mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-23 03:02:36 +01:00
b7b3e8e261
Record internal state based on register units. This is often more efficient as there are typically fewer register units to update compared to iterating over all the aliases of a register. Original patch by Matthias Braun, but I've been rebasing and fixing it for almost 2 years and fixed a few bugs causing intermediate failures to make this patch independent of the changes in https://reviews.llvm.org/D52010.
59 lines
1.9 KiB
LLVM
59 lines
1.9 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -O0 < %s | FileCheck %s
|
|
target datalayout = "e-m:e-i64:64-n32:64"
|
|
target triple = "powerpc64le-unknown-linux-gnu"
|
|
|
|
define void @bn_mul_comba8(i64* nocapture %r, i64* nocapture readonly %a, i64* nocapture readonly %b) {
|
|
; CHECK-LABEL: bn_mul_comba8:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: ld 6, 0(4)
|
|
; CHECK-NEXT: ld 7, 0(5)
|
|
; CHECK-NEXT: mulhdu 8, 7, 6
|
|
; CHECK-NEXT: ld 4, 8(4)
|
|
; CHECK-NEXT: mulld 9, 4, 6
|
|
; CHECK-NEXT: mulhdu 4, 4, 6
|
|
; CHECK-NEXT: addc 6, 9, 8
|
|
; CHECK-NEXT: addze 4, 4
|
|
; CHECK-NEXT: ld 5, 8(5)
|
|
; CHECK-NEXT: mulld 8, 5, 7
|
|
; CHECK-NEXT: mulhdu 5, 5, 7
|
|
; CHECK-NEXT: addc 6, 6, 8
|
|
; CHECK-NEXT: addze 5, 5
|
|
; CHECK-NEXT: add 4, 5, 4
|
|
; CHECK-NEXT: cmpld 7, 4, 5
|
|
; CHECK-NEXT: mfocrf 4, 1
|
|
; CHECK-NEXT: rlwinm 4, 4, 29, 31, 31
|
|
; CHECK-NEXT: # implicit-def: $x5
|
|
; CHECK-NEXT: mr 5, 4
|
|
; CHECK-NEXT: clrldi 4, 5, 32
|
|
; CHECK-NEXT: std 4, 0(3)
|
|
; CHECK-NEXT: blr
|
|
%1 = load i64, i64* %a, align 8
|
|
%conv = zext i64 %1 to i128
|
|
%2 = load i64, i64* %b, align 8
|
|
%conv2 = zext i64 %2 to i128
|
|
%mul = mul nuw i128 %conv2, %conv
|
|
%shr = lshr i128 %mul, 64
|
|
%agep = getelementptr inbounds i64, i64* %a, i64 1
|
|
%3 = load i64, i64* %agep, align 8
|
|
%conv14 = zext i64 %3 to i128
|
|
%mul15 = mul nuw i128 %conv14, %conv
|
|
%add17 = add i128 %mul15, %shr
|
|
%shr19 = lshr i128 %add17, 64
|
|
%conv20 = trunc i128 %shr19 to i64
|
|
%bgep = getelementptr inbounds i64, i64* %b, i64 1
|
|
%4 = load i64, i64* %bgep, align 8
|
|
%conv28 = zext i64 %4 to i128
|
|
%mul31 = mul nuw i128 %conv28, %conv2
|
|
%conv32 = and i128 %add17, 18446744073709551615
|
|
%add33 = add i128 %conv32, %mul31
|
|
%shr35 = lshr i128 %add33, 64
|
|
%conv36 = trunc i128 %shr35 to i64
|
|
%add37 = add i64 %conv36, %conv20
|
|
%cmp38 = icmp ult i64 %add37, %conv36
|
|
%conv148 = zext i1 %cmp38 to i64
|
|
store i64 %conv148, i64* %r, align 8
|
|
ret void
|
|
}
|
|
|