[AMDGPU] Update strict floating point tests to be more comprehensive

[chinmaydd]

No description provided.

[llvmbot]

@llvm/pr-subscribers-backend-amdgpu

Author: Chinmay Deshpande (chinmaydd)

Changes

---

Patch is 128.85 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/169578.diff

9 Files Affected:

• (modified) llvm/test/CodeGen/AMDGPU/strict_fadd.f16.ll (+684-108)
• (modified) llvm/test/CodeGen/AMDGPU/strict_fadd.f32.ll (+577-17)
• (modified) llvm/test/CodeGen/AMDGPU/strict_fadd.f64.ll (+8-3)
• (modified) llvm/test/CodeGen/AMDGPU/strict_fma.f16.ll (+16-5)
• (modified) llvm/test/CodeGen/AMDGPU/strict_fmul.f16.ll (+200)
• (modified) llvm/test/CodeGen/AMDGPU/strict_fmul.f32.ll (+221)
• (modified) llvm/test/CodeGen/AMDGPU/strict_fmul.f64.ll (+4-2)
• (modified) llvm/test/CodeGen/AMDGPU/strict_fsub.f16.ll (+248)
• (modified) llvm/test/CodeGen/AMDGPU/strict_fsub.f64.ll (+4-2)

diff --git a/llvm/test/CodeGen/AMDGPU/strict_fadd.f16.ll b/llvm/test/CodeGen/AMDGPU/strict_fadd.f16.ll
index a094631267e64..da2c31b374cb7 100644
--- a/llvm/test/CodeGen/AMDGPU/strict_fadd.f16.ll
+++ b/llvm/test/CodeGen/AMDGPU/strict_fadd.f16.ll
@@ -1,9 +1,21 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 < %s | FileCheck -check-prefixes=GCN,GFX9 %s
-; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji < %s | FileCheck -check-prefixes=GCN,GFX8 %s
-; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 < %s | FileCheck -check-prefixes=GFX10PLUS,GFX10 %s
-; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -mattr=+real-true16 -amdgpu-enable-delay-alu=0 < %s | FileCheck -check-prefixes=GFX10PLUS,GFX11,GFX11-TRUE16 %s
-; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -mattr=-real-true16 -amdgpu-enable-delay-alu=0 < %s | FileCheck -check-prefixes=GFX10PLUS,GFX11,GFX11-FAKE16 %s
+; RUN: llc -global-isel=0 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 < %s | FileCheck -check-prefixes=GCN,GFX9,GFX9-SDAG %s
+; RUN: llc -global-isel=1 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 < %s | FileCheck -check-prefixes=GCN,GFX9,GFX9-GISEL %s
+
+; RUN: llc -global-isel=0 -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji < %s | FileCheck -check-prefixes=GCN,GFX8,GFX8-SDAG %s
+; RUN: llc -global-isel=1 -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji < %s | FileCheck -check-prefixes=GCN,GFX8,GFX8-GISEL %s
+
+; RUN: llc -global-isel=0 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 < %s | FileCheck -check-prefixes=GFX10-SDAG %s
+; RUN: llc -global-isel=1 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 < %s | FileCheck -check-prefixes=GFX10-GISEL %s
+
+; RUN: llc -global-isel=0 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -mattr=+real-true16 -amdgpu-enable-delay-alu=0 < %s | FileCheck -check-prefixes=GFX11-SDAG-TRUE16 %s
+; RUN: llc -global-isel=0 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -mattr=-real-true16 -amdgpu-enable-delay-alu=0 < %s | FileCheck -check-prefixes=GFX11-SDAG-FAKE16 %s
+; RUN: llc -global-isel=1 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -mattr=+real-true16 -amdgpu-enable-delay-alu=0 < %s | FileCheck -check-prefixes=GFX11-GISEL,GFX11-GISEL-TRUE16 %s
+; RUN: llc -global-isel=1 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -mattr=-real-true16 -amdgpu-enable-delay-alu=0 < %s | FileCheck -check-prefixes=GFX11-GISEL,GFX11-GISEL-FAKE16 %s
+
+; RUN: llc -global-isel=0 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1200 < %s | FileCheck -check-prefixes=GFX12-SDAG %s
+; RUN: llc -global-isel=1 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1200 < %s | FileCheck -check-prefixes=GFX12-GISEL %s
+
 ; FIXME: promotion not handled without f16 insts
 
 define half @v_constained_fadd_f16_fpexcept_strict(half %x, half %y) #0 {
@@ -13,18 +25,71 @@ define half @v_constained_fadd_f16_fpexcept_strict(half %x, half %y) #0 {
 ; GCN-NEXT:    v_add_f16_e32 v0, v0, v1
 ; GCN-NEXT:    s_setpc_b64 s[30:31]
 ;
+; GFX10-SDAG-LABEL: v_constained_fadd_f16_fpexcept_strict:
+; GFX10-SDAG:       ; %bb.0:
+; GFX10-SDAG-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX10-SDAG-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX10-SDAG-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX10-GISEL-LABEL: v_constained_fadd_f16_fpexcept_strict:
+; GFX10-GISEL:       ; %bb.0:
+; GFX10-GISEL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX10-GISEL-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX10-GISEL-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-SDAG-TRUE16-LABEL: v_constained_fadd_f16_fpexcept_strict:
+; GFX11-SDAG-TRUE16:       ; %bb.0:
+; GFX11-SDAG-TRUE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-SDAG-TRUE16-NEXT:    v_add_f16_e32 v0.l, v0.l, v1.l
+; GFX11-SDAG-TRUE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-SDAG-FAKE16-LABEL: v_constained_fadd_f16_fpexcept_strict:
+; GFX11-SDAG-FAKE16:       ; %bb.0:
+; GFX11-SDAG-FAKE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-SDAG-FAKE16-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX11-SDAG-FAKE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-GISEL-TRUE16-LABEL: v_constained_fadd_f16_fpexcept_strict:
+; GFX11-GISEL-TRUE16:       ; %bb.0:
+; GFX11-GISEL-TRUE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-GISEL-TRUE16-NEXT:    v_add_f16_e32 v0.l, v0.l, v1.l
+; GFX11-GISEL-TRUE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-GISEL-FAKE16-LABEL: v_constained_fadd_f16_fpexcept_strict:
+; GFX11-GISEL-FAKE16:       ; %bb.0:
+; GFX11-GISEL-FAKE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-GISEL-FAKE16-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX11-GISEL-FAKE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX12-SDAG-LABEL: v_constained_fadd_f16_fpexcept_strict:
+; GFX12-SDAG:       ; %bb.0:
+; GFX12-SDAG-NEXT:    s_wait_loadcnt_dscnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_expcnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_samplecnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_bvhcnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_kmcnt 0x0
+; GFX12-SDAG-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX12-SDAG-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX12-GISEL-LABEL: v_constained_fadd_f16_fpexcept_strict:
+; GFX12-GISEL:       ; %bb.0:
+; GFX12-GISEL-NEXT:    s_wait_loadcnt_dscnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_expcnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_samplecnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_bvhcnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_kmcnt 0x0
+; GFX12-GISEL-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX12-GISEL-NEXT:    s_setpc_b64 s[30:31]
 ; GFX10-LABEL: v_constained_fadd_f16_fpexcept_strict:
 ; GFX10:       ; %bb.0:
 ; GFX10-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX10-NEXT:    v_add_f16_e32 v0, v0, v1
 ; GFX10-NEXT:    s_setpc_b64 s[30:31]
-;
 ; GFX11-TRUE16-LABEL: v_constained_fadd_f16_fpexcept_strict:
 ; GFX11-TRUE16:       ; %bb.0:
 ; GFX11-TRUE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX11-TRUE16-NEXT:    v_add_f16_e32 v0.l, v0.l, v1.l
 ; GFX11-TRUE16-NEXT:    s_setpc_b64 s[30:31]
-;
 ; GFX11-FAKE16-LABEL: v_constained_fadd_f16_fpexcept_strict:
 ; GFX11-FAKE16:       ; %bb.0:
 ; GFX11-FAKE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -41,18 +106,71 @@ define half @v_constained_fadd_f16_fpexcept_ignore(half %x, half %y) #0 {
 ; GCN-NEXT:    v_add_f16_e32 v0, v0, v1
 ; GCN-NEXT:    s_setpc_b64 s[30:31]
 ;
+; GFX10-SDAG-LABEL: v_constained_fadd_f16_fpexcept_ignore:
+; GFX10-SDAG:       ; %bb.0:
+; GFX10-SDAG-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX10-SDAG-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX10-SDAG-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX10-GISEL-LABEL: v_constained_fadd_f16_fpexcept_ignore:
+; GFX10-GISEL:       ; %bb.0:
+; GFX10-GISEL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX10-GISEL-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX10-GISEL-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-SDAG-TRUE16-LABEL: v_constained_fadd_f16_fpexcept_ignore:
+; GFX11-SDAG-TRUE16:       ; %bb.0:
+; GFX11-SDAG-TRUE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-SDAG-TRUE16-NEXT:    v_add_f16_e32 v0.l, v0.l, v1.l
+; GFX11-SDAG-TRUE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-SDAG-FAKE16-LABEL: v_constained_fadd_f16_fpexcept_ignore:
+; GFX11-SDAG-FAKE16:       ; %bb.0:
+; GFX11-SDAG-FAKE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-SDAG-FAKE16-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX11-SDAG-FAKE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-GISEL-TRUE16-LABEL: v_constained_fadd_f16_fpexcept_ignore:
+; GFX11-GISEL-TRUE16:       ; %bb.0:
+; GFX11-GISEL-TRUE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-GISEL-TRUE16-NEXT:    v_add_f16_e32 v0.l, v0.l, v1.l
+; GFX11-GISEL-TRUE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-GISEL-FAKE16-LABEL: v_constained_fadd_f16_fpexcept_ignore:
+; GFX11-GISEL-FAKE16:       ; %bb.0:
+; GFX11-GISEL-FAKE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-GISEL-FAKE16-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX11-GISEL-FAKE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX12-SDAG-LABEL: v_constained_fadd_f16_fpexcept_ignore:
+; GFX12-SDAG:       ; %bb.0:
+; GFX12-SDAG-NEXT:    s_wait_loadcnt_dscnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_expcnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_samplecnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_bvhcnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_kmcnt 0x0
+; GFX12-SDAG-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX12-SDAG-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX12-GISEL-LABEL: v_constained_fadd_f16_fpexcept_ignore:
+; GFX12-GISEL:       ; %bb.0:
+; GFX12-GISEL-NEXT:    s_wait_loadcnt_dscnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_expcnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_samplecnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_bvhcnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_kmcnt 0x0
+; GFX12-GISEL-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX12-GISEL-NEXT:    s_setpc_b64 s[30:31]
 ; GFX10-LABEL: v_constained_fadd_f16_fpexcept_ignore:
 ; GFX10:       ; %bb.0:
 ; GFX10-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX10-NEXT:    v_add_f16_e32 v0, v0, v1
 ; GFX10-NEXT:    s_setpc_b64 s[30:31]
-;
 ; GFX11-TRUE16-LABEL: v_constained_fadd_f16_fpexcept_ignore:
 ; GFX11-TRUE16:       ; %bb.0:
 ; GFX11-TRUE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX11-TRUE16-NEXT:    v_add_f16_e32 v0.l, v0.l, v1.l
 ; GFX11-TRUE16-NEXT:    s_setpc_b64 s[30:31]
-;
 ; GFX11-FAKE16-LABEL: v_constained_fadd_f16_fpexcept_ignore:
 ; GFX11-FAKE16:       ; %bb.0:
 ; GFX11-FAKE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -69,18 +187,71 @@ define half @v_constained_fadd_f16_fpexcept_maytrap(half %x, half %y) #0 {
 ; GCN-NEXT:    v_add_f16_e32 v0, v0, v1
 ; GCN-NEXT:    s_setpc_b64 s[30:31]
 ;
+; GFX10-SDAG-LABEL: v_constained_fadd_f16_fpexcept_maytrap:
+; GFX10-SDAG:       ; %bb.0:
+; GFX10-SDAG-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX10-SDAG-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX10-SDAG-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX10-GISEL-LABEL: v_constained_fadd_f16_fpexcept_maytrap:
+; GFX10-GISEL:       ; %bb.0:
+; GFX10-GISEL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX10-GISEL-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX10-GISEL-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-SDAG-TRUE16-LABEL: v_constained_fadd_f16_fpexcept_maytrap:
+; GFX11-SDAG-TRUE16:       ; %bb.0:
+; GFX11-SDAG-TRUE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-SDAG-TRUE16-NEXT:    v_add_f16_e32 v0.l, v0.l, v1.l
+; GFX11-SDAG-TRUE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-SDAG-FAKE16-LABEL: v_constained_fadd_f16_fpexcept_maytrap:
+; GFX11-SDAG-FAKE16:       ; %bb.0:
+; GFX11-SDAG-FAKE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-SDAG-FAKE16-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX11-SDAG-FAKE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-GISEL-TRUE16-LABEL: v_constained_fadd_f16_fpexcept_maytrap:
+; GFX11-GISEL-TRUE16:       ; %bb.0:
+; GFX11-GISEL-TRUE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-GISEL-TRUE16-NEXT:    v_add_f16_e32 v0.l, v0.l, v1.l
+; GFX11-GISEL-TRUE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-GISEL-FAKE16-LABEL: v_constained_fadd_f16_fpexcept_maytrap:
+; GFX11-GISEL-FAKE16:       ; %bb.0:
+; GFX11-GISEL-FAKE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-GISEL-FAKE16-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX11-GISEL-FAKE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX12-SDAG-LABEL: v_constained_fadd_f16_fpexcept_maytrap:
+; GFX12-SDAG:       ; %bb.0:
+; GFX12-SDAG-NEXT:    s_wait_loadcnt_dscnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_expcnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_samplecnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_bvhcnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_kmcnt 0x0
+; GFX12-SDAG-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX12-SDAG-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX12-GISEL-LABEL: v_constained_fadd_f16_fpexcept_maytrap:
+; GFX12-GISEL:       ; %bb.0:
+; GFX12-GISEL-NEXT:    s_wait_loadcnt_dscnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_expcnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_samplecnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_bvhcnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_kmcnt 0x0
+; GFX12-GISEL-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX12-GISEL-NEXT:    s_setpc_b64 s[30:31]
 ; GFX10-LABEL: v_constained_fadd_f16_fpexcept_maytrap:
 ; GFX10:       ; %bb.0:
 ; GFX10-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX10-NEXT:    v_add_f16_e32 v0, v0, v1
 ; GFX10-NEXT:    s_setpc_b64 s[30:31]
-;
 ; GFX11-TRUE16-LABEL: v_constained_fadd_f16_fpexcept_maytrap:
 ; GFX11-TRUE16:       ; %bb.0:
 ; GFX11-TRUE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GFX11-TRUE16-NEXT:    v_add_f16_e32 v0.l, v0.l, v1.l
 ; GFX11-TRUE16-NEXT:    s_setpc_b64 s[30:31]
-;
 ; GFX11-FAKE16-LABEL: v_constained_fadd_f16_fpexcept_maytrap:
 ; GFX11-FAKE16:       ; %bb.0:
 ; GFX11-FAKE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
@@ -97,19 +268,71 @@ define <2 x half> @v_constained_fadd_v2f16_fpexcept_strict(<2 x half> %x, <2 x h
 ; GFX9-NEXT:    v_pk_add_f16 v0, v0, v1
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
 ;
-; GFX8-LABEL: v_constained_fadd_v2f16_fpexcept_strict:
-; GFX8:       ; %bb.0:
-; GFX8-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX8-NEXT:    v_add_f16_sdwa v2, v0, v1 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
-; GFX8-NEXT:    v_add_f16_e32 v0, v0, v1
-; GFX8-NEXT:    v_or_b32_e32 v0, v0, v2
-; GFX8-NEXT:    s_setpc_b64 s[30:31]
-;
-; GFX10PLUS-LABEL: v_constained_fadd_v2f16_fpexcept_strict:
-; GFX10PLUS:       ; %bb.0:
-; GFX10PLUS-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX10PLUS-NEXT:    v_pk_add_f16 v0, v0, v1
-; GFX10PLUS-NEXT:    s_setpc_b64 s[30:31]
+; GFX8-SDAG-LABEL: v_constained_fadd_v2f16_fpexcept_strict:
+; GFX8-SDAG:       ; %bb.0:
+; GFX8-SDAG-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX8-SDAG-NEXT:    v_add_f16_sdwa v2, v0, v1 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; GFX8-SDAG-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX8-SDAG-NEXT:    v_or_b32_e32 v0, v0, v2
+; GFX8-SDAG-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX8-GISEL-LABEL: v_constained_fadd_v2f16_fpexcept_strict:
+; GFX8-GISEL:       ; %bb.0:
+; GFX8-GISEL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX8-GISEL-NEXT:    v_add_f16_e32 v2, v0, v1
+; GFX8-GISEL-NEXT:    v_add_f16_sdwa v0, v0, v1 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; GFX8-GISEL-NEXT:    v_or_b32_e32 v0, v2, v0
+; GFX8-GISEL-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX10-SDAG-LABEL: v_constained_fadd_v2f16_fpexcept_strict:
+; GFX10-SDAG:       ; %bb.0:
+; GFX10-SDAG-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX10-SDAG-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX10-SDAG-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX10-GISEL-LABEL: v_constained_fadd_v2f16_fpexcept_strict:
+; GFX10-GISEL:       ; %bb.0:
+; GFX10-GISEL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX10-GISEL-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX10-GISEL-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-SDAG-TRUE16-LABEL: v_constained_fadd_v2f16_fpexcept_strict:
+; GFX11-SDAG-TRUE16:       ; %bb.0:
+; GFX11-SDAG-TRUE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-SDAG-TRUE16-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX11-SDAG-TRUE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-SDAG-FAKE16-LABEL: v_constained_fadd_v2f16_fpexcept_strict:
+; GFX11-SDAG-FAKE16:       ; %bb.0:
+; GFX11-SDAG-FAKE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-SDAG-FAKE16-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX11-SDAG-FAKE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-GISEL-LABEL: v_constained_fadd_v2f16_fpexcept_strict:
+; GFX11-GISEL:       ; %bb.0:
+; GFX11-GISEL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-GISEL-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX11-GISEL-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX12-SDAG-LABEL: v_constained_fadd_v2f16_fpexcept_strict:
+; GFX12-SDAG:       ; %bb.0:
+; GFX12-SDAG-NEXT:    s_wait_loadcnt_dscnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_expcnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_samplecnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_bvhcnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_kmcnt 0x0
+; GFX12-SDAG-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX12-SDAG-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX12-GISEL-LABEL: v_constained_fadd_v2f16_fpexcept_strict:
+; GFX12-GISEL:       ; %bb.0:
+; GFX12-GISEL-NEXT:    s_wait_loadcnt_dscnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_expcnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_samplecnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_bvhcnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_kmcnt 0x0
+; GFX12-GISEL-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX12-GISEL-NEXT:    s_setpc_b64 s[30:31]
   %val = call <2 x half> @llvm.experimental.constrained.fadd.v2f16(<2 x half> %x, <2 x half> %y, metadata !"round.tonearest", metadata !"fpexcept.strict")
   ret <2 x half> %val
 }
@@ -121,19 +344,71 @@ define <2 x half> @v_constained_fadd_v2f16_fpexcept_ignore(<2 x half> %x, <2 x h
 ; GFX9-NEXT:    v_pk_add_f16 v0, v0, v1
 ; GFX9-NEXT:    s_setpc_b64 s[30:31]
 ;
-; GFX8-LABEL: v_constained_fadd_v2f16_fpexcept_ignore:
-; GFX8:       ; %bb.0:
-; GFX8-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX8-NEXT:    v_add_f16_sdwa v2, v0, v1 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
-; GFX8-NEXT:    v_add_f16_e32 v0, v0, v1
-; GFX8-NEXT:    v_or_b32_e32 v0, v0, v2
-; GFX8-NEXT:    s_setpc_b64 s[30:31]
-;
-; GFX10PLUS-LABEL: v_constained_fadd_v2f16_fpexcept_ignore:
-; GFX10PLUS:       ; %bb.0:
-; GFX10PLUS-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX10PLUS-NEXT:    v_pk_add_f16 v0, v0, v1
-; GFX10PLUS-NEXT:    s_setpc_b64 s[30:31]
+; GFX8-SDAG-LABEL: v_constained_fadd_v2f16_fpexcept_ignore:
+; GFX8-SDAG:       ; %bb.0:
+; GFX8-SDAG-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX8-SDAG-NEXT:    v_add_f16_sdwa v2, v0, v1 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; GFX8-SDAG-NEXT:    v_add_f16_e32 v0, v0, v1
+; GFX8-SDAG-NEXT:    v_or_b32_e32 v0, v0, v2
+; GFX8-SDAG-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX8-GISEL-LABEL: v_constained_fadd_v2f16_fpexcept_ignore:
+; GFX8-GISEL:       ; %bb.0:
+; GFX8-GISEL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX8-GISEL-NEXT:    v_add_f16_e32 v2, v0, v1
+; GFX8-GISEL-NEXT:    v_add_f16_sdwa v0, v0, v1 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
+; GFX8-GISEL-NEXT:    v_or_b32_e32 v0, v2, v0
+; GFX8-GISEL-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX10-SDAG-LABEL: v_constained_fadd_v2f16_fpexcept_ignore:
+; GFX10-SDAG:       ; %bb.0:
+; GFX10-SDAG-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX10-SDAG-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX10-SDAG-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX10-GISEL-LABEL: v_constained_fadd_v2f16_fpexcept_ignore:
+; GFX10-GISEL:       ; %bb.0:
+; GFX10-GISEL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX10-GISEL-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX10-GISEL-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-SDAG-TRUE16-LABEL: v_constained_fadd_v2f16_fpexcept_ignore:
+; GFX11-SDAG-TRUE16:       ; %bb.0:
+; GFX11-SDAG-TRUE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-SDAG-TRUE16-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX11-SDAG-TRUE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-SDAG-FAKE16-LABEL: v_constained_fadd_v2f16_fpexcept_ignore:
+; GFX11-SDAG-FAKE16:       ; %bb.0:
+; GFX11-SDAG-FAKE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-SDAG-FAKE16-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX11-SDAG-FAKE16-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-GISEL-LABEL: v_constained_fadd_v2f16_fpexcept_ignore:
+; GFX11-GISEL:       ; %bb.0:
+; GFX11-GISEL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-GISEL-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX11-GISEL-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX12-SDAG-LABEL: v_constained_fadd_v2f16_fpexcept_ignore:
+; GFX12-SDAG:       ; %bb.0:
+; GFX12-SDAG-NEXT:    s_wait_loadcnt_dscnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_expcnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_samplecnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_bvhcnt 0x0
+; GFX12-SDAG-NEXT:    s_wait_kmcnt 0x0
+; GFX12-SDAG-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX12-SDAG-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX12-GISEL-LABEL: v_constained_fadd_v2f16_fpexcept_ignore:
+; GFX12-GISEL:       ; %bb.0:
+; GFX12-GISEL-NEXT:    s_wait_loadcnt_dscnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_expcnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_samplecnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_bvhcnt 0x0
+; GFX12-GISEL-NEXT:    s_wait_kmcnt 0x0
+; GFX12-GISEL-NEXT:    v_pk_add_f16 v0, v0, v1
+; GFX12-GISEL-NEXT:    s_setpc_b64 s[30:31]
   %val = call <2 x half> @llvm.experimental.constrained.fadd.v2f16(<2 x half> %x, <2 x half> %y, metadata !"round.tonearest", metadata !"fpexcept.ignore")
   ret <2 x...
[truncated]