test/LongVectorLowering/frem.ll - master - redgpu/clspv

test/LongVectorLowering/frem.ll
99 строк4 KB
Marco Antognini
Lower BinaryOperator instructions on long-vector (#642)
29 сен 2020, 15:38
29 сен 2020, 15:389b29851
; RUN: clspv-opt --LongVectorLowering %s -o %t
; RUN: FileCheck %s < %t

target datalayout = "e-p:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024"
target triple = "spir-unknown-unknown"

define spir_func void @test8(<4 x float> addrspace(1)* %ptr) {
entry:
  %x0 = load <4 x float>, <4 x float> addrspace(1)* %ptr, align 16
  %ptr1 = getelementptr <4 x float>, <4 x float> addrspace(1)* %ptr, i32 1
  %x1 = load <4 x float>, <4 x float> addrspace(1)* %ptr1, align 16
  %a = shufflevector <4 x float> %x0, <4 x float> %x1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>

  %ptr2 = getelementptr <4 x float>, <4 x float> addrspace(1)* %ptr, i32 2
  %x2 = load <4 x float>, <4 x float> addrspace(1)* %ptr2, align 16
  %ptr3 = getelementptr <4 x float>, <4 x float> addrspace(1)* %ptr, i32 3
  %x3 = load <4 x float>, <4 x float> addrspace(1)* %ptr3, align 16
  %b = shufflevector <4 x float> %x2, <4 x float> %x3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>

  %frem = frem <8 x float> %a, %b

  %c = shufflevector <8 x float> %frem, <8 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  %ptr4 = getelementptr <4 x float>, <4 x float> addrspace(1)* %ptr, i32 4
  store <4 x float> %c, <4 x float> addrspace(1)* %ptr4, align 16

  %d = shufflevector <8 x float> %frem, <8 x float> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
  %ptr5 = getelementptr <4 x float>, <4 x float> addrspace(1)* %ptr, i32 5
  store <4 x float> %d, <4 x float> addrspace(1)* %ptr5, align 16

  ret void
}

; CHECK-LABEL: @test8
; CHECK-NOT: frem <8 x float>
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK-NOT: frem <8 x float>

define spir_func void @test16(<4 x float> addrspace(1)* %ptr) {
entry:
  %x0 = load <4 x float>, <4 x float> addrspace(1)* %ptr, align 16
  %ptr1 = getelementptr <4 x float>, <4 x float> addrspace(1)* %ptr, i32 1
  %x1 = load <4 x float>, <4 x float> addrspace(1)* %ptr1, align 16
  %a = shufflevector <4 x float> %x0, <4 x float> %x1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>

  %ptr2 = getelementptr <4 x float>, <4 x float> addrspace(1)* %ptr, i32 2
  %x2 = load <4 x float>, <4 x float> addrspace(1)* %ptr2, align 16
  %ptr3 = getelementptr <4 x float>, <4 x float> addrspace(1)* %ptr, i32 3
  %x3 = load <4 x float>, <4 x float> addrspace(1)* %ptr3, align 16
  %b = shufflevector <4 x float> %x2, <4 x float> %x3, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>

  %c = shufflevector <8 x float> %a, <8 x float> %b, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>

  %frem = frem <16 x float> %c, <float 0.0, float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0, float 8.0, float 9.0, float 10.0, float 11.0, float 12.0, float 13.0, float 14.0, float 15.0>

  %d = shufflevector <16 x float> %frem, <16 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  %ptr4 = getelementptr <4 x float>, <4 x float> addrspace(1)* %ptr, i32 4
  store <4 x float> %d, <4 x float> addrspace(1)* %ptr4, align 16

  %e = shufflevector <16 x float> %frem, <16 x float> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
  %ptr5 = getelementptr <4 x float>, <4 x float> addrspace(1)* %ptr, i32 5
  store <4 x float> %e, <4 x float> addrspace(1)* %ptr5, align 16

  %f = shufflevector <16 x float> %frem, <16 x float> undef, <4 x i32> <i32 8, i32 9, i32 10, i32 11>
  %ptr6 = getelementptr <4 x float>, <4 x float> addrspace(1)* %ptr, i32 6
  store <4 x float> %f, <4 x float> addrspace(1)* %ptr6, align 16

  %g = shufflevector <16 x float> %frem, <16 x float> undef, <4 x i32> <i32 12, i32 13, i32 14, i32 15>
  %ptr7 = getelementptr <4 x float>, <4 x float> addrspace(1)* %ptr, i32 7
  store <4 x float> %g, <4 x float> addrspace(1)* %ptr7, align 16

  ret void
}

; CHECK-LABEL: @test16
; CHECK-NOT: frem <16 x float>
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK: frem float
; CHECK-NOT: frem <16 x float>