RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-22 20:43:44 +02:00
Code Issues Projects Releases Wiki Activity
442bdb683b
llvm-mirror/test/CodeGen/X86/avx512ifmavl-intrinsics.ll
Gadi Haber 51f2170fda This is a large patch for X86 AVX-512 of an optimization for reducing code size by encoding EVEX AVX-512 instructions using the shorter VEX encoding when possible. 
There are cases of AVX-512 instructions that have two possible encodings. This is the case with instructions that use vector registers with low indexes of 0 - 15 and do not use the zmm registers or the mask k registers.
The EVEX encoding prefix requires 4 bytes whereas the VEX prefix can take only up to 3 bytes. Consequently, using the VEX encoding for these instructions results in a code size reduction of ~2 bytes even though it is compiled with the AVX-512 features enabled.

Reviewers: Craig Topper, Zvi Rackoover, Elena Demikhovsky 
Differential Revision: https://reviews.llvm.org/D27901

llvm-svn: 290663
2016-12-28 10:12:48 +00:00

227 lines
12 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=knl -mattr=+avx512vl -mattr=+avx512ifma | FileCheck %s
declare <2 x i64> @llvm.x86.avx512.mask.vpmadd52h.uq.128(<2 x i64>, <2 x i64>, <2 x i64>, i8)
define <2 x i64>@test_int_x86_avx512_mask_vpmadd52h_uq_128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_mask_vpmadd52h_uq_128:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vmovaps %xmm0, %xmm3
; CHECK-NEXT: vpmadd52huq %xmm2, %xmm1, %xmm3 {%k1}
; CHECK-NEXT: vmovaps %xmm0, %xmm4
; CHECK-NEXT: vpmadd52huq %xmm2, %xmm1, %xmm4
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpmadd52huq %xmm2, %xmm1, %xmm0 {%k1}
; CHECK-NEXT: vpmadd52huq %xmm2, %xmm1, %xmm2 {%k1} {z}
; CHECK-NEXT: vpaddq %xmm0, %xmm3, %xmm0
; CHECK-NEXT: vpaddq %xmm2, %xmm4, %xmm1
; CHECK-NEXT: vpaddq %xmm0, %xmm1, %xmm0
; CHECK-NEXT: retq
%res = call <2 x i64> @llvm.x86.avx512.mask.vpmadd52h.uq.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3)
%res1 = call <2 x i64> @llvm.x86.avx512.mask.vpmadd52h.uq.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> zeroinitializer, i8 %x3)
%res2 = call <2 x i64> @llvm.x86.avx512.mask.vpmadd52h.uq.128(<2 x i64> zeroinitializer, <2 x i64> %x1, <2 x i64> zeroinitializer, i8 %x3)
%res3 = call <2 x i64> @llvm.x86.avx512.mask.vpmadd52h.uq.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 -1)
%res4 = add <2 x i64> %res, %res1
%res5 = add <2 x i64> %res3, %res2
%res6 = add <2 x i64> %res5, %res4
ret <2 x i64> %res6
}
declare <4 x i64> @llvm.x86.avx512.mask.vpmadd52h.uq.256(<4 x i64>, <4 x i64>, <4 x i64>, i8)
define <4 x i64>@test_int_x86_avx512_mask_vpmadd52h_uq_256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_mask_vpmadd52h_uq_256:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vmovaps %ymm0, %ymm3
; CHECK-NEXT: vpmadd52huq %ymm2, %ymm1, %ymm3 {%k1}
; CHECK-NEXT: vmovaps %ymm0, %ymm4
; CHECK-NEXT: vpmadd52huq %ymm2, %ymm1, %ymm4
; CHECK-NEXT: vpxor %ymm2, %ymm2, %ymm2
; CHECK-NEXT: vpmadd52huq %ymm2, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: vpmadd52huq %ymm2, %ymm1, %ymm2 {%k1} {z}
; CHECK-NEXT: vpaddq %ymm0, %ymm3, %ymm0
; CHECK-NEXT: vpaddq %ymm2, %ymm4, %ymm1
; CHECK-NEXT: vpaddq %ymm0, %ymm1, %ymm0
; CHECK-NEXT: retq
%res = call <4 x i64> @llvm.x86.avx512.mask.vpmadd52h.uq.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3)
%res1 = call <4 x i64> @llvm.x86.avx512.mask.vpmadd52h.uq.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> zeroinitializer, i8 %x3)
%res2 = call <4 x i64> @llvm.x86.avx512.mask.vpmadd52h.uq.256(<4 x i64> zeroinitializer, <4 x i64> %x1, <4 x i64> zeroinitializer, i8 %x3)
%res3 = call <4 x i64> @llvm.x86.avx512.mask.vpmadd52h.uq.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 -1)
%res4 = add <4 x i64> %res, %res1
%res5 = add <4 x i64> %res3, %res2
%res6 = add <4 x i64> %res5, %res4
ret <4 x i64> %res6
}
declare <2 x i64> @llvm.x86.avx512.maskz.vpmadd52h.uq.128(<2 x i64>, <2 x i64>, <2 x i64>, i8)
define <2 x i64>@test_int_x86_avx512_maskz_vpmadd52h_uq_128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_maskz_vpmadd52h_uq_128:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vmovaps %xmm0, %xmm3
; CHECK-NEXT: vpmadd52huq %xmm2, %xmm1, %xmm3 {%k1} {z}
; CHECK-NEXT: vmovaps %xmm0, %xmm4
; CHECK-NEXT: vpmadd52huq %xmm2, %xmm1, %xmm4
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpmadd52huq %xmm2, %xmm1, %xmm0 {%k1} {z}
; CHECK-NEXT: vpmadd52huq %xmm2, %xmm1, %xmm2 {%k1} {z}
; CHECK-NEXT: vpaddq %xmm0, %xmm3, %xmm0
; CHECK-NEXT: vpaddq %xmm2, %xmm4, %xmm1
; CHECK-NEXT: vpaddq %xmm0, %xmm1, %xmm0
; CHECK-NEXT: retq
%res = call <2 x i64> @llvm.x86.avx512.maskz.vpmadd52h.uq.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3)
%res1 = call <2 x i64> @llvm.x86.avx512.maskz.vpmadd52h.uq.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> zeroinitializer, i8 %x3)
%res2 = call <2 x i64> @llvm.x86.avx512.maskz.vpmadd52h.uq.128(<2 x i64> zeroinitializer, <2 x i64> %x1, <2 x i64> zeroinitializer, i8 %x3)
%res3 = call <2 x i64> @llvm.x86.avx512.maskz.vpmadd52h.uq.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 -1)
%res4 = add <2 x i64> %res, %res1
%res5 = add <2 x i64> %res3, %res2
%res6 = add <2 x i64> %res5, %res4
ret <2 x i64> %res6
}
declare <4 x i64> @llvm.x86.avx512.maskz.vpmadd52h.uq.256(<4 x i64>, <4 x i64>, <4 x i64>, i8)
define <4 x i64>@test_int_x86_avx512_maskz_vpmadd52h_uq_256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_maskz_vpmadd52h_uq_256:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vmovaps %ymm0, %ymm3
; CHECK-NEXT: vpmadd52huq %ymm2, %ymm1, %ymm3 {%k1} {z}
; CHECK-NEXT: vmovaps %ymm0, %ymm4
; CHECK-NEXT: vpmadd52huq %ymm2, %ymm1, %ymm4
; CHECK-NEXT: vpxor %ymm2, %ymm2, %ymm2
; CHECK-NEXT: vpmadd52huq %ymm2, %ymm1, %ymm0 {%k1} {z}
; CHECK-NEXT: vpmadd52huq %ymm2, %ymm1, %ymm2 {%k1} {z}
; CHECK-NEXT: vpaddq %ymm0, %ymm3, %ymm0
; CHECK-NEXT: vpaddq %ymm2, %ymm4, %ymm1
; CHECK-NEXT: vpaddq %ymm0, %ymm1, %ymm0
; CHECK-NEXT: retq
%res = call <4 x i64> @llvm.x86.avx512.maskz.vpmadd52h.uq.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3)
%res1 = call <4 x i64> @llvm.x86.avx512.maskz.vpmadd52h.uq.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> zeroinitializer, i8 %x3)
%res2 = call <4 x i64> @llvm.x86.avx512.maskz.vpmadd52h.uq.256(<4 x i64> zeroinitializer, <4 x i64> %x1, <4 x i64> zeroinitializer, i8 %x3)
%res3 = call <4 x i64> @llvm.x86.avx512.maskz.vpmadd52h.uq.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 -1)
%res4 = add <4 x i64> %res, %res1
%res5 = add <4 x i64> %res3, %res2
%res6 = add <4 x i64> %res5, %res4
ret <4 x i64> %res6
}
declare <2 x i64> @llvm.x86.avx512.mask.vpmadd52l.uq.128(<2 x i64>, <2 x i64>, <2 x i64>, i8)
define <2 x i64>@test_int_x86_avx512_mask_vpmadd52l_uq_128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_mask_vpmadd52l_uq_128:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vmovaps %xmm0, %xmm3
; CHECK-NEXT: vpmadd52luq %xmm2, %xmm1, %xmm3 {%k1}
; CHECK-NEXT: vmovaps %xmm0, %xmm4
; CHECK-NEXT: vpmadd52luq %xmm2, %xmm1, %xmm4
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpmadd52luq %xmm2, %xmm1, %xmm0 {%k1}
; CHECK-NEXT: vpmadd52luq %xmm2, %xmm1, %xmm2 {%k1} {z}
; CHECK-NEXT: vpaddq %xmm0, %xmm3, %xmm0
; CHECK-NEXT: vpaddq %xmm2, %xmm4, %xmm1
; CHECK-NEXT: vpaddq %xmm0, %xmm1, %xmm0
; CHECK-NEXT: retq
%res = call <2 x i64> @llvm.x86.avx512.mask.vpmadd52l.uq.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3)
%res1 = call <2 x i64> @llvm.x86.avx512.mask.vpmadd52l.uq.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> zeroinitializer, i8 %x3)
%res2 = call <2 x i64> @llvm.x86.avx512.mask.vpmadd52l.uq.128(<2 x i64> zeroinitializer, <2 x i64> %x1, <2 x i64> zeroinitializer, i8 %x3)
%res3 = call <2 x i64> @llvm.x86.avx512.mask.vpmadd52l.uq.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 -1)
%res4 = add <2 x i64> %res, %res1
%res5 = add <2 x i64> %res3, %res2
%res6 = add <2 x i64> %res5, %res4
ret <2 x i64> %res6
}
declare <4 x i64> @llvm.x86.avx512.mask.vpmadd52l.uq.256(<4 x i64>, <4 x i64>, <4 x i64>, i8)
define <4 x i64>@test_int_x86_avx512_mask_vpmadd52l_uq_256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_mask_vpmadd52l_uq_256:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vmovaps %ymm0, %ymm3
; CHECK-NEXT: vpmadd52luq %ymm2, %ymm1, %ymm3 {%k1}
; CHECK-NEXT: vmovaps %ymm0, %ymm4
; CHECK-NEXT: vpmadd52luq %ymm2, %ymm1, %ymm4
; CHECK-NEXT: vpxor %ymm2, %ymm2, %ymm2
; CHECK-NEXT: vpmadd52luq %ymm2, %ymm1, %ymm0 {%k1}
; CHECK-NEXT: vpmadd52luq %ymm2, %ymm1, %ymm2 {%k1} {z}
; CHECK-NEXT: vpaddq %ymm0, %ymm3, %ymm0
; CHECK-NEXT: vpaddq %ymm2, %ymm4, %ymm1
; CHECK-NEXT: vpaddq %ymm0, %ymm1, %ymm0
; CHECK-NEXT: retq
%res = call <4 x i64> @llvm.x86.avx512.mask.vpmadd52l.uq.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3)
%res1 = call <4 x i64> @llvm.x86.avx512.mask.vpmadd52l.uq.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> zeroinitializer, i8 %x3)
%res2 = call <4 x i64> @llvm.x86.avx512.mask.vpmadd52l.uq.256(<4 x i64> zeroinitializer, <4 x i64> %x1, <4 x i64> zeroinitializer, i8 %x3)
%res3 = call <4 x i64> @llvm.x86.avx512.mask.vpmadd52l.uq.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 -1)
%res4 = add <4 x i64> %res, %res1
%res5 = add <4 x i64> %res3, %res2
%res6 = add <4 x i64> %res5, %res4
ret <4 x i64> %res6
}
declare <2 x i64> @llvm.x86.avx512.maskz.vpmadd52l.uq.128(<2 x i64>, <2 x i64>, <2 x i64>, i8)
define <2 x i64>@test_int_x86_avx512_maskz_vpmadd52l_uq_128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_maskz_vpmadd52l_uq_128:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vmovaps %xmm0, %xmm3
; CHECK-NEXT: vpmadd52luq %xmm2, %xmm1, %xmm3 {%k1} {z}
; CHECK-NEXT: vmovaps %xmm0, %xmm4
; CHECK-NEXT: vpmadd52luq %xmm2, %xmm1, %xmm4
; CHECK-NEXT: vpxor %xmm2, %xmm2, %xmm2
; CHECK-NEXT: vpmadd52luq %xmm2, %xmm1, %xmm0 {%k1} {z}
; CHECK-NEXT: vpmadd52luq %xmm2, %xmm1, %xmm2 {%k1} {z}
; CHECK-NEXT: vpaddq %xmm0, %xmm3, %xmm0
; CHECK-NEXT: vpaddq %xmm2, %xmm4, %xmm1
; CHECK-NEXT: vpaddq %xmm0, %xmm1, %xmm0
; CHECK-NEXT: retq
%res = call <2 x i64> @llvm.x86.avx512.maskz.vpmadd52l.uq.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3)
%res1 = call <2 x i64> @llvm.x86.avx512.maskz.vpmadd52l.uq.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> zeroinitializer, i8 %x3)
%res2 = call <2 x i64> @llvm.x86.avx512.maskz.vpmadd52l.uq.128(<2 x i64> zeroinitializer, <2 x i64> %x1, <2 x i64> zeroinitializer, i8 %x3)
%res3 = call <2 x i64> @llvm.x86.avx512.maskz.vpmadd52l.uq.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 -1)
%res4 = add <2 x i64> %res, %res1
%res5 = add <2 x i64> %res3, %res2
%res6 = add <2 x i64> %res5, %res4
ret <2 x i64> %res6
}
declare <4 x i64> @llvm.x86.avx512.maskz.vpmadd52l.uq.256(<4 x i64>, <4 x i64>, <4 x i64>, i8)
define <4 x i64>@test_int_x86_avx512_maskz_vpmadd52l_uq_256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_maskz_vpmadd52l_uq_256:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vmovaps %ymm0, %ymm3
; CHECK-NEXT: vpmadd52luq %ymm2, %ymm1, %ymm3 {%k1} {z}
; CHECK-NEXT: vmovaps %ymm0, %ymm4
; CHECK-NEXT: vpmadd52luq %ymm2, %ymm1, %ymm4
; CHECK-NEXT: vpxor %ymm2, %ymm2, %ymm2
; CHECK-NEXT: vpmadd52luq %ymm2, %ymm1, %ymm0 {%k1} {z}
; CHECK-NEXT: vpmadd52luq %ymm2, %ymm1, %ymm2 {%k1} {z}
; CHECK-NEXT: vpaddq %ymm0, %ymm3, %ymm0
; CHECK-NEXT: vpaddq %ymm2, %ymm4, %ymm1
; CHECK-NEXT: vpaddq %ymm0, %ymm1, %ymm0
; CHECK-NEXT: retq
%res = call <4 x i64> @llvm.x86.avx512.maskz.vpmadd52l.uq.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3)
%res1 = call <4 x i64> @llvm.x86.avx512.maskz.vpmadd52l.uq.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> zeroinitializer, i8 %x3)
%res2 = call <4 x i64> @llvm.x86.avx512.maskz.vpmadd52l.uq.256(<4 x i64> zeroinitializer, <4 x i64> %x1, <4 x i64> zeroinitializer, i8 %x3)
%res3 = call <4 x i64> @llvm.x86.avx512.maskz.vpmadd52l.uq.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 -1)
%res4 = add <4 x i64> %res, %res1
%res5 = add <4 x i64> %res3, %res2
%res6 = add <4 x i64> %res5, %res4
ret <4 x i64> %res6
}
Reference in New Issue View Git Blame Copy Permalink