vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

why nobody ever told me the makefile doesnt work outside x86 xD

Browse source
This commit is contained in:
Concedo 2023-04-05 17:15:42 +08:00
parent 1490cdd71d
commit 5c1920df43
1 changed files with 3 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

6
Makefile
View file

@ -36,7 +36,7 @@ CXXFLAGS = -I. -I./examples -Ofast -DNDEBUG -std=c++11 -fPIC
LDFLAGS =
#lets try enabling everything
CFLAGS += -pthread -mf16c -mfma -mavx2 -mavx -msse3
CFLAGS += -pthread
CXXFLAGS += -pthread
# OS specific
@ -71,7 +71,7 @@ endif
# feel free to update the Makefile for your architecture and send a pull request or issue
ifeq ($(UNAME_M),$(filter $(UNAME_M),x86_64 i686))
# Use all CPU extensions that are available:
CFLAGS +=
CFLAGS += -mf16c -mfma -mavx2 -mavx -msse3
endif
ifneq ($(filter ppc64%,$(UNAME_M)),)
POWER9_M := $(shell grep "POWER9" /proc/cpuinfo)
@ -146,7 +146,7 @@ default: llamalib quantize llamalib_blas
#
ggml.o: ggml.c ggml.h
$(CC) $(CFLAGS) -c ggml.c -o ggml.o
$(CC) $(CFLAGS) -c ggml.c -o ggml.o
ggml_blas.o: ggml.c ggml.h
$(CC) $(CFLAGS) -DGGML_USE_OPENBLAS -c ggml.c -o ggml_blas.o