kai/ukernels/matmul/matmul_clamp_f32_qsi8d32p_qsi4c32p/kai_matmul_clamp_f32_qsi8d32p1x4_qsi4c32p4vlx4_1x4vl_sme2_sdot.c
|
||||
${KLEIDIAI_SRC}/kai/ukernels/matmul/matmul_clamp_fp32_bf16p_bf16p/kai_matmul_clamp_f32_bf16p2vlx2_bf16p2vlx2_2vlx2vl_sme2_mopa.c
|
||||
${KLEIDIAI_SRC}/kai/ukernels/matmul/matmul_clamp_fp32_bf16p_bf16p/kai_matmul_clamp_f32_bf16p2vlx2_bf16p2vlx2_2vlx2vl_sme2_mopa_asm.S
|
||||
${KLEIDIAI_SRC}/kai/ukernels/matmul/pack/kai_lhs_pack_bf16p2vlx2_f32_sme.c
|
||||
${KLEIDIAI_SRC}/kai/ukernels/matmul/pack/kai_rhs_pack_kxn_bf16p2vlx2b_f32_x32_sme.c
|
||||
${KLEIDIAI_SRC}/kai/kai_common_sme_asm.S)
|
||||
|
||||
Reference in New Issue
Block a user