HIP: add cmake option to enable compiler output of kernel resource usage metrics...

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 7c4666b7edb2fe75f7db312d5be1ed76d967babb..1fb7abeaf088f34e1a19946d3ea1afee405e37c7 100644 (file)
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -176,6 +176,7 @@ option(GGML_HIP_NO_VMM                      "ggml: do not try to use HIP VMM"
 option(GGML_HIP_ROCWMMA_FATTN               "ggml: enable rocWMMA for FlashAttention"         OFF)
 option(GGML_HIP_FORCE_ROCWMMA_FATTN_GFX12   "ggml: enable rocWMMA FlashAttention on GFX12"    OFF)
 option(GGML_HIP_MMQ_MFMA                    "ggml: enable MFMA MMA for CDNA in MMQ"           ON)
+option(GGML_HIP_EXPORT_METRICS              "ggml: enable kernel perf metrics output"         OFF)
 option(GGML_MUSA_GRAPHS                     "ggml: use MUSA graph, experimental, unstable"    OFF)
 option(GGML_MUSA_MUDNN_COPY                 "ggml: enable muDNN for accelerated copy"         OFF)
 option(GGML_VULKAN                          "ggml: use Vulkan"                                OFF)

diff --git a/src/ggml-hip/CMakeLists.txt b/src/ggml-hip/CMakeLists.txt
index e92ec7faa33244169b349f4b1fcf4da461f618f3..852de973460252415da3db528b56b866e19cd720 100644 (file)
--- a/src/ggml-hip/CMakeLists.txt
+++ b/src/ggml-hip/CMakeLists.txt
@@ -121,6 +121,10 @@ if (GGML_HIP_FORCE_ROCWMMA_FATTN_GFX12 OR ${hip_VERSION} VERSION_GREATER_EQUAL 7
     add_compile_definitions(GGML_HIP_ROCWMMA_FATTN_GFX12)
 endif()
 
+if (GGML_HIP_EXPORT_METRICS)
+    set(CMAKE_HIP_FLAGS "${CMAKE_HIP_FLAGS} -Rpass-analysis=kernel-resource-usage --save-temps")
+endif()
+
 if (NOT GGML_CUDA_FA)
     add_compile_definitions(GGML_CUDA_NO_FA)
 endif()