update(llama.cpp): update server, correctly propagate LLAMA_VERSION (#1440)

* fix(Makefile): correctly propagate LLAMA_VERSION

Signed-off-by: Ettore Di Giacinto <mudler@users.noreply.github.com>

* update grpc-server.cpp

---------

Signed-off-by: Ettore Di Giacinto <mudler@users.noreply.github.com>

Author: mudler

Makefile

@@ -36,6 +36,7 @@ STABLEDIFFUSION_VERSION?=902db5f066fd137697e3b69d0fa10d4782bd2c2f
 export BUILD_TYPE?=
 export STABLE_BUILD_TYPE?=$(BUILD_TYPE)
 export CMAKE_ARGS?=
+
 CGO_LDFLAGS?=
 CUDA_LIBPATH?=/usr/local/cuda/lib64/
 GO_TAGS?=
@@ -229,7 +230,7 @@ sources/go-piper/libpiper_binding.a: sources/go-piper
 	$(MAKE) -C sources/go-piper libpiper_binding.a example/main
 
 backend/cpp/llama/llama.cpp:
-	$(MAKE) -C backend/cpp/llama llama.cpp	
+	LLAMA_VERSION=$(CPPLLAMA_VERSION) $(MAKE) -C backend/cpp/llama llama.cpp	
 
 get-sources: backend/cpp/llama/llama.cpp sources/go-llama sources/go-llama-ggml sources/go-ggml-transformers sources/gpt4all sources/go-piper sources/go-rwkv sources/whisper.cpp sources/go-bert sources/go-stable-diffusion
 	touch $@

backend/cpp/llama/Makefile

@@ -1,5 +1,5 @@
 
-LLAMA_VERSION?=d9b33fe95bd257b36c84ee5769cc048230067d6f
+LLAMA_VERSION?=
 
 CMAKE_ARGS?=
 BUILD_TYPE?=
@@ -21,6 +21,9 @@ endif
 
 llama.cpp:
 	git clone --recurse-submodules https://github.com/ggerganov/llama.cpp llama.cpp
+	if [ -z "$(LLAMA_VERSION)" ]; then \
+		exit 1; \
+	fi
 	cd llama.cpp && git checkout -b build $(LLAMA_VERSION) && git submodule update --init --recursive --depth 1
 
 llama.cpp/examples/grpc-server: