diff --git a/Makefile b/Makefile
index b70ebaed5..68197fef8 100644
--- a/Makefile
+++ b/Makefile
@@ -554,7 +554,7 @@ 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
@@ -983,9 +983,9 @@ 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 aae5b8e9f..c5ee7e425 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 c0aced3d2..1bb731e16 100644
--- a/ggml/src/ggml.c
+++ b/ggml/src/ggml.c
@@ -6,7 +6,6 @@
 #include "ggml.h"
 #include "ggml-aarch64.h"
 
-
 #if defined(_MSC_VER) || defined(__MINGW32__)
 #include <malloc.h> // using malloc.h with MSC/MINGW
 #elif !defined(__FreeBSD__) && !defined(__NetBSD__) && !defined(__OpenBSD__)
@@ -43,7 +42,7 @@
 #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