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
311 lines
11 KiB
LLVM
311 lines
11 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
|
|
; RUN: llc -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+mve -verify-machineinstrs %s -o - | FileCheck %s
|
|
|
|
define arm_aapcs_vfpcc <2 x i64> @zext_v2i1(i1 %a, i1 %b, i1 %c, i1 %d) {
|
|
; CHECK-LABEL: zext_v2i1:
|
|
; CHECK: @ %bb.0:
|
|
; CHECK-NEXT: vldr s1, .LCPI0_0
|
|
; CHECK-NEXT: and r1, r1, #1
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.f32 s3, s1
|
|
; CHECK-NEXT: vmov s2, r1
|
|
; CHECK-NEXT: vmov s0, r0
|
|
; CHECK-NEXT: bx lr
|
|
; CHECK-NEXT: .p2align 2
|
|
; CHECK-NEXT: @ %bb.1:
|
|
; CHECK-NEXT: .LCPI0_0:
|
|
; CHECK-NEXT: .long 0x00000000 @ float 0
|
|
%w = insertelement <2 x i1> poison, i1 %a, i32 0
|
|
%x = insertelement <2 x i1> %w, i1 %b, i32 1
|
|
%s = zext <2 x i1> %x to <2 x i64>
|
|
ret <2 x i64> %s
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <2 x i64> @sext_v2i1(i1 %a, i1 %b, i1 %c, i1 %d) {
|
|
; CHECK-LABEL: sext_v2i1:
|
|
; CHECK: @ %bb.0:
|
|
; CHECK-NEXT: and r1, r1, #1
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r1, r1, #0
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov q0[2], q0[0], r0, r1
|
|
; CHECK-NEXT: vmov q0[3], q0[1], r0, r1
|
|
; CHECK-NEXT: bx lr
|
|
%w = insertelement <2 x i1> poison, i1 %a, i32 0
|
|
%x = insertelement <2 x i1> %w, i1 %b, i32 1
|
|
%s = sext <2 x i1> %x to <2 x i64>
|
|
ret <2 x i64> %s
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <4 x i32> @zext_v4i1(i1 %a, i1 %b, i1 %c, i1 %d) {
|
|
; CHECK-LABEL: zext_v4i1:
|
|
; CHECK: @ %bb.0:
|
|
; CHECK-NEXT: and r2, r2, #1
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov q0[2], q0[0], r0, r2
|
|
; CHECK-NEXT: and r0, r3, #1
|
|
; CHECK-NEXT: and r1, r1, #1
|
|
; CHECK-NEXT: vmov q0[3], q0[1], r1, r0
|
|
; CHECK-NEXT: bx lr
|
|
%w = insertelement <4 x i1> poison, i1 %a, i32 0
|
|
%x = insertelement <4 x i1> %w, i1 %b, i32 1
|
|
%y = insertelement <4 x i1> %x, i1 %c, i32 2
|
|
%z = insertelement <4 x i1> %y, i1 %d, i32 3
|
|
%s = zext <4 x i1> %z to <4 x i32>
|
|
ret <4 x i32> %s
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <4 x i32> @sext_v4i1(i1 %a, i1 %b, i1 %c, i1 %d) {
|
|
; CHECK-LABEL: sext_v4i1:
|
|
; CHECK: @ %bb.0:
|
|
; CHECK-NEXT: and r2, r2, #1
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r2, r2, #0
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: and r1, r1, #1
|
|
; CHECK-NEXT: vmov q0[2], q0[0], r0, r2
|
|
; CHECK-NEXT: and r0, r3, #1
|
|
; CHECK-NEXT: rsbs r1, r1, #0
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov q0[3], q0[1], r1, r0
|
|
; CHECK-NEXT: bx lr
|
|
%w = insertelement <4 x i1> poison, i1 %a, i32 0
|
|
%x = insertelement <4 x i1> %w, i1 %b, i32 1
|
|
%y = insertelement <4 x i1> %x, i1 %c, i32 2
|
|
%z = insertelement <4 x i1> %y, i1 %d, i32 3
|
|
%s = sext <4 x i1> %z to <4 x i32>
|
|
ret <4 x i32> %s
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <8 x i16> @zext_v8i1(i1 %a, i1 %b, i1 %c, i1 %d, i1 %e, i1 %f, i1 %g, i1 %h) {
|
|
; CHECK-LABEL: zext_v8i1:
|
|
; CHECK: @ %bb.0:
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.16 q0[0], r0
|
|
; CHECK-NEXT: and r0, r1, #1
|
|
; CHECK-NEXT: vmov.16 q0[1], r0
|
|
; CHECK-NEXT: and r0, r2, #1
|
|
; CHECK-NEXT: vmov.16 q0[2], r0
|
|
; CHECK-NEXT: and r0, r3, #1
|
|
; CHECK-NEXT: vmov.16 q0[3], r0
|
|
; CHECK-NEXT: ldr r0, [sp]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.16 q0[4], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #4]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.16 q0[5], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #8]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.16 q0[6], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #12]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.16 q0[7], r0
|
|
; CHECK-NEXT: bx lr
|
|
%w = insertelement <8 x i1> poison, i1 %a, i32 0
|
|
%x = insertelement <8 x i1> %w, i1 %b, i32 1
|
|
%y = insertelement <8 x i1> %x, i1 %c, i32 2
|
|
%z = insertelement <8 x i1> %y, i1 %d, i32 3
|
|
%v = insertelement <8 x i1> %z, i1 %e, i32 4
|
|
%u = insertelement <8 x i1> %v, i1 %f, i32 5
|
|
%t = insertelement <8 x i1> %u, i1 %g, i32 6
|
|
%r = insertelement <8 x i1> %t, i1 %h, i32 7
|
|
%s = zext <8 x i1> %r to <8 x i16>
|
|
ret <8 x i16> %s
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <8 x i16> @sext_v8i1(i1 %a, i1 %b, i1 %c, i1 %d, i1 %e, i1 %f, i1 %g, i1 %h) {
|
|
; CHECK-LABEL: sext_v8i1:
|
|
; CHECK: @ %bb.0:
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.16 q0[0], r0
|
|
; CHECK-NEXT: and r0, r1, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.16 q0[1], r0
|
|
; CHECK-NEXT: and r0, r2, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.16 q0[2], r0
|
|
; CHECK-NEXT: and r0, r3, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.16 q0[3], r0
|
|
; CHECK-NEXT: ldr r0, [sp]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.16 q0[4], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #4]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.16 q0[5], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #8]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.16 q0[6], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #12]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.16 q0[7], r0
|
|
; CHECK-NEXT: bx lr
|
|
%w = insertelement <8 x i1> poison, i1 %a, i32 0
|
|
%x = insertelement <8 x i1> %w, i1 %b, i32 1
|
|
%y = insertelement <8 x i1> %x, i1 %c, i32 2
|
|
%z = insertelement <8 x i1> %y, i1 %d, i32 3
|
|
%v = insertelement <8 x i1> %z, i1 %e, i32 4
|
|
%u = insertelement <8 x i1> %v, i1 %f, i32 5
|
|
%t = insertelement <8 x i1> %u, i1 %g, i32 6
|
|
%r = insertelement <8 x i1> %t, i1 %h, i32 7
|
|
%s = sext <8 x i1> %r to <8 x i16>
|
|
ret <8 x i16> %s
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <16 x i8> @zext_v16i1(i1 %a, i1 %b, i1 %c, i1 %d, i1 %e, i1 %f, i1 %g, i1 %h, i1 %i, i1 %j, i1 %k, i1 %l, i1 %m, i1 %n, i1 %o, i1 %p) {
|
|
; CHECK-LABEL: zext_v16i1:
|
|
; CHECK: @ %bb.0:
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[0], r0
|
|
; CHECK-NEXT: and r0, r1, #1
|
|
; CHECK-NEXT: vmov.8 q0[1], r0
|
|
; CHECK-NEXT: and r0, r2, #1
|
|
; CHECK-NEXT: vmov.8 q0[2], r0
|
|
; CHECK-NEXT: and r0, r3, #1
|
|
; CHECK-NEXT: vmov.8 q0[3], r0
|
|
; CHECK-NEXT: ldr r0, [sp]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[4], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #4]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[5], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #8]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[6], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #12]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[7], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #16]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[8], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #20]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[9], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #24]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[10], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #28]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[11], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #32]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[12], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #36]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[13], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #40]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[14], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #44]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: vmov.8 q0[15], r0
|
|
; CHECK-NEXT: bx lr
|
|
%w = insertelement <16 x i1> poison, i1 %a, i32 0
|
|
%x = insertelement <16 x i1> %w, i1 %b, i32 1
|
|
%y = insertelement <16 x i1> %x, i1 %c, i32 2
|
|
%z = insertelement <16 x i1> %y, i1 %d, i32 3
|
|
%v = insertelement <16 x i1> %z, i1 %e, i32 4
|
|
%u = insertelement <16 x i1> %v, i1 %f, i32 5
|
|
%t = insertelement <16 x i1> %u, i1 %g, i32 6
|
|
%r = insertelement <16 x i1> %t, i1 %h, i32 7
|
|
%w1 = insertelement <16 x i1> %r, i1 %i, i32 8
|
|
%x1 = insertelement <16 x i1> %w1, i1 %j, i32 9
|
|
%y1 = insertelement <16 x i1> %x1, i1 %k, i32 10
|
|
%z1 = insertelement <16 x i1> %y1, i1 %l, i32 11
|
|
%v1 = insertelement <16 x i1> %z1, i1 %m, i32 12
|
|
%u1 = insertelement <16 x i1> %v1, i1 %n, i32 13
|
|
%t1 = insertelement <16 x i1> %u1, i1 %o, i32 14
|
|
%r1 = insertelement <16 x i1> %t1, i1 %p, i32 15
|
|
%s = zext <16 x i1> %r1 to <16 x i8>
|
|
ret <16 x i8> %s
|
|
}
|
|
|
|
define arm_aapcs_vfpcc <16 x i8> @sext_v16i1(i1 %a, i1 %b, i1 %c, i1 %d, i1 %e, i1 %f, i1 %g, i1 %h, i1 %i, i1 %j, i1 %k, i1 %l, i1 %m, i1 %n, i1 %o, i1 %p) {
|
|
; CHECK-LABEL: sext_v16i1:
|
|
; CHECK: @ %bb.0:
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[0], r0
|
|
; CHECK-NEXT: and r0, r1, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[1], r0
|
|
; CHECK-NEXT: and r0, r2, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[2], r0
|
|
; CHECK-NEXT: and r0, r3, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[3], r0
|
|
; CHECK-NEXT: ldr r0, [sp]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[4], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #4]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[5], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #8]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[6], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #12]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[7], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #16]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[8], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #20]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[9], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #24]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[10], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #28]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[11], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #32]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[12], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #36]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[13], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #40]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[14], r0
|
|
; CHECK-NEXT: ldr r0, [sp, #44]
|
|
; CHECK-NEXT: and r0, r0, #1
|
|
; CHECK-NEXT: rsbs r0, r0, #0
|
|
; CHECK-NEXT: vmov.8 q0[15], r0
|
|
; CHECK-NEXT: bx lr
|
|
%w = insertelement <16 x i1> poison, i1 %a, i32 0
|
|
%x = insertelement <16 x i1> %w, i1 %b, i32 1
|
|
%y = insertelement <16 x i1> %x, i1 %c, i32 2
|
|
%z = insertelement <16 x i1> %y, i1 %d, i32 3
|
|
%v = insertelement <16 x i1> %z, i1 %e, i32 4
|
|
%u = insertelement <16 x i1> %v, i1 %f, i32 5
|
|
%t = insertelement <16 x i1> %u, i1 %g, i32 6
|
|
%r = insertelement <16 x i1> %t, i1 %h, i32 7
|
|
%w1 = insertelement <16 x i1> %r, i1 %i, i32 8
|
|
%x1 = insertelement <16 x i1> %w1, i1 %j, i32 9
|
|
%y1 = insertelement <16 x i1> %x1, i1 %k, i32 10
|
|
%z1 = insertelement <16 x i1> %y1, i1 %l, i32 11
|
|
%v1 = insertelement <16 x i1> %z1, i1 %m, i32 12
|
|
%u1 = insertelement <16 x i1> %v1, i1 %n, i32 13
|
|
%t1 = insertelement <16 x i1> %u1, i1 %o, i32 14
|
|
%r1 = insertelement <16 x i1> %t1, i1 %p, i32 15
|
|
%s = sext <16 x i1> %r1 to <16 x i8>
|
|
ret <16 x i8> %s
|
|
}
|