search

llvm / llvm-project

The LLVM Project is a collection of modular and reusable compiler and toolchain technologies.
http://llvm.org
Other
27.32k stars 11.21k forks source link

LLVM MCA does not match uops.info/Agner measurements #63877

Closed SeeSpring closed 1 year ago

SeeSpring commented 1 year ago

According to uops.info and Agner these instructions all have a throughput of 0.50 on Zen 4

vminps        %ymm0, %ymm1, %ymm2
vfmadd213ps   %ymm0, %ymm1, %ymm2
vcmpltps      %ymm0, %ymm1, %ymm2

llvm-mca claims they have a RThroughput of 1.00 Godbolt of llvm-mca,l:'5',n:'0',o:'Analysis+source+%231',t:'0')),k:37.5,l:'4',n:'0',o:'',s:0,t:'0'),(g:!((h:compiler,i:(compiler:llvm-mcatrunk,deviceViewOpen:'1',filters:(b:'0',binary:'1',binaryObject:'1',commentOnly:'0',debugCalls:'1',demangle:'0',directives:'0',execute:'1',intel:'1',libraryCode:'0',trim:'1'),flagsViewOpen:'1',fontScale:14,fontUsePx:'0',j:2,lang:analysis,libs:!(),options:'--mcpu%3Dznver4',overrides:!(),selection:(endColumn:1,endLineNumber:1,positionColumn:1,positionLineNumber:1,selectionStartColumn:1,selectionStartLineNumber:1,startColumn:1,startLineNumber:1),source:1),l:'5',n:'0',o:'+llvm-mca+(trunk)+(Editor+%231)',t:'0')),k:62.5,l:'4',m:100,n:'0',o:'',s:1,t:'0')),l:'2',m:100,n:'0',o:'',t:'0')),version:4)

Version ``` LLVM (http://llvm.org/): LLVM version 17.0.0git Optimized build. Default target: x86_64-unknown-linux-gnu Host CPU: skylake-avx512 Registered Targets: aarch64 - AArch64 (little endian) aarch64_32 - AArch64 (little endian ILP32) aarch64_be - AArch64 (big endian) amdgcn - AMD GCN GPUs arm - ARM arm64 - ARM64 (little endian) arm64_32 - ARM64 (little endian ILP32) armeb - ARM (big endian) avr - Atmel AVR Microcontroller bpf - BPF (host endian) bpfeb - BPF (big endian) bpfel - BPF (little endian) hexagon - Hexagon lanai - Lanai loongarch32 - 32-bit LoongArch loongarch64 - 64-bit LoongArch mips - MIPS (32-bit big endian) mips64 - MIPS (64-bit big endian) mips64el - MIPS (64-bit little endian) mipsel - MIPS (32-bit little endian) msp430 - MSP430 [experimental] nvptx - NVIDIA PTX 32-bit nvptx64 - NVIDIA PTX 64-bit ppc32 - PowerPC 32 ppc32le - PowerPC 32 LE ppc64 - PowerPC 64 ppc64le - PowerPC 64 LE r600 - AMD GPUs HD2XXX-HD6XXX riscv32 - 32-bit RISC-V riscv64 - 64-bit RISC-V sparc - Sparc sparcel - Sparc LE sparcv9 - Sparc V9 systemz - SystemZ thumb - Thumb thumbeb - Thumb (big endian) ve - VE wasm32 - WebAssembly 32-bit wasm64 - WebAssembly 64-bit x86 - 32-bit X86: Pentium-Pro and above x86-64 - 64-bit X86: EM64T and AMD64 xcore - XCore Compiler returned: 0 ```
llvmbot commented 1 year ago

@llvm/issue-subscribers-backend-x86