haraldwolff/llama.cpp
1
0
Fork 0
Code Issues Pull requests Packages Projects Releases Wiki Activity

metal : switch back to default.metallib (ggml/681)

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2024-01-05 16:30:52 +02:00
parent d061bf9405
commit 91d38876df
4 changed files with 11 additions and 26 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
CMakeLists.txt
View file

@ -177,27 +177,29 @@ if (LLAMA_METAL)
if (LLAMA_METAL_SHADER_DEBUG)
# custom command to do the following:
# xcrun -sdk macosx metal -fno-fast-math -c ggml-metal.metal -o ggml-metal.air
# xcrun -sdk macosx metallib ggml-metal.air -o ggml.metallib
# xcrun -sdk macosx metallib ggml-metal.air -o default.metallib
#
# note: this is the only way I found to disable fast-math in Metal. it's ugly, but at least it works
# disabling fast math is needed in order to pass tests/test-backend-ops
# note: adding -fno-inline fixes the tests when using MTL_SHADER_VALIDATION=1
# note: unfortunately, we have to call it default.metallib instead of ggml.metallib
# ref: https://github.com/ggerganov/whisper.cpp/issues/1720
set(XC_FLAGS -fno-fast-math -fno-inline -g)
if (LLAMA_QKK_64)
set(XC_FLAGS ${XC_FLAGS} -DQK_K=64)
endif()
add_custom_command(
OUTPUT ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/ggml.metallib
OUTPUT ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/default.metallib
COMMAND xcrun -sdk macosx metal ${XC_FLAGS} -c ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/ggml-metal.metal -o ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/ggml-metal.air
COMMAND xcrun -sdk macosx metallib ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/ggml-metal.air -o ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/ggml.metallib
COMMAND xcrun -sdk macosx metallib ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/ggml-metal.air -o ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/default.metallib
DEPENDS ggml-metal.metal
COMMENT "Compiling Metal kernels"
)
add_custom_target(
ggml-metal ALL
DEPENDS ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/ggml.metallib
DEPENDS ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/default.metallib
)
endif()

19
examples/llama.swiftui/llama.swiftui.xcodeproj/project.pbxproj
View file

@ -23,26 +23,10 @@
8A3F84242AC4C891005E2EE8 /* models in Resources */ = {isa = PBXBuildFile; fileRef = 8A3F84232AC4C891005E2EE8 /* models */; };
8A907F332AC7138A006146EA /* LibLlama.swift in Sources */ = {isa = PBXBuildFile; fileRef = 8A907F322AC7134E006146EA /* LibLlama.swift */; };
8A9F7C4D2AC332EE008AE1EA /* LlamaState.swift in Sources */ = {isa = PBXBuildFile; fileRef = 8A9F7C4C2AC332EE008AE1EA /* LlamaState.swift */; };
F1FE20E22B465ECA00B45541 /* LoadCustomButton.swift in Sources */ = {isa = PBXBuildFile; fileRef = F1FE20E12B465EC900B45541 /* LoadCustomButton.swift */; };
F1FE20DC2B465C4500B45541 /* ggml-metal.metal in Resources */ = {isa = PBXBuildFile; fileRef = 549479C82AC9E10B00E0F78B /* ggml-metal.metal */; };
F1FE20E22B465ECA00B45541 /* LoadCustomButton.swift in Sources */ = {isa = PBXBuildFile; fileRef = F1FE20E12B465EC900B45541 /* LoadCustomButton.swift */; };
/* End PBXBuildFile section */
/* Begin PBXBuildRule section */
F1FE20DB2B465C2100B45541 /* PBXBuildRule */ = {
isa = PBXBuildRule;
compilerSpec = com.apple.compilers.proxy.script;
fileType = sourcecode.metal;
inputFiles = (
);
isEditable = 1;
outputFiles = (
"${DERIVED_FILES_DIR}/ggml-metal.air",
"${DERIVED_FILES_DIR}/ggml.metallib",
);
script = "# metal\nxcrun metal -c \"${INPUT_FILE_PATH}\" -o \"${DERIVED_FILES_DIR}/${INPUT_FILE_BASE}.air\"\nxcrun metallib -o \"${DERIVED_FILES_DIR}/${INPUT_FILE_BASE%-metal}.metallib\" \"${DERIVED_FILES_DIR}/${INPUT_FILE_BASE}.air\"\n";
};
/* End PBXBuildRule section */
/* Begin PBXFileReference section */
542376062B0D9BEA008E6A1C /* ggml-quants.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = "ggml-quants.h"; path = "../../ggml-quants.h"; sourceTree = "<group>"; };
542376072B0D9BFB008E6A1C /* ggml-quants.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; name = "ggml-quants.c"; path = "../../ggml-quants.c"; sourceTree = "<group>"; };
@ -209,7 +193,6 @@
8A1C83712AC328BD0096AF73 /* Resources */,
);
buildRules = (
F1FE20DB2B465C2100B45541 /* PBXBuildRule */,
);
dependencies = (
);

6
ggml-metal.m
View file

@ -258,14 +258,14 @@ struct ggml_metal_context * ggml_metal_init(int n_cb) {
bundle = [NSBundle bundleForClass:[GGMLMetalClass class]];
#endif
NSError * error = nil;
NSString * libPath = [bundle pathForResource:@"ggml" ofType:@"metallib"];
NSString * libPath = [bundle pathForResource:@"default" ofType:@"metallib"];
if (libPath != nil) {
// pre-compiled library found
NSURL * libURL = [NSURL fileURLWithPath:libPath];
GGML_METAL_LOG_INFO("%s: loading '%s'\n", __func__, [libPath UTF8String]);
ctx->library = [ctx->device newLibraryWithURL:libURL error:&error];
} else {
GGML_METAL_LOG_INFO("%s: ggml.metallib not found, loading from source\n", __func__);
GGML_METAL_LOG_INFO("%s: default.metallib not found, loading from source\n", __func__);
NSString * sourcePath;
NSString * ggmlMetalPathResources = [[NSProcessInfo processInfo].environment objectForKey:@"GGML_METAL_PATH_RESOURCES"];
@ -295,7 +295,7 @@ struct ggml_metal_context * ggml_metal_init(int n_cb) {
#endif
// try to disable fast-math
// NOTE: this seems to have no effect whatsoever
// instead, in order to disable fast-math, we have to build ggml.metallib from the command line
// instead, in order to disable fast-math, we have to build default.metallib from the command line
// using xcrun -sdk macosx metal -fno-fast-math -c ggml-metal.metal -o ggml-metal.air
// and go through the "pre-compiled library found" path above
//[options setFastMathEnabled:false];

2
scripts/sync-ggml.last
View file

@ -1 +1 @@
3fd01e00e40583ccd4b393a7c6502d6a4455a1d5
f96711108d55bdbbd277e6be07204dce6a94fb93