run.remoting: update to run llama-server

Author: kpouget

build.remoting.sh

@@ -0,0 +1,25 @@
+# force isatty-->true, so that $0 |& head -50 has colors ...
+rm -f READY FAILED
+
+echo "int isatty(int fd) { return 1; }" | gcc -O2 -fpic -shared -ldl -o /tmp/isatty.so -xc -
+export LD_PRELOAD=/tmp/isatty.so
+
+TARGETS="ggml-remotingfrontend"
+
+TARGETS="$BUILD_TARGET llama-run"
+set -x
+if [[ "${BENCH_MODE:-}" == "bench" ]]; then
+    TARGETS="$TARGETS llama-bench"
+elif [[ "${BENCH_MODE:-}" == "server" ]]; then
+    TARGETS="$TARGETS llama-server"
+elif [[ "${BENCH_MODE:-}" == "perf" ]]; then
+    TARGETS="$TARGETS test-backend-ops"
+fi
+
+cmake --build ../build.remoting-frontend$FLAVOR --parallel 8 --target $TARGETS "$@"
+
+if [[ $? == 0 ]]; then
+    touch READY
+else
+    touch FAILED
+fi

run.remoting.sh

@@ -18,6 +18,20 @@ set -x
 if [[ "${BENCH_MODE:-}" == "bench" ]]; then
     cat <<EOF
 ###
+### Running llama-server
+###
+
+EOF
+    $prefix \
+        $LLAMA_BUILD_DIR/bin/llama-server \
+        --host 0.0.0.0 \
+        --port 8080 \
+        --model "$MODEL_HOME/$MODEL" \
+        --n-gpu-layers 99 \
+        --threads 1
+elif [[ "${BENCH_MODE:-}" == "bench" ]]; then
+    cat <<EOF
+###
 ### Running llama-bench
 ###