; Test 32-bit square root.
;
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z10 \
; RUN: | FileCheck -check-prefix=CHECK -check-prefix=CHECK-SCALAR %s
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z14 | FileCheck %s
declare float @llvm.sqrt.f32(float)
declare float @sqrtf(float)
; Check register square root.
define float @f1(float %val) {
; CHECK-LABEL: f1:
; CHECK: sqebr %f0, %f0
; CHECK: br %r14
%res = call float @llvm.sqrt.f32(float %val)
ret float %res
}
; Check the low end of the SQEB range.
define float @f2(ptr %ptr) {
; CHECK-LABEL: f2:
; CHECK: sqeb %f0, 0(%r2)
; CHECK: br %r14
%val = load float, ptr %ptr
%res = call float @llvm.sqrt.f32(float %val)
ret float %res
}
; Check the high end of the aligned SQEB range.
define float @f3(ptr %base) {
; CHECK-LABEL: f3:
; CHECK: sqeb %f0, 4092(%r2)
; CHECK: br %r14
%ptr = getelementptr float, ptr %base, i64 1023
%val = load float, ptr %ptr
%res = call float @llvm.sqrt.f32(float %val)
ret float %res
}
; Check the next word up, which needs separate address logic.
; Other sequences besides this one would be OK.
define float @f4(ptr %base) {
; CHECK-LABEL: f4:
; CHECK: aghi %r2, 4096
; CHECK: sqeb %f0, 0(%r2)
; CHECK: br %r14
%ptr = getelementptr float, ptr %base, i64 1024
%val = load float, ptr %ptr
%res = call float @llvm.sqrt.f32(float %val)
ret float %res
}
; Check negative displacements, which also need separate address logic.
define float @f5(ptr %base) {
; CHECK-LABEL: f5:
; CHECK: aghi %r2, -4
; CHECK: sqeb %f0, 0(%r2)
; CHECK: br %r14
%ptr = getelementptr float, ptr %base, i64 -1
%val = load float, ptr %ptr
%res = call float @llvm.sqrt.f32(float %val)
ret float %res
}
; Check that SQEB allows indices.
define float @f6(ptr %base, i64 %index) {
; CHECK-LABEL: f6:
; CHECK: sllg %r1, %r3, 2
; CHECK: sqeb %f0, 400(%r1,%r2)
; CHECK: br %r14
%ptr1 = getelementptr float, ptr %base, i64 %index
%ptr2 = getelementptr float, ptr %ptr1, i64 100
%val = load float, ptr %ptr2
%res = call float @llvm.sqrt.f32(float %val)
ret float %res
}
; Test a case where we spill the source of at least one SQEBR. We want
; to use SQEB if possible.
define void @f7(ptr %ptr) {
; CHECK-LABEL: f7:
; CHECK-SCALAR: sqeb {{%f[0-9]+}}, 16{{[04]}}(%r15)
; CHECK: br %r14
%val0 = load volatile float, ptr %ptr
%val1 = load volatile float, ptr %ptr
%val2 = load volatile float, ptr %ptr
%val3 = load volatile float, ptr %ptr
%val4 = load volatile float, ptr %ptr
%val5 = load volatile float, ptr %ptr
%val6 = load volatile float, ptr %ptr
%val7 = load volatile float, ptr %ptr
%val8 = load volatile float, ptr %ptr
%val9 = load volatile float, ptr %ptr
%val10 = load volatile float, ptr %ptr
%val11 = load volatile float, ptr %ptr
%val12 = load volatile float, ptr %ptr
%val13 = load volatile float, ptr %ptr
%val14 = load volatile float, ptr %ptr
%val15 = load volatile float, ptr %ptr
%val16 = load volatile float, ptr %ptr
%sqrt0 = call float @llvm.sqrt.f32(float %val0)
%sqrt1 = call float @llvm.sqrt.f32(float %val1)
%sqrt2 = call float @llvm.sqrt.f32(float %val2)
%sqrt3 = call float @llvm.sqrt.f32(float %val3)
%sqrt4 = call float @llvm.sqrt.f32(float %val4)
%sqrt5 = call float @llvm.sqrt.f32(float %val5)
%sqrt6 = call float @llvm.sqrt.f32(float %val6)
%sqrt7 = call float @llvm.sqrt.f32(float %val7)
%sqrt8 = call float @llvm.sqrt.f32(float %val8)
%sqrt9 = call float @llvm.sqrt.f32(float %val9)
%sqrt10 = call float @llvm.sqrt.f32(float %val10)
%sqrt11 = call float @llvm.sqrt.f32(float %val11)
%sqrt12 = call float @llvm.sqrt.f32(float %val12)
%sqrt13 = call float @llvm.sqrt.f32(float %val13)
%sqrt14 = call float @llvm.sqrt.f32(float %val14)
%sqrt15 = call float @llvm.sqrt.f32(float %val15)
%sqrt16 = call float @llvm.sqrt.f32(float %val16)
store volatile float %val0, ptr %ptr
store volatile float %val1, ptr %ptr
store volatile float %val2, ptr %ptr
store volatile float %val3, ptr %ptr
store volatile float %val4, ptr %ptr
store volatile float %val5, ptr %ptr
store volatile float %val6, ptr %ptr
store volatile float %val7, ptr %ptr
store volatile float %val8, ptr %ptr
store volatile float %val9, ptr %ptr
store volatile float %val10, ptr %ptr
store volatile float %val11, ptr %ptr
store volatile float %val12, ptr %ptr
store volatile float %val13, ptr %ptr
store volatile float %val14, ptr %ptr
store volatile float %val15, ptr %ptr
store volatile float %val16, ptr %ptr
store volatile float %sqrt0, ptr %ptr
store volatile float %sqrt1, ptr %ptr
store volatile float %sqrt2, ptr %ptr
store volatile float %sqrt3, ptr %ptr
store volatile float %sqrt4, ptr %ptr
store volatile float %sqrt5, ptr %ptr
store volatile float %sqrt6, ptr %ptr
store volatile float %sqrt7, ptr %ptr
store volatile float %sqrt8, ptr %ptr
store volatile float %sqrt9, ptr %ptr
store volatile float %sqrt10, ptr %ptr
store volatile float %sqrt11, ptr %ptr
store volatile float %sqrt12, ptr %ptr
store volatile float %sqrt13, ptr %ptr
store volatile float %sqrt14, ptr %ptr
store volatile float %sqrt15, ptr %ptr
store volatile float %sqrt16, ptr %ptr
ret void
}
; Check that a call to the normal sqrtf function is lowered.
define float @f8(float %dummy, float %val) {
; CHECK-LABEL: f8:
; CHECK: sqebr %f0, %f2
; CHECK: cebr %f0, %f0
; CHECK: bnor %r14
; CHECK: {{ler|ldr}} %f0, %f2
; CHECK: jg sqrtf@PLT
%res = tail call float @sqrtf(float %val)
ret float %res
}
Codebase Browser
llvm