From d4c15504ee76931cc164f7893c3bc880df524991 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Tue, 9 Jul 2024 16:01:11 +0300
Subject: [PATCH] ggml : move sgemm sources to llamafile subfolder

ggml-ci
---
 Makefile                           | 8 ++++----
 ggml/CMakeLists.txt                | 2 +-
 ggml/src/CMakeLists.txt            | 6 +++---
 ggml/src/ggml.c                    | 3 +--
 ggml/src/{ => llamafile}/sgemm.cpp | 0
 ggml/src/{ => llamafile}/sgemm.h   | 0
 6 files changed, 9 insertions(+), 10 deletions(-)
 rename ggml/src/{ => llamafile}/sgemm.cpp (100%)
 rename ggml/src/{ => llamafile}/sgemm.h (100%)

diff --git a/Makefile b/Makefile
index bb6e2f968..8c3e60b77 100644
--- a/Makefile
+++ b/Makefile
@@ -545,7 +545,7 @@ endif # GGML_BLIS
 
 ifndef GGML_NO_LLAMAFILE
 	MK_CPPFLAGS += -DGGML_USE_LLAMAFILE
-	OBJ_GGML    += ggml/src/sgemm.o
+	OBJ_GGML    += ggml/src/llamafile/sgemm.o
 endif
 
 ifdef GGML_RPC
@@ -965,9 +965,9 @@ ggml/src/ggml-blas.o: \
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 
 ifndef GGML_NO_LLAMAFILE
-ggml/src/sgemm.o: \
-	ggml/src/sgemm.cpp \
-	ggml/src/sgemm.h \
+ggml/src/llamafile/sgemm.o: \
+	ggml/src/llamafile/sgemm.cpp \
+	ggml/src/llamafile/sgemm.h \
 	ggml/include/ggml.h
 	$(CXX) $(CXXFLAGS) -c $< -o $@
 endif # GGML_NO_LLAMAFILE
diff --git a/ggml/CMakeLists.txt b/ggml/CMakeLists.txt
index 0d0d52d57..649ac3dcc 100644
--- a/ggml/CMakeLists.txt
+++ b/ggml/CMakeLists.txt
@@ -104,7 +104,7 @@ option(GGML_ACCELERATE                      "ggml: enable Accelerate framework"
 option(GGML_BLAS                            "ggml: use BLAS"                                  ${GGML_BLAS_DEFAULT})
 set(GGML_BLAS_VENDOR ${GGML_BLAS_VENDOR_DEFAULT} CACHE STRING
                                             "ggml: BLAS library vendor")
-option(GGML_LLAMAFILE                       "ggml: use ggml SGEMM"                            OFF)
+option(GGML_LLAMAFILE                       "ggml: use LLAMAFILE"                             OFF)
 
 option(GGML_CUDA                            "ggml: use CUDA"                                  OFF)
 option(GGML_CUDA_FORCE_DMMV                 "ggml: use dmmv instead of mmvq CUDA kernels"     OFF)
diff --git a/ggml/src/CMakeLists.txt b/ggml/src/CMakeLists.txt
index c6694df67..083326d68 100644
--- a/ggml/src/CMakeLists.txt
+++ b/ggml/src/CMakeLists.txt
@@ -238,12 +238,12 @@ if (GGML_BLAS)
 endif()
 
 if (GGML_LLAMAFILE)
-    message(STATUS "Using ggml SGEMM")
+    message(STATUS "Using llamafile")
 
     add_compile_definitions(GGML_USE_LLAMAFILE)
 
-    set(GGML_HEADERS_LLAMAFILE sgemm.h)
-    set(GGML_SOURCES_LLAMAFILE sgemm.cpp)
+    set(GGML_HEADERS_LLAMAFILE llamafile/sgemm.h)
+    set(GGML_SOURCES_LLAMAFILE llamafile/sgemm.cpp)
 endif()
 
 if (GGML_CUDA)
diff --git a/ggml/src/ggml.c b/ggml/src/ggml.c
index bc91ac3a7..013b219fd 100644
--- a/ggml/src/ggml.c
+++ b/ggml/src/ggml.c
@@ -5,7 +5,6 @@
 #include "ggml-quants.h"
 #include "ggml.h"
 
-
 #if defined(_MSC_VER) || defined(__MINGW32__)
 #include <malloc.h> // using malloc.h with MSC/MINGW
 #elif !defined(__FreeBSD__) && !defined(__NetBSD__) && !defined(__OpenBSD__)
@@ -42,7 +41,7 @@
 #endif
 
 #ifdef GGML_USE_LLAMAFILE
-#include "sgemm.h"
+#include <llamafile/sgemm.h>
 #endif
 
 #if defined(_MSC_VER)
diff --git a/ggml/src/sgemm.cpp b/ggml/src/llamafile/sgemm.cpp
similarity index 100%
rename from ggml/src/sgemm.cpp
rename to ggml/src/llamafile/sgemm.cpp
diff --git a/ggml/src/sgemm.h b/ggml/src/llamafile/sgemm.h
similarity index 100%
rename from ggml/src/sgemm.h
rename to ggml/src/llamafile/sgemm.h