derek.holloway
7502bf64a4
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
380 lines
27 KiB
LLVM
380 lines
27 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6
|
|
; RUN: opt -mattr=+sve,+bf16 < %s -passes=msan -S | FileCheck %s
|
|
;
|
|
; Forked from llvm/test/CodeGen/AArch64/sve-intrinsics-bfloat.ll
|
|
;
|
|
; Strictly handled:
|
|
; - llvm.aarch64.sve.bfdot
|
|
; - llvm.aarch64.sve.bfdot.lane.v2
|
|
; - llvm.aarch64.sve.bfmlalb
|
|
; - llvm.aarch64.sve.bfmlalb.lane.v2
|
|
; - llvm.aarch64.sve.bfmlalt
|
|
; - llvm.aarch64.sve.bfmlalt.lane.v2
|
|
; - llvm.aarch64.sve.fmmla.nxv4f32.nxv8bf16
|
|
; - llvm.aarch64.sve.convert.from.svbool.nxv4i1
|
|
; - llvm.aarch64.sve.convert.to.svbool.nxv8i1
|
|
; - llvm.aarch64.sve.fcvt.bf16f32.v2
|
|
; - llvm.aarch64.sve.fcvtnt.bf16f32.v2
|
|
;
|
|
; Heuristically handled: (none)
|
|
|
|
target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
|
|
target triple = "aarch64--linux-android9001"
|
|
|
|
define <vscale x 4 x float> @bfdot_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfdot_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0:[0-9]+]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]])
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfdot_lane_0_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfdot_lane_0_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 0)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 0)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfdot_lane_1_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfdot_lane_1_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 1)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 1)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfdot_lane_2_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfdot_lane_2_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 2)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 2)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfdot_lane_3_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfdot_lane_3_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 3)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 3)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
|
|
define <vscale x 4 x float> @bfmlalb_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalb_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]])
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalb_lane_0_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalb_lane_0_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 0)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 0)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalb_lane_1_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalb_lane_1_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 1)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 1)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalb_lane_2_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalb_lane_2_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 2)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 2)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalb_lane_3_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalb_lane_3_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 3)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 3)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalb_lane_4_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalb_lane_4_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 4)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 4)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalb_lane_5_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalb_lane_5_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 5)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 5)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalb_lane_6_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalb_lane_6_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 6)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 6)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalb_lane_7_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalb_lane_7_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 7)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 7)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
|
|
define <vscale x 4 x float> @bfmlalt_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalt_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]])
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalt_lane_0_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalt_lane_0_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 0)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 0)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalt_lane_1_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalt_lane_1_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 1)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 1)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalt_lane_2_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalt_lane_2_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 2)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 2)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalt_lane_3_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalt_lane_3_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 3)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 3)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalt_lane_4_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalt_lane_4_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 4)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 4)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalt_lane_5_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalt_lane_5_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 5)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 5)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalt_lane_6_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalt_lane_6_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 6)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 6)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
define <vscale x 4 x float> @bfmlalt_lane_7_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmlalt_lane_7_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]], i32 7)
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c, i32 7)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
|
|
define <vscale x 4 x float> @bfmmla_f32(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 4 x float> @bfmmla_f32(
|
|
; CHECK-SAME: <vscale x 4 x float> [[A:%.*]], <vscale x 8 x bfloat> [[B:%.*]], <vscale x 8 x bfloat> [[C:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmmla.nxv4f32.nxv8bf16(<vscale x 4 x float> [[A]], <vscale x 8 x bfloat> [[B]], <vscale x 8 x bfloat> [[C]])
|
|
; CHECK-NEXT: store <vscale x 4 x i32> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 4 x float> [[OUT]]
|
|
;
|
|
%out = call <vscale x 4 x float> @llvm.aarch64.sve.fmmla.nxv4f32.nxv8bf16(<vscale x 4 x float> %a, <vscale x 8 x bfloat> %b, <vscale x 8 x bfloat> %c)
|
|
ret <vscale x 4 x float> %out
|
|
}
|
|
|
|
|
|
define <vscale x 8 x bfloat> @fcvt_bf16_f32(<vscale x 8 x bfloat> %a, <vscale x 4 x i1> %pg, <vscale x 4 x float> %b) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 8 x bfloat> @fcvt_bf16_f32(
|
|
; CHECK-SAME: <vscale x 8 x bfloat> [[A:%.*]], <vscale x 4 x i1> [[PG:%.*]], <vscale x 4 x float> [[B:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvt.bf16f32.v2(<vscale x 8 x bfloat> [[A]], <vscale x 4 x i1> [[PG]], <vscale x 4 x float> [[B]])
|
|
; CHECK-NEXT: store <vscale x 8 x i16> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 8 x bfloat> [[OUT]]
|
|
;
|
|
%out = call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvt.bf16f32.v2(<vscale x 8 x bfloat> %a, <vscale x 4 x i1> %pg, <vscale x 4 x float> %b)
|
|
ret <vscale x 8 x bfloat> %out
|
|
}
|
|
|
|
define <vscale x 8 x bfloat> @fcvt_bf16_f32_deprecated(<vscale x 8 x bfloat> %a, <vscale x 8 x i1> %pg, <vscale x 4 x float> %b) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 8 x bfloat> @fcvt_bf16_f32_deprecated(
|
|
; CHECK-SAME: <vscale x 8 x bfloat> [[A:%.*]], <vscale x 8 x i1> [[PG:%.*]], <vscale x 4 x float> [[B:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[PG]])
|
|
; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[TMP1]])
|
|
; CHECK-NEXT: [[OUT1:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvt.bf16f32.v2(<vscale x 8 x bfloat> [[A]], <vscale x 4 x i1> [[TMP2]], <vscale x 4 x float> [[B]])
|
|
; CHECK-NEXT: store <vscale x 8 x i16> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 8 x bfloat> [[OUT1]]
|
|
;
|
|
%out = call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvt.bf16f32(<vscale x 8 x bfloat> %a, <vscale x 8 x i1> %pg, <vscale x 4 x float> %b)
|
|
ret <vscale x 8 x bfloat> %out
|
|
}
|
|
|
|
|
|
define <vscale x 8 x bfloat> @fcvtnt_bf16_f32(<vscale x 8 x bfloat> %a, <vscale x 4 x i1> %pg, <vscale x 4 x float> %b) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 8 x bfloat> @fcvtnt_bf16_f32(
|
|
; CHECK-SAME: <vscale x 8 x bfloat> [[A:%.*]], <vscale x 4 x i1> [[PG:%.*]], <vscale x 4 x float> [[B:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[OUT:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvtnt.bf16f32.v2(<vscale x 8 x bfloat> [[A]], <vscale x 4 x i1> [[PG]], <vscale x 4 x float> [[B]])
|
|
; CHECK-NEXT: store <vscale x 8 x i16> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 8 x bfloat> [[OUT]]
|
|
;
|
|
%out = call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvtnt.bf16f32.v2(<vscale x 8 x bfloat> %a, <vscale x 4 x i1> %pg, <vscale x 4 x float> %b)
|
|
ret <vscale x 8 x bfloat> %out
|
|
}
|
|
|
|
define <vscale x 8 x bfloat> @fcvtnt_bf16_f32_deprecated(<vscale x 8 x bfloat> %a, <vscale x 8 x i1> %pg, <vscale x 4 x float> %b) nounwind sanitize_memory {
|
|
; CHECK-LABEL: define <vscale x 8 x bfloat> @fcvtnt_bf16_f32_deprecated(
|
|
; CHECK-SAME: <vscale x 8 x bfloat> [[A:%.*]], <vscale x 8 x i1> [[PG:%.*]], <vscale x 4 x float> [[B:%.*]]) #[[ATTR0]] {
|
|
; CHECK-NEXT: call void @llvm.donothing()
|
|
; CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[PG]])
|
|
; CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[TMP1]])
|
|
; CHECK-NEXT: [[OUT1:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvtnt.bf16f32.v2(<vscale x 8 x bfloat> [[A]], <vscale x 4 x i1> [[TMP2]], <vscale x 4 x float> [[B]])
|
|
; CHECK-NEXT: store <vscale x 8 x i16> zeroinitializer, ptr @__msan_retval_tls, align 8
|
|
; CHECK-NEXT: ret <vscale x 8 x bfloat> [[OUT1]]
|
|
;
|
|
%out = call <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvtnt.bf16f32(<vscale x 8 x bfloat> %a, <vscale x 8 x i1> %pg, <vscale x 4 x float> %b)
|
|
ret <vscale x 8 x bfloat> %out
|
|
}
|
|
|
|
declare <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>)
|
|
declare <vscale x 4 x float> @llvm.aarch64.sve.bfdot.lane.v2(<vscale x 4 x float>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, i32)
|
|
declare <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb(<vscale x 4 x float>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>)
|
|
declare <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb.lane.v2(<vscale x 4 x float>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, i32)
|
|
declare <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt(<vscale x 4 x float>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>)
|
|
declare <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt.lane.v2(<vscale x 4 x float>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, i32)
|
|
declare <vscale x 4 x float> @llvm.aarch64.sve.fmmla.nxv4f32.nxv8bf16(<vscale x 4 x float>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>)
|
|
declare <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvt.bf16f32(<vscale x 8 x bfloat>, <vscale x 8 x i1>, <vscale x 4 x float>)
|
|
declare <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvt.bf16f32.v2(<vscale x 8 x bfloat>, <vscale x 4 x i1>, <vscale x 4 x float>)
|
|
declare <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvtnt.bf16f32(<vscale x 8 x bfloat>, <vscale x 8 x i1>, <vscale x 4 x float>)
|
|
declare <vscale x 8 x bfloat> @llvm.aarch64.sve.fcvtnt.bf16f32.v2(<vscale x 8 x bfloat>, <vscale x 4 x i1>, <vscale x 4 x float>)
|