ggml-cpu : add check for ARM MATMUL_INT8/i8mm support (llama/15922)

This commit adds a check for GGML_MACHINE_SUPPORTS_i8mm when enabling
MATMUL_INT8 features, ensuring that i8mm intrinsics are only used when
the target hardware actually supports them.

The motivation for this is to fix ggml CI build failures where the
feature detection correctly identifies that i8mm is not supported,
adding the +noi8mm flag, but MATMUL_INT8 preprocessor definitions are
still enabled, causing the compiler to attempt to use vmmlaq_s32
intrinsics without i8mm support.

Refs: https://github.com/ggml-org/ggml/actions/runs/17525174120/job/49909199499

diff --git a/src/ggml-cpu/CMakeLists.txt b/src/ggml-cpu/CMakeLists.txt
index 388675f5f0911820830cb8dd366bb3e5fe2d9020..369905750754fb48df1fbac3dc1d5296e1766670 100644 (file)
--- a/src/ggml-cpu/CMakeLists.txt
+++ b/src/ggml-cpu/CMakeLists.txt
@@ -224,7 +224,13 @@ function(ggml_add_cpu_backend_variant_impl tag_name)
                 foreach(feature DOTPROD SVE MATMUL_INT8 FMA FP16_VECTOR_ARITHMETIC SME)
                     string(FIND "${ARM_FEATURE}" "__ARM_FEATURE_${feature} 1" feature_pos)
                     if (NOT ${feature_pos} EQUAL -1)
-                        message(STATUS "ARM feature ${feature} enabled")
+                        # Special handling for MATMUL_INT8 when machine doesn't support i8mm
+                        if ("${feature}" STREQUAL "MATMUL_INT8" AND GGML_MACHINE_SUPPORTS_noi8mm)
+                            message(STATUS "ARM feature ${feature} detected but unsetting due to machine not supporting i8mm")
+                            list(APPEND ARCH_FLAGS -U__ARM_FEATURE_MATMUL_INT8)
+                        else()
+                            message(STATUS "ARM feature ${feature} enabled")
+                        endif()
                     endif()
                 endforeach()
             endif()