Wire up ccache and pigz in the docker based build

This should help speed things up a little

Dockerfile

@@ -19,7 +19,8 @@ COPY --from=llm-code / /go/src/github.com/ollama/ollama/
 WORKDIR /go/src/github.com/ollama/ollama/llm/generate
 ARG CGO_CFLAGS
 ENV GOARCH amd64 
-RUN OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_SKIP_CPU_GENERATE=1 sh gen_linux.sh
+RUN --mount=type=cache,target=/root/.ccache \
+    OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_SKIP_CPU_GENERATE=1 bash gen_linux.sh
 
 FROM --platform=linux/arm64 nvidia/cuda:$CUDA_VERSION-devel-rockylinux8 AS cuda-build-arm64
 ARG CMAKE_VERSION
@@ -30,7 +31,12 @@ COPY --from=llm-code / /go/src/github.com/ollama/ollama/
 WORKDIR /go/src/github.com/ollama/ollama/llm/generate
 ARG CGO_CFLAGS
 ENV GOARCH arm64 
-RUN OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_SKIP_CPU_GENERATE=1 sh gen_linux.sh
+RUN --mount=type=cache,target=/root/.ccache \
+    OLLAMA_SKIP_STATIC_GENERATE=1 \
+    OLLAMA_SKIP_CPU_GENERATE=1 \
+    CMAKE_CUDA_ARCHITECTURES="${CUDA_V11_ARCHITECTURES}" \
+    CUDA_VARIANT="_v11" \
+    bash gen_linux.sh
 
 FROM --platform=linux/amd64 rocm/dev-centos-7:${ROCM_VERSION}-complete AS rocm-build-amd64
 ARG CMAKE_VERSION
@@ -43,7 +49,8 @@ WORKDIR /go/src/github.com/ollama/ollama/llm/generate
 ARG CGO_CFLAGS
 ARG AMDGPU_TARGETS
 ENV GOARCH amd64 
-RUN OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_SKIP_CPU_GENERATE=1 sh gen_linux.sh
+RUN --mount=type=cache,target=/root/.ccache \
+    OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_SKIP_CPU_GENERATE=1 bash gen_linux.sh
 RUN mkdir -p ../../dist/linux-amd64/ollama_libs && \
     (cd /opt/rocm/lib && tar cf - rocblas/library) | (cd ../../dist/linux-amd64/ollama_libs && tar xf - )
 
@@ -60,13 +67,17 @@ ENV GOARCH amd64
 WORKDIR /go/src/github.com/ollama/ollama/llm/generate
 
 FROM --platform=linux/amd64 cpu-builder-amd64 AS static-build-amd64
-RUN OLLAMA_CPU_TARGET="static" sh gen_linux.sh
+RUN --mount=type=cache,target=/root/.ccache \
+    OLLAMA_CPU_TARGET="static" bash gen_linux.sh
 FROM --platform=linux/amd64 cpu-builder-amd64 AS cpu-build-amd64
-RUN OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu" sh gen_linux.sh
+RUN --mount=type=cache,target=/root/.ccache \
+    OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu" bash gen_linux.sh
 FROM --platform=linux/amd64 cpu-builder-amd64 AS cpu_avx-build-amd64
-RUN OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu_avx" sh gen_linux.sh
+RUN --mount=type=cache,target=/root/.ccache \
+    OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu_avx" bash gen_linux.sh
 FROM --platform=linux/amd64 cpu-builder-amd64 AS cpu_avx2-build-amd64
-RUN OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu_avx2" sh gen_linux.sh
+RUN --mount=type=cache,target=/root/.ccache \
+    OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu_avx2" bash gen_linux.sh
 
 FROM --platform=linux/arm64 rockylinux:8 AS cpu-builder-arm64
 ARG CMAKE_VERSION
@@ -81,9 +92,11 @@ ENV GOARCH arm64
 WORKDIR /go/src/github.com/ollama/ollama/llm/generate
 
 FROM --platform=linux/arm64 cpu-builder-arm64 AS static-build-arm64
-RUN OLLAMA_CPU_TARGET="static" sh gen_linux.sh
+RUN --mount=type=cache,target=/root/.ccache \
+    OLLAMA_CPU_TARGET="static" bash gen_linux.sh
 FROM --platform=linux/arm64 cpu-builder-arm64 AS cpu-build-arm64
-RUN OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu" sh gen_linux.sh
+RUN --mount=type=cache,target=/root/.ccache \
+    OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu" bash gen_linux.sh
 
 
 # Intermediate stage used for ./scripts/build_linux.sh
@@ -100,7 +113,8 @@ COPY --from=rocm-build-amd64 /go/src/github.com/ollama/ollama/dist/ dist/
 COPY --from=rocm-build-amd64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
 ARG GOFLAGS
 ARG CGO_CFLAGS
-RUN go build -trimpath -o dist/linux-amd64/ollama .
+RUN --mount=type=cache,target=/root/.ccache \
+    go build -trimpath -o dist/linux-amd64/ollama .
 
 # Intermediate stage used for ./scripts/build_linux.sh
 FROM --platform=linux/arm64 cpu-build-arm64 AS build-arm64
@@ -113,7 +127,8 @@ COPY --from=cuda-build-arm64 /go/src/github.com/ollama/ollama/dist/ dist/
 COPY --from=cuda-build-arm64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
 ARG GOFLAGS
 ARG CGO_CFLAGS
-RUN go build -trimpath -o dist/linux-arm64/ollama .
+RUN --mount=type=cache,target=/root/.ccache \
+    go build -trimpath -o dist/linux-arm64/ollama .
 
 # Runtime stages
 FROM --platform=linux/amd64 ubuntu:22.04 as runtime-amd64

llm/generate/gen_common.sh

@@ -47,6 +47,7 @@ init_vars() {
     if [ -z "${CMAKE_CUDA_ARCHITECTURES}" ] ; then
         CMAKE_CUDA_ARCHITECTURES="50;52;61;70;75;80"
     fi
+    GZIP=$(which pigz 2>/dev/null || echo "gzip")
 }
 
 git_module_setup() {
@@ -90,21 +91,23 @@ build() {
 
 compress() {
     echo "Compressing payloads to reduce overall binary size..."
-    pids=""
     rm -rf ${BUILD_DIR}/bin/*.gz
     for f in ${BUILD_DIR}/bin/* ; do
-        gzip -n --best -f ${f} &
+        ${GZIP} -n --best -f ${f} &
-        pids+=" $!"
+        compress_pids+=" $!"
     done
     # check for lib directory
     if [ -d ${BUILD_DIR}/lib ]; then
         for f in ${BUILD_DIR}/lib/* ; do
-            gzip -n --best -f ${f} &
+            ${GZIP} -n --best -f ${f} &
-            pids+=" $!"
+            compress_pids+=" $!"
         done
     fi
     echo
-    for pid in ${pids}; do
+}
+
+wait_for_compress() {
+    for pid in ${compress_pids}; do
         wait $pid
     done
     echo "Finished compression"

llm/generate/gen_darwin.sh

@@ -6,6 +6,7 @@
 
 set -ex
 set -o pipefail
+compress_pids=""
 echo "Starting darwin generate script"
 source $(dirname $0)/gen_common.sh
 init_vars
@@ -98,4 +99,5 @@ case "${GOARCH}" in
 esac
 
 cleanup
+wait_for_compress
 echo "go generate completed.  LLM runners: $(cd ${BUILD_DIR}/..; echo *)"

llm/generate/gen_linux.sh

@@ -13,6 +13,7 @@
 
 set -ex
 set -o pipefail
+compress_pids=""
 
 # See https://llvm.org/docs/AMDGPUUsage.html#processors for reference
 amdGPUs() {
@@ -274,4 +275,5 @@ if [ -z "${OLLAMA_SKIP_ROCM_GENERATE}" -a -d "${ROCM_PATH}" ]; then
 fi
 
 cleanup
+wait_for_compress
 echo "go generate completed.  LLM runners: $(cd ${BUILD_DIR}/..; echo *)"

scripts/build_linux.sh

@@ -4,6 +4,7 @@ set -eu
 
 export VERSION=${VERSION:-$(git describe --tags --first-parent --abbrev=7 --long --dirty --always | sed -e "s/^v//g")}
 export GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=$VERSION\" \"-X=github.com/ollama/ollama/server.mode=release\"'"
+GZIP=$(which pigz 2>/dev/null || echo "gzip")
 
 BUILD_ARCH=${BUILD_ARCH:-"amd64 arm64"}
 export AMDGPU_TARGETS=${AMDGPU_TARGETS:=""}
@@ -25,5 +26,5 @@ for TARGETARCH in ${BUILD_ARCH}; do
     docker rm builder-$TARGETARCH
     echo "Compressing final linux bundle..."
     rm -f ./dist/ollama-linux-$TARGETARCH.tgz
-    (cd dist/linux-$TARGETARCH && tar cf - . | gzip --best > ../ollama-linux-$TARGETARCH.tgz )
+    (cd dist/linux-$TARGETARCH && tar cf - . | ${GZIP} --best > ../ollama-linux-$TARGETARCH.tgz )
 done

scripts/rh_linux_deps.sh

@@ -3,6 +3,7 @@
 # Script for common Dockerfile dependency installation in redhat linux based images
 
 set -ex
+set -o pipefail
 MACHINE=$(uname -m)
 
 if grep -i "centos" /etc/system-release >/dev/null; then
@@ -29,7 +30,7 @@ if grep -i "centos" /etc/system-release >/dev/null; then
         dnf install -y rh-git227-git
         ln -s /opt/rh/rh-git227/root/usr/bin/git /usr/local/bin/git
     fi
-    dnf install -y devtoolset-10-gcc devtoolset-10-gcc-c++
+    dnf install -y devtoolset-10-gcc devtoolset-10-gcc-c++ pigz
 elif grep -i "rocky" /etc/system-release >/dev/null; then
     # Temporary workaround until rocky 8 AppStream ships GCC 10.4 (10.3 is incompatible with NVCC)
     cat << EOF > /etc/yum.repos.d/Rocky-Vault.repo
@@ -43,12 +44,21 @@ gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-rockyofficial
 EOF
     dnf install -y git \
         gcc-toolset-10-gcc-10.2.1-8.2.el8 \
-        gcc-toolset-10-gcc-c++-10.2.1-8.2.el8
+        gcc-toolset-10-gcc-c++-10.2.1-8.2.el8 \
+        pigz
 else
     echo "ERROR Unexpected distro"
     exit 1
 fi
 
+if [ "${MACHINE}" = "x86_64" ] ; then
+    curl -s -L https://github.com/ccache/ccache/releases/download/v4.10.2/ccache-4.10.2-linux-x86_64.tar.xz | tar -Jx -C /tmp --strip-components 1 && \
+    mv /tmp/ccache /usr/local/bin/
+else
+    yum -y install epel-release
+    yum install -y ccache
+fi
+
 if [ -n "${CMAKE_VERSION}" ]; then
     curl -s -L https://github.com/Kitware/CMake/releases/download/v${CMAKE_VERSION}/cmake-${CMAKE_VERSION}-linux-$(uname -m).tar.gz | tar -zx -C /usr --strip-components 1
 fi