mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-22 18:54:02 +01:00
5d7ef4b589
The scalarization overhead was set deliberately high for MVE, whilst the codegen was new. It helps protect us against the negative ramifications of mixing scalar and vector instructions. This decreases that, especially for floating point where the cost of extracting/inserting lane elements can be low. For integer the cost is still fairly high due to the cross-register-bank copy, but is no longer n^2 in the length of the vector. In general, this will decrease the cost of scalarizing floats and long integer vectors. i64 increase in cost, having a high cost before and after this patch. For floats this allows up to start doing things like vectorizing fdiv instructions, even if they are scalarized. Differential Revision: https://reviews.llvm.org/D98245
173 lines
11 KiB
LLVM
173 lines
11 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
|
|
; RUN: opt -cost-model -analyze -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve < %s | FileCheck %s --check-prefix=CHECK-MVE
|
|
; RUN: opt -cost-model -analyze -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve.fp < %s | FileCheck %s --check-prefix=CHECK-MVEFP
|
|
|
|
define void @f32() {
|
|
; CHECK-MVE-LABEL: 'f32'
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %c = fadd float undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %d = fsub float undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %e = fmul float undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
|
;
|
|
; CHECK-MVEFP-LABEL: 'f32'
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %c = fadd float undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %d = fsub float undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %e = fmul float undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
|
;
|
|
%c = fadd float undef, undef
|
|
%d = fsub float undef, undef
|
|
%e = fmul float undef, undef
|
|
ret void
|
|
}
|
|
|
|
define void @f16() {
|
|
; CHECK-MVE-LABEL: 'f16'
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %c = fadd half undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %d = fsub half undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %e = fmul half undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
|
;
|
|
; CHECK-MVEFP-LABEL: 'f16'
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %c = fadd half undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %d = fsub half undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %e = fmul half undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
|
;
|
|
%c = fadd half undef, undef
|
|
%d = fsub half undef, undef
|
|
%e = fmul half undef, undef
|
|
ret void
|
|
}
|
|
|
|
define void @f64() {
|
|
; CHECK-MVE-LABEL: 'f64'
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %c = fadd double undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %d = fsub double undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %e = fmul double undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
|
;
|
|
; CHECK-MVEFP-LABEL: 'f64'
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %c = fadd double undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %d = fsub double undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %e = fmul double undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
|
;
|
|
%c = fadd double undef, undef
|
|
%d = fsub double undef, undef
|
|
%e = fmul double undef, undef
|
|
ret void
|
|
}
|
|
|
|
define void @vf32() {
|
|
; CHECK-MVE-LABEL: 'vf32'
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %c2 = fadd <2 x float> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %d2 = fsub <2 x float> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %e2 = fmul <2 x float> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %c4 = fadd <4 x float> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %d4 = fsub <4 x float> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %e4 = fmul <4 x float> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %c8 = fadd <8 x float> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %d8 = fsub <8 x float> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %e8 = fmul <8 x float> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
|
;
|
|
; CHECK-MVEFP-LABEL: 'vf32'
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %c2 = fadd <2 x float> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %d2 = fsub <2 x float> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %e2 = fmul <2 x float> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %c4 = fadd <4 x float> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %d4 = fsub <4 x float> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %e4 = fmul <4 x float> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %c8 = fadd <8 x float> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %d8 = fsub <8 x float> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %e8 = fmul <8 x float> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
|
;
|
|
%c2 = fadd <2 x float> undef, undef
|
|
%d2 = fsub <2 x float> undef, undef
|
|
%e2 = fmul <2 x float> undef, undef
|
|
%c4 = fadd <4 x float> undef, undef
|
|
%d4 = fsub <4 x float> undef, undef
|
|
%e4 = fmul <4 x float> undef, undef
|
|
%c8 = fadd <8 x float> undef, undef
|
|
%d8 = fsub <8 x float> undef, undef
|
|
%e8 = fmul <8 x float> undef, undef
|
|
ret void
|
|
}
|
|
|
|
define void @vf16() {
|
|
; CHECK-MVE-LABEL: 'vf16'
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %c2 = fadd <2 x half> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %d2 = fsub <2 x half> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %e2 = fmul <2 x half> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %c4 = fadd <4 x half> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %d4 = fsub <4 x half> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %e4 = fmul <4 x half> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %c8 = fadd <8 x half> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %d8 = fsub <8 x half> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %e8 = fmul <8 x half> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
|
;
|
|
; CHECK-MVEFP-LABEL: 'vf16'
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %c2 = fadd <2 x half> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %d2 = fsub <2 x half> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %e2 = fmul <2 x half> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %c4 = fadd <4 x half> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %d4 = fsub <4 x half> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %e4 = fmul <4 x half> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %c8 = fadd <8 x half> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %d8 = fsub <8 x half> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %e8 = fmul <8 x half> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
|
;
|
|
%c2 = fadd <2 x half> undef, undef
|
|
%d2 = fsub <2 x half> undef, undef
|
|
%e2 = fmul <2 x half> undef, undef
|
|
%c4 = fadd <4 x half> undef, undef
|
|
%d4 = fsub <4 x half> undef, undef
|
|
%e4 = fmul <4 x half> undef, undef
|
|
%c8 = fadd <8 x half> undef, undef
|
|
%d8 = fsub <8 x half> undef, undef
|
|
%e8 = fmul <8 x half> undef, undef
|
|
ret void
|
|
}
|
|
|
|
define void @vf64() {
|
|
; CHECK-MVE-LABEL: 'vf64'
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %c2 = fadd <2 x double> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %d2 = fsub <2 x double> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %e2 = fmul <2 x double> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %c4 = fadd <4 x double> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %d4 = fsub <4 x double> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %e4 = fmul <4 x double> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %c8 = fadd <8 x double> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %d8 = fsub <8 x double> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %e8 = fmul <8 x double> undef, undef
|
|
; CHECK-MVE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
|
;
|
|
; CHECK-MVEFP-LABEL: 'vf64'
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %c2 = fadd <2 x double> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %d2 = fsub <2 x double> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %e2 = fmul <2 x double> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %c4 = fadd <4 x double> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %d4 = fsub <4 x double> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %e4 = fmul <4 x double> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %c8 = fadd <8 x double> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %d8 = fsub <8 x double> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %e8 = fmul <8 x double> undef, undef
|
|
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
|
;
|
|
%c2 = fadd <2 x double> undef, undef
|
|
%d2 = fsub <2 x double> undef, undef
|
|
%e2 = fmul <2 x double> undef, undef
|
|
%c4 = fadd <4 x double> undef, undef
|
|
%d4 = fsub <4 x double> undef, undef
|
|
%e4 = fmul <4 x double> undef, undef
|
|
%c8 = fadd <8 x double> undef, undef
|
|
%d8 = fsub <8 x double> undef, undef
|
|
%e8 = fmul <8 x double> undef, undef
|
|
ret void
|
|
}
|
|
|