Open RDearnaley opened 5 days ago
Hi, sorry about that, does it work if you change this additional line into the makefile like this
https://github.com/LostRuins/koboldcpp/commit/06d9068c97bdadbe5830d5a4a9a107c7bc112365#
ggml-metal.o: ggml/src/ggml-metal.m ggml/include/ggml-metal.h
@echo "== Preparing merged Metal file =="
@sed -e '/#include "ggml-common.h"/r ggml/src/ggml-common.h' -e '/#include "ggml-common.h"/d' < ggml/src/ggml-metal.metal > ggml/src/ggml-metal-merged.metal
@cp ggml/src/ggml-metal-merged.metal ./ggml-metal-merged.metal
$(CC) $(CFLAGS) -c $< -o $@
@beebopkim do you have the same issue?
@LostRuins That fixes it for me after a clean build.
I got the same error message as #744, again on Mac using Metal:
llama_new_context_with_model: failed to initialize Metal backend
Removing
-DGGML_METAL_NDEBUG
from theMakefile
as was done in #744 I got a different set of debug messages:Setting the environment variable
GGML_METAL_PATH_RESOURCES
to the location~/Documents/GitHub/koboldcpp/ggml/src
containing the file it's looking for solved this problem. So I have a workaround, but presumably this shouldn't be necessary.