From 9e83e550e1b0022e39de9bf2b84961cec3a0071c Mon Sep 17 00:00:00 2001
From: Daniel Hiltgen <dhiltgen@users.noreply.github.com>
Date: Thu, 7 Nov 2024 10:20:50 -0800
Subject: [PATCH] Align rocm compiler flags (#7467)

Bring consistency with the old generate script behavior
---
 llama/make/Makefile.rocm | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/llama/make/Makefile.rocm b/llama/make/Makefile.rocm
index 947c43a6..136ce015 100644
--- a/llama/make/Makefile.rocm
+++ b/llama/make/Makefile.rocm
@@ -58,6 +58,8 @@ endif
 GPU_COMPILER_CUFLAGS = \
 	$(GPU_COMPILER_FPIC) \
 	$(addprefix -m,$(GPU_RUNNER_CPU_FLAGS)) \
+	-mf16c \
+	-mfma \
 	-parallel-jobs=2 \
 	-c \
 	-O3 \
@@ -77,6 +79,9 @@ GPU_COMPILER_CUFLAGS = \
 	-D_CRT_SECURE_NO_WARNINGS \
 	-D_GNU_SOURCE \
 	-D_XOPEN_SOURCE=600 \
+	-DUSE_PROF_API=1 \
+	-std=gnu++14 \
+	-x hip \
 	-mllvm=-amdgpu-early-inline-all=true \
 	-mllvm=-amdgpu-function-calls=false \
 	-Wno-expansion-to-defined \