uArch definition (PR 1/N) #2
Conversation
mshahneo
force-pushed
the
uarch_definition
branch
from
4666e33 to
5dd0ebe
Compare
| } // namespace mlir | ||
|
|
||
| #endif // MLIR_DIALECT_XEGPU_UTILS_INTEL_GPU_PVC_H | ||
| //===--- IntelGpuPVC.h ---------------------------------------*- C++ -*-===// |
There was a problem hiding this comment.
this one should be on the top I guess
There was a problem hiding this comment.
Ah, yes, I think I auto-added by co-pilot suggestion. Will remove it :)
|
A general question on the structure - what's the benefit of maintaining both yaml and C++ definitions? |
| Range systolic_depth; | ||
| Range repreat_count; | ||
| Range execution_size; | ||
| std::map<std::string, uint> ops_per_channel; |
There was a problem hiding this comment.
Opaque string mapping doesn't feel user friendly. Especially as it seems to be wrapping numerical bit width that could be a numerical key directly - I assume it's some limitation due to using (conversion from) yaml.
matrix_size is even worse offender in that regard. These wrappers are hard to use without yaml definitions which again makes me wonder if we really need to split it into two separate files.
There was a problem hiding this comment.
Opaque string mapping doesn't feel user friendly. Especially as it seems to be wrapping numerical bit width that could be a numerical key directly - I assume it's some limitation due to using (conversion from) yaml.
You are right, the limitation is due to the yaml mapping.
matrix_size is even worse offender in that regard. These wrappers are hard to use without yaml definitions which again makes me wonder if we really need to split it into two separate files.
Sorry, the matrix size is a mistake. The value should be a vector of uint. But yes, in general you are right, the structs and and yaml to some extent have to be used in conjunction.
The C++ structure actually comes out of a necessity. To use yaml mapping utlity in LLVM, we have to have a C++ object mapped to it, hence the structures. And since we need the C++ structures anyway, I wanted to make some of them to be re-usable across uArchs.
Not really, not in this context anyway. But we wanted to keep the base structs in a such a way that they can be used in a non-LLVM cases, but that's more of hope not necessasity.
I thought about this, but it takes me back to tablegen. Should we use tablegen? |
It feels like the overhead of this approach in its complexity and possible maintenance cost is not really justified. On its own it seems fine but these C++ bindings are not great. 😅
It's worth a try if what we need easily translates to tablegen entries and structure. That is, if you find yourself at any point fighting against tablegen or in need to introduce custom "hacks", I would not necessarily double down on it. I would suggest starting with user interfaces to flesh out what's really needed. Then it should become clearer where to store all the uarch details e.g., embed shared memory size directly in Next quick test would be to see how well it all holds up for another target like battlemage. |
Thanks, Adam. I think that's a good idea. Let me try to collect the uArch info JGS and battlemage. I looked into them during the design, but more details may help us finding the approach that would be easy to maintain. |
llvm#138091) Check this error for more context (https://github.com/compiler-research/CppInterOp/actions/runs/14749797085/job/41407625681?pr=491#step:10:531) This fails with ``` * thread #1, name = 'CppInterOpTests', stop reason = signal SIGSEGV: address not mapped to object (fault address: 0x55500356d6d3) * frame #0: 0x00007fffee41cfe3 libclangCppInterOp.so.21.0gitclang::PragmaNamespace::~PragmaNamespace() + 99 frame #1: 0x00007fffee435666 libclangCppInterOp.so.21.0gitclang::Preprocessor::~Preprocessor() + 3830 frame #2: 0x00007fffee20917a libclangCppInterOp.so.21.0gitstd::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release() + 58 frame llvm#3: 0x00007fffee224796 libclangCppInterOp.so.21.0gitclang::CompilerInstance::~CompilerInstance() + 838 frame llvm#4: 0x00007fffee22494d libclangCppInterOp.so.21.0gitclang::CompilerInstance::~CompilerInstance() + 13 frame llvm#5: 0x00007fffed95ec62 libclangCppInterOp.so.21.0gitclang::IncrementalCUDADeviceParser::~IncrementalCUDADeviceParser() + 98 frame llvm#6: 0x00007fffed9551b6 libclangCppInterOp.so.21.0gitclang::Interpreter::~Interpreter() + 102 frame llvm#7: 0x00007fffed95598d libclangCppInterOp.so.21.0gitclang::Interpreter::~Interpreter() + 13 frame llvm#8: 0x00007fffed9181e7 libclangCppInterOp.so.21.0gitcompat::createClangInterpreter(std::vector<char const*, std::allocator<char const*>>&) + 2919 ``` Problem : 1) The destructor currently handles no clearance for the DeviceParser and the DeviceAct. We currently only have this https://github.com/llvm/llvm-project/blob/976493822443c52a71ed3c67aaca9a555b20c55d/clang/lib/Interpreter/Interpreter.cpp#L416-L419 2) The ownership for DeviceCI currently is present in IncrementalCudaDeviceParser. But this should be similar to how the combination for hostCI, hostAction and hostParser are managed by the Interpreter. As on master the DeviceAct and DeviceParser are managed by the Interpreter but not DeviceCI. This is problematic because : IncrementalParser holds a Sema& which points into the DeviceCI. On master, DeviceCI is destroyed before the base class ~IncrementalParser() runs, causing Parser::reset() to access a dangling Sema (and as Sema holds a reference to Preprocessor which owns PragmaNamespace) we see this ``` * frame #0: 0x00007fffee41cfe3 libclangCppInterOp.so.21.0gitclang::PragmaNamespace::~PragmaNamespace() + 99 frame #1: 0x00007fffee435666 libclangCppInterOp.so.21.0gitclang::Preprocessor::~Preprocessor() + 3830 ```
Fix for: `Assertion failed: (false && "Architecture or OS not supported"), function CreateRegisterContextForFrame, file /usr/src/contrib/llvm-project/lldb/source/Plugins/Process/elf-core/ThreadElfCore.cpp, line 182. PLEASE submit a bug report to https://bugs.freebsd.org/submit/ and include the crash backtrace. #0 0x000000080cd857c8 llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) /usr/src/contrib/llvm-project/llvm/lib/Support/Unix/Signals.inc:723:13 #1 0x000000080cd85ed4 /usr/src/contrib/llvm-project/llvm/lib/Support/Unix/Signals.inc:797:3 #2 0x000000080cd82ae8 llvm::sys::RunSignalHandlers() /usr/src/contrib/llvm-project/llvm/lib/Support/Signals.cpp:104:5 llvm#3 0x000000080cd861f0 SignalHandler /usr/src/contrib/llvm-project/llvm/lib/Support/Unix/Signals.inc:403:3 llvm#4 0x000000080f159644 handle_signal /usr/src/lib/libthr/thread/thr_sig.c:298:3 `
The mcmodel=tiny memory model is only valid on ARM targets. While trying this on X86 compiler throws an internal error along with stack dump. llvm#125641 This patch resolves the issue. Reduced test case: ``` #include <stdio.h> int main( void ) { printf( "Hello, World!\n" ); return 0; } ``` ``` 0. Program arguments: /opt/compiler-explorer/clang-trunk/bin/clang++ -gdwarf-4 -g -o /app/output.s -fno-verbose-asm -S --gcc-toolchain=/opt/compiler-explorer/gcc-snapshot -fcolor-diagnostics -fno-crash-diagnostics -mcmodel=tiny <source> 1. <eof> parser at end of file #0 0x0000000003b10218 llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3b10218) #1 0x0000000003b0e35c llvm::sys::CleanupOnSignal(unsigned long) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3b0e35c) #2 0x0000000003a5dbc3 llvm::CrashRecoveryContext::HandleExit(int) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3a5dbc3) llvm#3 0x0000000003b05cfe llvm::sys::Process::Exit(int, bool) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3b05cfe) llvm#4 0x0000000000d4e3eb LLVMErrorHandler(void*, char const*, bool) cc1_main.cpp:0:0 llvm#5 0x0000000003a67c93 llvm::report_fatal_error(llvm::Twine const&, bool) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3a67c93) llvm#6 0x0000000003a67df8 (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3a67df8) llvm#7 0x0000000002549148 llvm::X86TargetMachine::X86TargetMachine(llvm::Target const&, llvm::Triple const&, llvm::StringRef, llvm::StringRef, llvm::TargetOptions const&, std::optional<llvm::Reloc::Model>, std::optional<llvm::CodeModel::Model>, llvm::CodeGenOptLevel, bool) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x2549148) llvm#8 0x00000000025491fc llvm::RegisterTargetMachine<llvm::X86TargetMachine>::Allocator(llvm::Target const&, llvm::Triple const&, llvm::StringRef, llvm::StringRef, llvm::TargetOptions const&, std::optional<llvm::Reloc::Model>, std::optional<llvm::CodeModel::Model>, llvm::CodeGenOptLevel, bool) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x25491fc) llvm#9 0x0000000003db74cc clang::emitBackendOutput(clang::CompilerInstance&, clang::CodeGenOptions&, llvm::StringRef, llvm::Module*, clang::BackendAction, llvm::IntrusiveRefCntPtr<llvm::vfs::FileSystem>, std::unique_ptr<llvm::raw_pwrite_stream, std::default_delete<llvm::raw_pwrite_stream>>, clang::BackendConsumer*) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3db74cc) llvm#10 0x0000000004460d95 clang::BackendConsumer::HandleTranslationUnit(clang::ASTContext&) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x4460d95) llvm#11 0x00000000060005ec clang::ParseAST(clang::Sema&, bool, bool) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x60005ec) llvm#12 0x00000000044614b5 clang::CodeGenAction::ExecuteAction() (/opt/compiler-explorer/clang-trunk/bin/clang+++0x44614b5) llvm#13 0x0000000004737121 clang::FrontendAction::Execute() (/opt/compiler-explorer/clang-trunk/bin/clang+++0x4737121) llvm#14 0x00000000046b777b clang::CompilerInstance::ExecuteAction(clang::FrontendAction&) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x46b777b) llvm#15 0x00000000048229e3 clang::ExecuteCompilerInvocation(clang::CompilerInstance*) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x48229e3) llvm#16 0x0000000000d50621 cc1_main(llvm::ArrayRef<char const*>, char const*, void*) (/opt/compiler-explorer/clang-trunk/bin/clang+++0xd50621) llvm#17 0x0000000000d48e2d ExecuteCC1Tool(llvm::SmallVectorImpl<char const*>&, llvm::ToolContext const&) driver.cpp:0:0 llvm#18 0x00000000044acc99 void llvm::function_ref<void ()>::callback_fn<clang::driver::CC1Command::Execute(llvm::ArrayRef<std::optional<llvm::StringRef>>, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char>>*, bool*) const::'lambda'()>(long) Job.cpp:0:0 llvm#19 0x0000000003a5dac3 llvm::CrashRecoveryContext::RunSafely(llvm::function_ref<void ()>) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x3a5dac3) llvm#20 0x00000000044aceb9 clang::driver::CC1Command::Execute(llvm::ArrayRef<std::optional<llvm::StringRef>>, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char>>*, bool*) const (.part.0) Job.cpp:0:0 llvm#21 0x00000000044710dd clang::driver::Compilation::ExecuteCommand(clang::driver::Command const&, clang::driver::Command const*&, bool) const (/opt/compiler-explorer/clang-trunk/bin/clang+++0x44710dd) llvm#22 0x0000000004472071 clang::driver::Compilation::ExecuteJobs(clang::driver::JobList const&, llvm::SmallVectorImpl<std::pair<int, clang::driver::Command const*>>&, bool) const (/opt/compiler-explorer/clang-trunk/bin/clang+++0x4472071) llvm#23 0x000000000447c3fc clang::driver::Driver::ExecuteCompilation(clang::driver::Compilation&, llvm::SmallVectorImpl<std::pair<int, clang::driver::Command const*>>&) (/opt/compiler-explorer/clang-trunk/bin/clang+++0x447c3fc) llvm#24 0x0000000000d4d2b1 clang_main(int, char**, llvm::ToolContext const&) (/opt/compiler-explorer/clang-trunk/bin/clang+++0xd4d2b1) llvm#25 0x0000000000c12464 main (/opt/compiler-explorer/clang-trunk/bin/clang+++0xc12464) llvm#26 0x00007ae43b029d90 (/lib/x86_64-linux-gnu/libc.so.6+0x29d90) llvm#27 0x00007ae43b029e40 __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x29e40) llvm#28 0x0000000000d488c5 _start (/opt/compiler-explorer/clang-trunk/bin/clang+++0xd488c5) ``` --------- Co-authored-by: Shashwathi N <nshashwa@pe31.hpc.amslabs.hpecorp.net>
…142952) This was removed in llvm#135343 in favour of making it a format variable, which we do here. This follows the precedent of the `[opt]` and `[artificial]` markers. Before: ``` thread #1, queue = 'com.apple.main-thread', stop reason = breakpoint 1.2 * frame #0: 0x000000010000037c a.out`inlined1() at inline.cpp:4:3 frame #1: 0x000000010000037c a.out`regular() at inline.cpp:6:17 frame #2: 0x00000001000003b8 a.out`inlined2() at inline.cpp:7:43 frame llvm#3: 0x00000001000003b4 a.out`main at inline.cpp:10:3 frame llvm#4: 0x0000000186345be4 dyld`start + 7040 ``` After (note the `[inlined]` markers): ``` thread #1, queue = 'com.apple.main-thread', stop reason = breakpoint 1.2 * frame #0: 0x000000010000037c a.out`inlined1() at inline.cpp:4:3 [inlined] frame #1: 0x000000010000037c a.out`regular() at inline.cpp:6:17 frame #2: 0x00000001000003b8 a.out`inlined2() at inline.cpp:7:43 [inlined] frame llvm#3: 0x00000001000003b4 a.out`main at inline.cpp:10:3 frame llvm#4: 0x0000000186345be4 dyld`start + 7040 ``` rdar://152642178
Hi All,
This is the initial PR for uArch definition. The primary purpose for this PR is to have a discussion and finalize the design. So please think of it as an RFC as well.
First target we want to achieve here is to finalize the skeleton:
Due to this purpose, the main focus point of this PR are the following files:
mlir/lib/Dialect/XeGPU/Utils/intel_gpu_pvc.yaml: holds the YAML definition of uArch (PVC is used for this use case)mlir/include/mlir/Dialect/XeGPU/Utils/uArch.h: holds the base uArch struct that can be re-used by specialized uArchs such as PVC.mlir/include/mlir/Dialect/XeGPU/Utils/IntelGpuPVC.h: holds the PVC-specific structs.mlir/lib/Dialect/XeGPU/Utils/IntelGpuPVC.cpp is added as placeholder for YAML mapping. This file is intentionally not complete. We want to reach a consensus on the YAML, and C++ structures before filling this one in.
Please provide your valuable feedback as we try to finalize this together.