cgo: use O3

discover/gpu.go

@@ -3,6 +3,7 @@
 package discover
 
 /*
+#cgo CPPFLAGS: -O3
 #cgo linux LDFLAGS: -lrt -lpthread -ldl -lstdc++ -lm
 #cgo windows LDFLAGS: -lpthread
 

discover/gpu_darwin.go

@@ -4,6 +4,7 @@ package discover
 
 /*
 #cgo CFLAGS: -x objective-c
+#cgo CPPFLAGS: -O3
 #cgo LDFLAGS: -framework Foundation -framework CoreGraphics -framework Metal
 #include "gpu_info_darwin.h"
 */

llama/llama.cpp/examples/llava/llava.go

 package llava
 
 // #cgo CXXFLAGS: -std=c++11
+// #cgo CPPFLAGS: -O3
 // #cgo CPPFLAGS: -I${SRCDIR}/../../include -I${SRCDIR}/../../common
 // #cgo CPPFLAGS: -I${SRCDIR}/../../../../ml/backend/ggml/ggml/include
 import "C"

llama/llama.cpp/src/llama.go

 package llama
 
 // #cgo CXXFLAGS: -std=c++17
+// #cgo CPPFLAGS: -O3
 // #cgo CPPFLAGS: -I${SRCDIR}/../include
 // #cgo CPPFLAGS: -I${SRCDIR}/../../../ml/backend/ggml/ggml/include
 // #cgo windows CPPFLAGS: -D_WIN32_WINNT=0x0602

llama/llama.go

@@ -3,6 +3,7 @@ package llama
 /*
 #cgo CFLAGS: -std=c11
 #cgo CXXFLAGS: -std=c++17
+#cgo CPPFLAGS: -O3
 #cgo CPPFLAGS: -I${SRCDIR}/llama.cpp/include
 #cgo CPPFLAGS: -I${SRCDIR}/llama.cpp/common
 #cgo CPPFLAGS: -I${SRCDIR}/llama.cpp/examples/llava

ml/backend/ggml/ggml/src/ggml-blas/blas.go

@@ -3,6 +3,7 @@
 package blas
 
 // #cgo CXXFLAGS: -std=c++11
+// #cgo CPPFLAGS: -O3
 // #cgo CPPFLAGS: -DGGML_USE_BLAS
 // #cgo CPPFLAGS: -I${SRCDIR}/.. -I${SRCDIR}/../../include
 // #cgo darwin,arm64 CPPFLAGS: -DGGML_BLAS_USE_ACCELERATE -DACCELERATE_NEW_LAPACK -DACCELERATE_LAPACK_ILP64

ml/backend/ggml/ggml/src/ggml-cpu/cpu.go

@@ -2,9 +2,11 @@ package cpu
 
 // #cgo CFLAGS: -Wno-implicit-function-declaration
 // #cgo CXXFLAGS: -std=c++17
-// #cgo CPPFLAGS: -I${SRCDIR}/amx -I${SRCDIR}/llamafile -I${SRCDIR}/.. -I${SRCDIR}/../../include
+// #cgo CPPFLAGS: -O3
 // #cgo CPPFLAGS: -DGGML_USE_LLAMAFILE
+// #cgo CPPFLAGS: -I${SRCDIR}/amx -I${SRCDIR}/llamafile -I${SRCDIR}/.. -I${SRCDIR}/../../include
 // #cgo linux CPPFLAGS: -D_GNU_SOURCE
+// #cgo arm64 CPPFLAGS: -DGGML_USE_AARCH64
 // #cgo darwin,arm64 CPPFLAGS: -DGGML_USE_ACCELERATE -DACCELERATE_NEW_LAPACK -DACCELERATE_LAPACK_ILP64
 // #cgo darwin,arm64 LDFLAGS: -framework Accelerate
 import "C"

ml/backend/ggml/ggml/src/ggml-cpu/llamafile/llamafile.go

 package llamafile
 
 // #cgo CXXFLAGS: -std=c++17
+// #cgo CPPFLAGS: -O3
 // #cgo CPPFLAGS: -I${SRCDIR}/.. -I${SRCDIR}/../.. -I${SRCDIR}/../../../include
 import "C"

ml/backend/ggml/ggml/src/ggml-metal/metal.go

@@ -4,6 +4,7 @@ package metal
 
 //go:generate sh -c "{ echo // Code generated $(date). DO NOT EDIT.; sed -e '/__embed_ggml-common.h__/r ../ggml-common.h' -e '/__embed_ggml-common.h__/d' -e '/#include \"ggml-metal-impl.h\"/r ggml-metal-impl.h' -e '/#include \"ggml-metal-impl.h\"/d' ggml-metal.metal; } >ggml-metal-embed.metal"
 
+// #cgo CPPFLAGS: -O3
 // #cgo CPPFLAGS: -DGGML_METAL_EMBED_LIBRARY -I.. -I../../include
 // #cgo LDFLAGS: -framework Metal -framework MetalKit
 import "C"

ml/backend/ggml/ggml/src/ggml.go

 package ggml
 
 // #cgo CXXFLAGS: -std=c++17
+// #cgo CPPFLAGS: -O3
 // #cgo CPPFLAGS: -DNDEBUG -DGGML_USE_CPU
 // #cgo CPPFLAGS: -I${SRCDIR}/../include -I${SRCDIR}/ggml-cpu
 // #cgo windows LDFLAGS: -lmsvcrt -static -static-libgcc -static-libstdc++

ml/backend/ggml/ggml/src/ggml_darwin_arm64.go

 package ggml
 
+// #cgo CPPFLAGS: -O3
 // #cgo CPPFLAGS: -DGGML_USE_METAL -DGGML_USE_BLAS
 // #cgo LDFLAGS: -framework Foundation
 import "C"