| ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py |
| ; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve2p1 -mattr=+b16b16 -verify-machineinstrs < %s | FileCheck %s |
| |
| define <vscale x 4 x float> @bfmlslb_f32(<vscale x 4 x float> %zda, <vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm) { |
| ; CHECK-LABEL: bfmlslb_f32: |
| ; CHECK: // %bb.0: |
| ; CHECK-NEXT: bfmlslb z0.s, z1.h, z2.h |
| ; CHECK-NEXT: ret |
| %out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlslb(<vscale x 4 x float> %zda, <vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm) |
| ret <vscale x 4 x float> %out |
| } |
| |
| define <vscale x 4 x float> @bfmlslt_f32(<vscale x 4 x float> %zda, <vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm) { |
| ; CHECK-LABEL: bfmlslt_f32: |
| ; CHECK: // %bb.0: |
| ; CHECK-NEXT: bfmlslt z0.s, z1.h, z2.h |
| ; CHECK-NEXT: ret |
| %out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlslt(<vscale x 4 x float> %zda, <vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm) |
| ret <vscale x 4 x float> %out |
| } |
| |
| define <vscale x 4 x float> @bfmlslb_lane_f32(<vscale x 4 x float> %zda, <vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm) { |
| ; CHECK-LABEL: bfmlslb_lane_f32: |
| ; CHECK: // %bb.0: |
| ; CHECK-NEXT: bfmlslb z0.s, z1.h, z2.h[7] |
| ; CHECK-NEXT: ret |
| %out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlslb.lane(<vscale x 4 x float> %zda, <vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm, i32 7) |
| ret <vscale x 4 x float> %out |
| } |
| |
| define <vscale x 4 x float> @bfmlslt_lane_f32(<vscale x 4 x float> %zda, <vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm) { |
| ; CHECK-LABEL: bfmlslt_lane_f32: |
| ; CHECK: // %bb.0: |
| ; CHECK-NEXT: bfmlslt z0.s, z1.h, z2.h[7] |
| ; CHECK-NEXT: ret |
| %out = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlslt.lane(<vscale x 4 x float> %zda, <vscale x 8 x bfloat> %zn, <vscale x 8 x bfloat> %zm, i32 7) |
| ret <vscale x 4 x float> %out |
| } |
| |
| declare <vscale x 4 x float> @llvm.aarch64.sve.bfmlslb(<vscale x 4 x float>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>) |
| declare <vscale x 4 x float> @llvm.aarch64.sve.bfmlslt(<vscale x 4 x float>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>) |
| declare <vscale x 4 x float> @llvm.aarch64.sve.bfmlslb.lane(<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.bfmlslt.lane(<vscale x 4 x float>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, i32) |