From: fxzjshm <redacted>
Date: Tue, 4 Feb 2025 18:18:38 +0000 (+0800)
Subject: HIP: force max threads per block to be 1024 (llama/11621)
X-Git-Tag: upstream/0.0.1722~26
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=9e65e640914f820c5715ea9366d258516c657ef3;p=pkg%2Fggml%2Fsources%2Fggml

HIP: force max threads per block to be 1024 (llama/11621)

Some old/vendor forked version of llvm still use 256. Explicitly set it to 1024 to align with upstream llvm.

Signed-off-by: fxzjshm <redacted>
---

diff --git a/src/ggml-hip/CMakeLists.txt b/src/ggml-hip/CMakeLists.txt
index eb03e10f..f4a46836 100644
--- a/src/ggml-hip/CMakeLists.txt
+++ b/src/ggml-hip/CMakeLists.txt
@@ -46,6 +46,9 @@ endif()
 
 message(STATUS "HIP and hipBLAS found")
 
+# Workaround old compilers
+set(CMAKE_HIP_FLAGS "${CMAKE_HIP_FLAGS} --gpu-max-threads-per-block=1024")
+
 file(GLOB   GGML_HEADERS_ROCM "../ggml-cuda/*.cuh")
 list(APPEND GGML_HEADERS_ROCM "../../include/ggml-cuda.h")