1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2025-02-01 05:01:59 +01:00
llvm-mirror/test/CodeGen/X86/bmi2-x86_64.ll
Craig Topper 43dcc4cc5f [X86] Don't emit MULX by default with BMI2
MULX has somewhat improved register allocation constraints compared to the legacy MUL instruction. Both output registers are encoded instead of fixed to EAX/EDX, but EDX is used as input. It also doesn't touch flags. Unfortunately, the encoding is longer.

Prefering it whenever BMI2 is enabled is probably not optimal. Choosing it should somehow be a function of register allocation constraints like converting adds to three address. gcc and icc definitely don't pick MULX by default. Not sure what if any rules they have for using it.

Differential Revision: https://reviews.llvm.org/D55565

llvm-svn: 348975
2018-12-12 21:21:31 +00:00

103 lines
2.7 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+bmi,+bmi2 | FileCheck %s --check-prefixes=CHECK
define i64 @bzhi64(i64 %x, i64 %y) {
; CHECK-LABEL: bzhi64:
; CHECK: # %bb.0:
; CHECK-NEXT: bzhiq %rsi, %rdi, %rax
; CHECK-NEXT: retq
%tmp = tail call i64 @llvm.x86.bmi.bzhi.64(i64 %x, i64 %y)
ret i64 %tmp
}
define i64 @bzhi64_load(i64* %x, i64 %y) {
; CHECK-LABEL: bzhi64_load:
; CHECK: # %bb.0:
; CHECK-NEXT: bzhiq %rsi, (%rdi), %rax
; CHECK-NEXT: retq
%x1 = load i64, i64* %x
%tmp = tail call i64 @llvm.x86.bmi.bzhi.64(i64 %x1, i64 %y)
ret i64 %tmp
}
declare i64 @llvm.x86.bmi.bzhi.64(i64, i64)
define i64 @pdep64(i64 %x, i64 %y) {
; CHECK-LABEL: pdep64:
; CHECK: # %bb.0:
; CHECK-NEXT: pdepq %rsi, %rdi, %rax
; CHECK-NEXT: retq
%tmp = tail call i64 @llvm.x86.bmi.pdep.64(i64 %x, i64 %y)
ret i64 %tmp
}
define i64 @pdep64_load(i64 %x, i64* %y) {
; CHECK-LABEL: pdep64_load:
; CHECK: # %bb.0:
; CHECK-NEXT: pdepq (%rsi), %rdi, %rax
; CHECK-NEXT: retq
%y1 = load i64, i64* %y
%tmp = tail call i64 @llvm.x86.bmi.pdep.64(i64 %x, i64 %y1)
ret i64 %tmp
}
declare i64 @llvm.x86.bmi.pdep.64(i64, i64)
define i64 @pext64(i64 %x, i64 %y) {
; CHECK-LABEL: pext64:
; CHECK: # %bb.0:
; CHECK-NEXT: pextq %rsi, %rdi, %rax
; CHECK-NEXT: retq
%tmp = tail call i64 @llvm.x86.bmi.pext.64(i64 %x, i64 %y)
ret i64 %tmp
}
define i64 @pext64_load(i64 %x, i64* %y) {
; CHECK-LABEL: pext64_load:
; CHECK: # %bb.0:
; CHECK-NEXT: pextq (%rsi), %rdi, %rax
; CHECK-NEXT: retq
%y1 = load i64, i64* %y
%tmp = tail call i64 @llvm.x86.bmi.pext.64(i64 %x, i64 %y1)
ret i64 %tmp
}
declare i64 @llvm.x86.bmi.pext.64(i64, i64)
define i64 @mulx64(i64 %x, i64 %y, i64* %p) {
; CHECK-LABEL: mulx64:
; CHECK: # %bb.0:
; CHECK-NEXT: movq %rdx, %rcx
; CHECK-NEXT: movq %rdi, %rax
; CHECK-NEXT: mulq %rsi
; CHECK-NEXT: movq %rdx, (%rcx)
; CHECK-NEXT: retq
%x1 = zext i64 %x to i128
%y1 = zext i64 %y to i128
%r1 = mul i128 %x1, %y1
%h1 = lshr i128 %r1, 64
%h = trunc i128 %h1 to i64
%l = trunc i128 %r1 to i64
store i64 %h, i64* %p
ret i64 %l
}
define i64 @mulx64_load(i64 %x, i64* %y, i64* %p) {
; CHECK-LABEL: mulx64_load:
; CHECK: # %bb.0:
; CHECK-NEXT: movq %rdx, %rcx
; CHECK-NEXT: movq %rdi, %rax
; CHECK-NEXT: mulq (%rsi)
; CHECK-NEXT: movq %rdx, (%rcx)
; CHECK-NEXT: retq
%y1 = load i64, i64* %y
%x2 = zext i64 %x to i128
%y2 = zext i64 %y1 to i128
%r1 = mul i128 %x2, %y2
%h1 = lshr i128 %r1, 64
%h = trunc i128 %h1 to i64
%l = trunc i128 %r1 to i64
store i64 %h, i64* %p
ret i64 %l
}