ggml-hip : remove -funsafe-math-optimizations (#26696)

It enables -fassociative-math, which reassociates FP reductions and can flip
greedy argmax on RDNA3.5 (e.g. MTP speculative decode diverging from the
non-speculative baseline). Drop it so HIP builds are IEEE-conformant.

Co-authored-by: Jim Wu <redacted>

diff --git a/ggml/src/ggml-hip/CMakeLists.txt b/ggml/src/ggml-hip/CMakeLists.txt
index bbc51797c1827961ac4541361f4e0fde3974ca51..47f16f56c470f263816bab3fb9faae874f3c2d0a 100644 (file)
--- a/ggml/src/ggml-hip/CMakeLists.txt
+++ b/ggml/src/ggml-hip/CMakeLists.txt
@@ -126,9 +126,6 @@ if (GGML_HIP_EXPORT_METRICS)
     set(CMAKE_HIP_FLAGS "${CMAKE_HIP_FLAGS} -Rpass-analysis=kernel-resource-usage --save-temps")
 endif()
 
-# Fast math for HIP, like CUDA's -use_fast_math. Not -ffast-math: that implies -ffinite-math-only, which breaks ggml's INFINITY masking and produces NaNs.
-set(CMAKE_HIP_FLAGS "${CMAKE_HIP_FLAGS} -funsafe-math-optimizations")
-
 if (NOT GGML_CUDA_FA)
     add_compile_definitions(GGML_CUDA_NO_FA)
 endif()