mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2025-01-31 20:51:52 +01:00
f83e7bf498
Summary: Some constants can be handled with less instructions than our current results. And it seems our original approach is not very easy to extend. Therefore this patch proposes to materialize all 64-bit constants by enumerated patterns. I traversed almost all constants to verified the functionality of these pattens. A traversed comparison of the number of instructions used by the original method and the new method has also been completed, where no degradation was caused by this patch. This patch also passed Bootstrap test and SPEC test. Improvements of this patch are shown in llvm/test/CodeGen/PowerPC/constants-i64.ll Reviewed By: steven.zhang, stefanp Differential Revision: https://reviews.llvm.org/D92089
272 lines
8.9 KiB
LLVM
272 lines
8.9 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -verify-machineinstrs -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr \
|
|
; RUN: < %s -mtriple=powerpc64le-unknown-linux -mcpu=pwr9 | FileCheck %s
|
|
; RUN: llc -verify-machineinstrs -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr \
|
|
; RUN: < %s -mtriple=powerpc64le-unknown-linux -mcpu=pwr8 | FileCheck %s --check-prefix=CHECK-P8
|
|
|
|
declare fp128 @llvm.experimental.constrained.fadd.f128(fp128, fp128, metadata, metadata)
|
|
declare fp128 @llvm.experimental.constrained.fsub.f128(fp128, fp128, metadata, metadata)
|
|
declare fp128 @llvm.experimental.constrained.fmul.f128(fp128, fp128, metadata, metadata)
|
|
declare fp128 @llvm.experimental.constrained.fdiv.f128(fp128, fp128, metadata, metadata)
|
|
|
|
declare fp128 @llvm.experimental.constrained.fma.f128(fp128, fp128, fp128, metadata, metadata)
|
|
declare fp128 @llvm.experimental.constrained.sqrt.f128(fp128, metadata, metadata)
|
|
|
|
define fp128 @fadd_f128(fp128 %f1, fp128 %f2) #0 {
|
|
; CHECK-LABEL: fadd_f128:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: xsaddqp v2, v2, v3
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; CHECK-P8-LABEL: fadd_f128:
|
|
; CHECK-P8: # %bb.0:
|
|
; CHECK-P8-NEXT: mflr r0
|
|
; CHECK-P8-NEXT: std r0, 16(r1)
|
|
; CHECK-P8-NEXT: stdu r1, -32(r1)
|
|
; CHECK-P8-NEXT: .cfi_def_cfa_offset 32
|
|
; CHECK-P8-NEXT: .cfi_offset lr, 16
|
|
; CHECK-P8-NEXT: bl __addkf3
|
|
; CHECK-P8-NEXT: nop
|
|
; CHECK-P8-NEXT: addi r1, r1, 32
|
|
; CHECK-P8-NEXT: ld r0, 16(r1)
|
|
; CHECK-P8-NEXT: mtlr r0
|
|
; CHECK-P8-NEXT: blr
|
|
%res = call fp128 @llvm.experimental.constrained.fadd.f128(
|
|
fp128 %f1, fp128 %f2,
|
|
metadata !"round.dynamic",
|
|
metadata !"fpexcept.strict") #0
|
|
ret fp128 %res
|
|
}
|
|
|
|
define fp128 @fsub_f128(fp128 %f1, fp128 %f2) #0 {
|
|
; CHECK-LABEL: fsub_f128:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: xssubqp v2, v2, v3
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; CHECK-P8-LABEL: fsub_f128:
|
|
; CHECK-P8: # %bb.0:
|
|
; CHECK-P8-NEXT: mflr r0
|
|
; CHECK-P8-NEXT: std r0, 16(r1)
|
|
; CHECK-P8-NEXT: stdu r1, -32(r1)
|
|
; CHECK-P8-NEXT: .cfi_def_cfa_offset 32
|
|
; CHECK-P8-NEXT: .cfi_offset lr, 16
|
|
; CHECK-P8-NEXT: bl __subkf3
|
|
; CHECK-P8-NEXT: nop
|
|
; CHECK-P8-NEXT: addi r1, r1, 32
|
|
; CHECK-P8-NEXT: ld r0, 16(r1)
|
|
; CHECK-P8-NEXT: mtlr r0
|
|
; CHECK-P8-NEXT: blr
|
|
%res = call fp128 @llvm.experimental.constrained.fsub.f128(
|
|
fp128 %f1, fp128 %f2,
|
|
metadata !"round.dynamic",
|
|
metadata !"fpexcept.strict") #0
|
|
ret fp128 %res
|
|
}
|
|
|
|
define fp128 @fmul_f128(fp128 %f1, fp128 %f2) #0 {
|
|
; CHECK-LABEL: fmul_f128:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: xsmulqp v2, v2, v3
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; CHECK-P8-LABEL: fmul_f128:
|
|
; CHECK-P8: # %bb.0:
|
|
; CHECK-P8-NEXT: mflr r0
|
|
; CHECK-P8-NEXT: std r0, 16(r1)
|
|
; CHECK-P8-NEXT: stdu r1, -32(r1)
|
|
; CHECK-P8-NEXT: .cfi_def_cfa_offset 32
|
|
; CHECK-P8-NEXT: .cfi_offset lr, 16
|
|
; CHECK-P8-NEXT: bl __mulkf3
|
|
; CHECK-P8-NEXT: nop
|
|
; CHECK-P8-NEXT: addi r1, r1, 32
|
|
; CHECK-P8-NEXT: ld r0, 16(r1)
|
|
; CHECK-P8-NEXT: mtlr r0
|
|
; CHECK-P8-NEXT: blr
|
|
%res = call fp128 @llvm.experimental.constrained.fmul.f128(
|
|
fp128 %f1, fp128 %f2,
|
|
metadata !"round.dynamic",
|
|
metadata !"fpexcept.strict") #0
|
|
ret fp128 %res
|
|
}
|
|
|
|
define fp128 @fdiv_f128(fp128 %f1, fp128 %f2) #0 {
|
|
; CHECK-LABEL: fdiv_f128:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: xsdivqp v2, v2, v3
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; CHECK-P8-LABEL: fdiv_f128:
|
|
; CHECK-P8: # %bb.0:
|
|
; CHECK-P8-NEXT: mflr r0
|
|
; CHECK-P8-NEXT: std r0, 16(r1)
|
|
; CHECK-P8-NEXT: stdu r1, -32(r1)
|
|
; CHECK-P8-NEXT: .cfi_def_cfa_offset 32
|
|
; CHECK-P8-NEXT: .cfi_offset lr, 16
|
|
; CHECK-P8-NEXT: bl __divkf3
|
|
; CHECK-P8-NEXT: nop
|
|
; CHECK-P8-NEXT: addi r1, r1, 32
|
|
; CHECK-P8-NEXT: ld r0, 16(r1)
|
|
; CHECK-P8-NEXT: mtlr r0
|
|
; CHECK-P8-NEXT: blr
|
|
%res = call fp128 @llvm.experimental.constrained.fdiv.f128(
|
|
fp128 %f1, fp128 %f2,
|
|
metadata !"round.dynamic",
|
|
metadata !"fpexcept.strict") #0
|
|
ret fp128 %res
|
|
}
|
|
|
|
define fp128 @fmadd_f128(fp128 %f0, fp128 %f1, fp128 %f2) #0 {
|
|
; CHECK-LABEL: fmadd_f128:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: xsmaddqp v4, v2, v3
|
|
; CHECK-NEXT: vmr v2, v4
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; CHECK-P8-LABEL: fmadd_f128:
|
|
; CHECK-P8: # %bb.0:
|
|
; CHECK-P8-NEXT: mflr r0
|
|
; CHECK-P8-NEXT: std r0, 16(r1)
|
|
; CHECK-P8-NEXT: stdu r1, -32(r1)
|
|
; CHECK-P8-NEXT: .cfi_def_cfa_offset 32
|
|
; CHECK-P8-NEXT: .cfi_offset lr, 16
|
|
; CHECK-P8-NEXT: bl fmaf128
|
|
; CHECK-P8-NEXT: nop
|
|
; CHECK-P8-NEXT: addi r1, r1, 32
|
|
; CHECK-P8-NEXT: ld r0, 16(r1)
|
|
; CHECK-P8-NEXT: mtlr r0
|
|
; CHECK-P8-NEXT: blr
|
|
%res = call fp128 @llvm.experimental.constrained.fma.f128(
|
|
fp128 %f0, fp128 %f1, fp128 %f2,
|
|
metadata !"round.dynamic",
|
|
metadata !"fpexcept.strict") #0
|
|
ret fp128 %res
|
|
}
|
|
|
|
define fp128 @fmsub_f128(fp128 %f0, fp128 %f1, fp128 %f2) #0 {
|
|
; CHECK-LABEL: fmsub_f128:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: xsmsubqp v4, v2, v3
|
|
; CHECK-NEXT: vmr v2, v4
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; CHECK-P8-LABEL: fmsub_f128:
|
|
; CHECK-P8: # %bb.0:
|
|
; CHECK-P8-NEXT: mflr r0
|
|
; CHECK-P8-NEXT: std r0, 16(r1)
|
|
; CHECK-P8-NEXT: stdu r1, -32(r1)
|
|
; CHECK-P8-NEXT: .cfi_def_cfa_offset 32
|
|
; CHECK-P8-NEXT: .cfi_offset lr, 16
|
|
; CHECK-P8-NEXT: li r9, 1
|
|
; CHECK-P8-NEXT: rldic r9, r9, 63, 0
|
|
; CHECK-P8-NEXT: xor r8, r8, r9
|
|
; CHECK-P8-NEXT: bl fmaf128
|
|
; CHECK-P8-NEXT: nop
|
|
; CHECK-P8-NEXT: addi r1, r1, 32
|
|
; CHECK-P8-NEXT: ld r0, 16(r1)
|
|
; CHECK-P8-NEXT: mtlr r0
|
|
; CHECK-P8-NEXT: blr
|
|
%neg = fneg fp128 %f2
|
|
%res = call fp128 @llvm.experimental.constrained.fma.f128(
|
|
fp128 %f0, fp128 %f1, fp128 %neg,
|
|
metadata !"round.dynamic",
|
|
metadata !"fpexcept.strict") #0
|
|
ret fp128 %res
|
|
}
|
|
|
|
define fp128 @fnmadd_f128(fp128 %f0, fp128 %f1, fp128 %f2) #0 {
|
|
; CHECK-LABEL: fnmadd_f128:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: xsnmaddqp v4, v2, v3
|
|
; CHECK-NEXT: vmr v2, v4
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; CHECK-P8-LABEL: fnmadd_f128:
|
|
; CHECK-P8: # %bb.0:
|
|
; CHECK-P8-NEXT: mflr r0
|
|
; CHECK-P8-NEXT: std r0, 16(r1)
|
|
; CHECK-P8-NEXT: stdu r1, -32(r1)
|
|
; CHECK-P8-NEXT: .cfi_def_cfa_offset 32
|
|
; CHECK-P8-NEXT: .cfi_offset lr, 16
|
|
; CHECK-P8-NEXT: bl fmaf128
|
|
; CHECK-P8-NEXT: nop
|
|
; CHECK-P8-NEXT: li r5, 1
|
|
; CHECK-P8-NEXT: rldic r5, r5, 63, 0
|
|
; CHECK-P8-NEXT: xor r4, r4, r5
|
|
; CHECK-P8-NEXT: addi r1, r1, 32
|
|
; CHECK-P8-NEXT: ld r0, 16(r1)
|
|
; CHECK-P8-NEXT: mtlr r0
|
|
; CHECK-P8-NEXT: blr
|
|
%fma = call fp128 @llvm.experimental.constrained.fma.f128(
|
|
fp128 %f0, fp128 %f1, fp128 %f2,
|
|
metadata !"round.dynamic",
|
|
metadata !"fpexcept.strict") #0
|
|
%res = fneg fp128 %fma
|
|
ret fp128 %res
|
|
}
|
|
|
|
define fp128 @fnmsub_f128(fp128 %f0, fp128 %f1, fp128 %f2) #0 {
|
|
; CHECK-LABEL: fnmsub_f128:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: xsnmsubqp v4, v2, v3
|
|
; CHECK-NEXT: vmr v2, v4
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; CHECK-P8-LABEL: fnmsub_f128:
|
|
; CHECK-P8: # %bb.0:
|
|
; CHECK-P8-NEXT: mflr r0
|
|
; CHECK-P8-NEXT: .cfi_def_cfa_offset 48
|
|
; CHECK-P8-NEXT: .cfi_offset lr, 16
|
|
; CHECK-P8-NEXT: .cfi_offset r30, -16
|
|
; CHECK-P8-NEXT: std r30, -16(r1) # 8-byte Folded Spill
|
|
; CHECK-P8-NEXT: std r0, 16(r1)
|
|
; CHECK-P8-NEXT: stdu r1, -48(r1)
|
|
; CHECK-P8-NEXT: li r9, 1
|
|
; CHECK-P8-NEXT: rldic r30, r9, 63, 0
|
|
; CHECK-P8-NEXT: xor r8, r8, r30
|
|
; CHECK-P8-NEXT: bl fmaf128
|
|
; CHECK-P8-NEXT: nop
|
|
; CHECK-P8-NEXT: xor r4, r4, r30
|
|
; CHECK-P8-NEXT: addi r1, r1, 48
|
|
; CHECK-P8-NEXT: ld r0, 16(r1)
|
|
; CHECK-P8-NEXT: ld r30, -16(r1) # 8-byte Folded Reload
|
|
; CHECK-P8-NEXT: mtlr r0
|
|
; CHECK-P8-NEXT: blr
|
|
%neg = fneg fp128 %f2
|
|
%fma = call fp128 @llvm.experimental.constrained.fma.f128(
|
|
fp128 %f0, fp128 %f1, fp128 %neg,
|
|
metadata !"round.dynamic",
|
|
metadata !"fpexcept.strict") #0
|
|
%res = fneg fp128 %fma
|
|
ret fp128 %res
|
|
}
|
|
|
|
|
|
define fp128 @fsqrt_f128(fp128 %f1) #0 {
|
|
; CHECK-LABEL: fsqrt_f128:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: xssqrtqp v2, v2
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; CHECK-P8-LABEL: fsqrt_f128:
|
|
; CHECK-P8: # %bb.0:
|
|
; CHECK-P8-NEXT: mflr r0
|
|
; CHECK-P8-NEXT: std r0, 16(r1)
|
|
; CHECK-P8-NEXT: stdu r1, -32(r1)
|
|
; CHECK-P8-NEXT: .cfi_def_cfa_offset 32
|
|
; CHECK-P8-NEXT: .cfi_offset lr, 16
|
|
; CHECK-P8-NEXT: bl sqrtf128
|
|
; CHECK-P8-NEXT: nop
|
|
; CHECK-P8-NEXT: addi r1, r1, 32
|
|
; CHECK-P8-NEXT: ld r0, 16(r1)
|
|
; CHECK-P8-NEXT: mtlr r0
|
|
; CHECK-P8-NEXT: blr
|
|
%res = call fp128 @llvm.experimental.constrained.sqrt.f128(
|
|
fp128 %f1,
|
|
metadata !"round.dynamic",
|
|
metadata !"fpexcept.strict") #0
|
|
ret fp128 %res
|
|
}
|
|
|
|
attributes #0 = { strictfp }
|