darwin: restore multiple runners for x86 (#8125)

In 0.5.2 we simplified packaging to have avx only for macos x86.  It looks like
there may still be some non-AVX systems out there, so this puts back the prior
logic of building no-AVX for the primary binary, and now 2 runners for avx and avx2.
These will be packaged in the App bundle only, so the stand-alone binary will now be
without AVX support on macos.  On arm, we'll also see these runners reported
as available in the log, but they're dormant and will never be used at runtime.

Makefile

@@ -8,11 +8,9 @@ include make/cuda-v12-defs.make
 include make/rocm-defs.make
 
 ifeq ($(CUSTOM_CPU_FLAGS),)
-ifneq ($(OS),darwin)
 ifeq ($(ARCH),amd64)
 	RUNNER_TARGETS=cpu
 endif
-endif
 # Without CUSTOM_CPU_FLAGS we default to build both v11 and v12 if present
 ifeq ($(OLLAMA_SKIP_CUDA_GENERATE),)
 ifneq ($(CUDA_11_COMPILER),)

macapp/forge.config.ts

@@ -19,6 +19,7 @@ const config: ForgeConfig = {
     icon: './assets/icon.icns',
     extraResource: [
       '../dist/ollama',
+      '../dist/darwin-amd64/lib',
       path.join(__dirname, './assets/iconTemplate.png'),
       path.join(__dirname, './assets/iconTemplate@2x.png'),
       path.join(__dirname, './assets/iconUpdateTemplate.png'),
@@ -42,7 +43,7 @@ const config: ForgeConfig = {
         }
       : {}),
     osxUniversal: {
-      x64ArchFiles: '**/ollama',
+      x64ArchFiles: '**/ollama*',
     },
   },
   rebuildConfig: {},

runners/common.go

@@ -72,6 +72,7 @@ func locateRunnersOnce() {
 	paths := []string{
 		filepath.Join(filepath.Dir(exe), "llama", "build", runtime.GOOS+"-"+runtime.GOARCH, "runners"),
 		filepath.Join(filepath.Dir(exe), envconfig.LibRelativeToExe(), "lib", "ollama", "runners"),
+		filepath.Join(filepath.Dir(exe), "lib", "ollama", "runners"),
 	}
 	for _, path := range paths {
 		if _, err := os.Stat(path); err == nil {

scripts/build_darwin.sh

@@ -18,10 +18,18 @@ rm -rf llama/build dist/darwin-*
 echo "Building darwin arm64"
 GOOS=darwin ARCH=arm64 GOARCH=arm64 make -j 8 dist
 echo "Building darwin amd64 with AVX enabled"
-GOOS=darwin ARCH=amd64 GOARCH=amd64 CUSTOM_CPU_FLAGS="avx" make -j 8 dist
+GOOS=darwin ARCH=amd64 GOARCH=amd64 CUSTOM_CPU_FLAGS="avx" make -j 8 dist_exe
 
+# Generate the universal ollama binary for stand-alone usage: metal + avx
+lipo -create -output dist/ollama-darwin dist/darwin-arm64/bin/ollama dist/darwin-amd64/bin/ollama
 
+echo "Building darwin amd64 with runners"
+rm dist/darwin-amd64/bin/ollama
+GOOS=darwin ARCH=amd64 GOARCH=amd64 make -j 8 dist
+# Generate the universal ollama binary for the app bundle: metal + no-avx
 lipo -create -output dist/ollama dist/darwin-arm64/bin/ollama dist/darwin-amd64/bin/ollama
+
+
 if [ -n "$APPLE_IDENTITY" ]; then
     codesign --deep --force --options=runtime --sign "$APPLE_IDENTITY" --timestamp dist/ollama
 else
@@ -48,5 +56,4 @@ ditto -c -k --keepParent dist/ollama dist/temp.zip
 if [ -n "$APPLE_IDENTITY" ]; then
     xcrun notarytool submit dist/temp.zip --wait --timeout 10m --apple-id $APPLE_ID --password $APPLE_PASSWORD --team-id $APPLE_TEAM_ID
 fi
-mv dist/ollama dist/ollama-darwin
 rm -f dist/temp.zip