aboutsummaryrefslogtreecommitdiff
path: root/llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll')
-rw-r--r--llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll41
1 files changed, 41 insertions, 0 deletions
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll b/llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll
index 47777e3..0d58afd 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll
@@ -2,6 +2,7 @@
; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=tahiti -verify-machineinstrs < %s | FileCheck -check-prefixes=SI %s
; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=fiji -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefixes=VI %s
; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX11 %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn -mcpu=gfx1200 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX12 %s
declare half @llvm.trunc.f16(half %a)
declare <2 x half> @llvm.trunc.v2f16(<2 x half> %a)
@@ -62,6 +63,24 @@ define amdgpu_kernel void @trunc_f16(
; GFX11-NEXT: v_trunc_f16_e32 v0, v0
; GFX11-NEXT: buffer_store_b16 v0, off, s[4:7], 0
; GFX11-NEXT: s_endpgm
+;
+; GFX12-LABEL: trunc_f16:
+; GFX12: ; %bb.0: ; %entry
+; GFX12-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
+; GFX12-NEXT: s_mov_b32 s6, -1
+; GFX12-NEXT: s_mov_b32 s7, 0x31016000
+; GFX12-NEXT: s_mov_b32 s10, s6
+; GFX12-NEXT: s_mov_b32 s11, s7
+; GFX12-NEXT: s_wait_kmcnt 0x0
+; GFX12-NEXT: s_mov_b32 s8, s2
+; GFX12-NEXT: s_mov_b32 s9, s3
+; GFX12-NEXT: s_mov_b32 s4, s0
+; GFX12-NEXT: buffer_load_u16 v0, off, s[8:11], null
+; GFX12-NEXT: s_mov_b32 s5, s1
+; GFX12-NEXT: s_wait_loadcnt 0x0
+; GFX12-NEXT: v_trunc_f16_e32 v0, v0
+; GFX12-NEXT: buffer_store_b16 v0, off, s[4:7], null
+; GFX12-NEXT: s_endpgm
ptr addrspace(1) %r,
ptr addrspace(1) %a) {
entry:
@@ -147,6 +166,28 @@ define amdgpu_kernel void @trunc_v2f16(
; GFX11-NEXT: v_pack_b32_f16 v0, v0, v1
; GFX11-NEXT: buffer_store_b32 v0, off, s[4:7], 0
; GFX11-NEXT: s_endpgm
+;
+; GFX12-LABEL: trunc_v2f16:
+; GFX12: ; %bb.0: ; %entry
+; GFX12-NEXT: s_load_b128 s[0:3], s[4:5], 0x24
+; GFX12-NEXT: s_mov_b32 s6, -1
+; GFX12-NEXT: s_mov_b32 s7, 0x31016000
+; GFX12-NEXT: s_mov_b32 s10, s6
+; GFX12-NEXT: s_mov_b32 s11, s7
+; GFX12-NEXT: s_wait_kmcnt 0x0
+; GFX12-NEXT: s_mov_b32 s8, s2
+; GFX12-NEXT: s_mov_b32 s9, s3
+; GFX12-NEXT: s_mov_b32 s4, s0
+; GFX12-NEXT: buffer_load_b32 v0, off, s[8:11], null
+; GFX12-NEXT: s_mov_b32 s5, s1
+; GFX12-NEXT: s_wait_loadcnt 0x0
+; GFX12-NEXT: v_lshrrev_b32_e32 v1, 16, v0
+; GFX12-NEXT: v_trunc_f16_e32 v0, v0
+; GFX12-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1)
+; GFX12-NEXT: v_trunc_f16_e32 v1, v1
+; GFX12-NEXT: v_pack_b32_f16 v0, v0, v1
+; GFX12-NEXT: buffer_store_b32 v0, off, s[4:7], null
+; GFX12-NEXT: s_endpgm
ptr addrspace(1) %r,
ptr addrspace(1) %a) {
entry: