CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
136 lines
3.8 KiB
LLVM
136 lines
3.8 KiB
LLVM
; Test multiplication of two f64s, producing an f64 result.
|
|
;
|
|
; 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=z13 | FileCheck %s
|
|
|
|
declare double @foo()
|
|
|
|
; Check register multiplication.
|
|
define double @f1(double %f1, double %f2) {
|
|
; CHECK-LABEL: f1:
|
|
; CHECK: mdbr %f0, %f2
|
|
; CHECK: br %r14
|
|
%res = fmul double %f1, %f2
|
|
ret double %res
|
|
}
|
|
|
|
; Check the low end of the MDB range.
|
|
define double @f2(double %f1, ptr %ptr) {
|
|
; CHECK-LABEL: f2:
|
|
; CHECK: mdb %f0, 0(%r2)
|
|
; CHECK: br %r14
|
|
%f2 = load double, ptr %ptr
|
|
%res = fmul double %f1, %f2
|
|
ret double %res
|
|
}
|
|
|
|
; Check the high end of the aligned MDB range.
|
|
define double @f3(double %f1, ptr %base) {
|
|
; CHECK-LABEL: f3:
|
|
; CHECK: mdb %f0, 4088(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr double, ptr %base, i64 511
|
|
%f2 = load double, ptr %ptr
|
|
%res = fmul double %f1, %f2
|
|
ret double %res
|
|
}
|
|
|
|
; Check the next doubleword up, which needs separate address logic.
|
|
; Other sequences besides this one would be OK.
|
|
define double @f4(double %f1, ptr %base) {
|
|
; CHECK-LABEL: f4:
|
|
; CHECK: aghi %r2, 4096
|
|
; CHECK: mdb %f0, 0(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr double, ptr %base, i64 512
|
|
%f2 = load double, ptr %ptr
|
|
%res = fmul double %f1, %f2
|
|
ret double %res
|
|
}
|
|
|
|
; Check negative displacements, which also need separate address logic.
|
|
define double @f5(double %f1, ptr %base) {
|
|
; CHECK-LABEL: f5:
|
|
; CHECK: aghi %r2, -8
|
|
; CHECK: mdb %f0, 0(%r2)
|
|
; CHECK: br %r14
|
|
%ptr = getelementptr double, ptr %base, i64 -1
|
|
%f2 = load double, ptr %ptr
|
|
%res = fmul double %f1, %f2
|
|
ret double %res
|
|
}
|
|
|
|
; Check that MDB allows indices.
|
|
define double @f6(double %f1, ptr %base, i64 %index) {
|
|
; CHECK-LABEL: f6:
|
|
; CHECK: sllg %r1, %r3, 3
|
|
; CHECK: mdb %f0, 800(%r1,%r2)
|
|
; CHECK: br %r14
|
|
%ptr1 = getelementptr double, ptr %base, i64 %index
|
|
%ptr2 = getelementptr double, ptr %ptr1, i64 100
|
|
%f2 = load double, ptr %ptr2
|
|
%res = fmul double %f1, %f2
|
|
ret double %res
|
|
}
|
|
|
|
; Check that multiplications of spilled values can use MDB rather than MDBR.
|
|
define double @f7(ptr %ptr0) {
|
|
; CHECK-LABEL: f7:
|
|
; CHECK: brasl %r14, foo@PLT
|
|
; CHECK-SCALAR: mdb %f0, 160(%r15)
|
|
; CHECK: br %r14
|
|
%ptr1 = getelementptr double, ptr %ptr0, i64 2
|
|
%ptr2 = getelementptr double, ptr %ptr0, i64 4
|
|
%ptr3 = getelementptr double, ptr %ptr0, i64 6
|
|
%ptr4 = getelementptr double, ptr %ptr0, i64 8
|
|
%ptr5 = getelementptr double, ptr %ptr0, i64 10
|
|
%ptr6 = getelementptr double, ptr %ptr0, i64 12
|
|
%ptr7 = getelementptr double, ptr %ptr0, i64 14
|
|
%ptr8 = getelementptr double, ptr %ptr0, i64 16
|
|
%ptr9 = getelementptr double, ptr %ptr0, i64 18
|
|
%ptr10 = getelementptr double, ptr %ptr0, i64 20
|
|
|
|
%val0 = load double, ptr %ptr0
|
|
%val1 = load double, ptr %ptr1
|
|
%val2 = load double, ptr %ptr2
|
|
%val3 = load double, ptr %ptr3
|
|
%val4 = load double, ptr %ptr4
|
|
%val5 = load double, ptr %ptr5
|
|
%val6 = load double, ptr %ptr6
|
|
%val7 = load double, ptr %ptr7
|
|
%val8 = load double, ptr %ptr8
|
|
%val9 = load double, ptr %ptr9
|
|
%val10 = load double, ptr %ptr10
|
|
|
|
%ret = call double @foo()
|
|
|
|
%mul0 = fmul double %ret, %val0
|
|
%mul1 = fmul double %mul0, %val1
|
|
%mul2 = fmul double %mul1, %val2
|
|
%mul3 = fmul double %mul2, %val3
|
|
%mul4 = fmul double %mul3, %val4
|
|
%mul5 = fmul double %mul4, %val5
|
|
%mul6 = fmul double %mul5, %val6
|
|
%mul7 = fmul double %mul6, %val7
|
|
%mul8 = fmul double %mul7, %val8
|
|
%mul9 = fmul double %mul8, %val9
|
|
%mul10 = fmul double %mul9, %val10
|
|
|
|
ret double %mul10
|
|
}
|
|
|
|
; Check that reassociation flags do not get in the way of MDB.
|
|
define double @f8(ptr %x) {
|
|
; CHECK-LABEL: f8:
|
|
; CHECK: ld %f0
|
|
; CHECK: mdb %f0
|
|
; CHECK: br %r14
|
|
entry:
|
|
%0 = load double, ptr %x, align 8
|
|
%arrayidx1 = getelementptr inbounds double, ptr %x, i64 1
|
|
%1 = load double, ptr %arrayidx1, align 8
|
|
%add = fmul reassoc nsz arcp contract afn double %1, %0
|
|
ret double %add
|
|
}
|