From: Joe Nash <Joseph.Nash@amd.com>
Date: Mon, 13 Jun 2022 20:38:18 +0000 (-0400)
Subject: [AMDGPU] gfx11 support add_f16
X-Git-Tag: upstream/15.0.7~4803
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=989bd57f987977321f95d5414680d3f6970f4b36;p=platform%2Fupstream%2Fllvm.git

[AMDGPU] gfx11 support add_f16

The instruction was skipped in the earlier large patch adding
VOP2, https://reviews.llvm.org/D126917.

Reviewed By: rampitec, #amdgpu

Differential Revision: https://reviews.llvm.org/D127697
---

diff --git a/llvm/lib/Target/AMDGPU/VOP2Instructions.td b/llvm/lib/Target/AMDGPU/VOP2Instructions.td
index ea2e497..2b3444e 100644
--- a/llvm/lib/Target/AMDGPU/VOP2Instructions.td
+++ b/llvm/lib/Target/AMDGPU/VOP2Instructions.td
@@ -1548,7 +1548,7 @@ defm V_XNOR_B32        : VOP2_Real_gfx10_gfx11<0x01e>;
 defm V_FMAC_F32        : VOP2_Real_gfx10_gfx11<0x02b>;
 defm V_FMAMK_F32       : VOP2Only_Real_MADK_gfx10_gfx11<0x02c>;
 defm V_FMAAK_F32       : VOP2Only_Real_MADK_gfx10_gfx11<0x02d>;
-defm V_ADD_F16         : VOP2_Real_gfx10<0x032>;
+defm V_ADD_F16         : VOP2_Real_gfx10_gfx11<0x032>;
 defm V_SUB_F16         : VOP2_Real_gfx10_gfx11<0x033>;
 defm V_SUBREV_F16      : VOP2_Real_gfx10_gfx11<0x034>;
 defm V_MUL_F16         : VOP2_Real_gfx10_gfx11<0x035>;
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_dpp16.s b/llvm/test/MC/AMDGPU/gfx11_asm_dpp16.s
index 1519835..5783666 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_dpp16.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_dpp16.s
@@ -312,6 +312,9 @@ v_subrev_co_ci_u32 v5, vcc_lo, v1, v2, vcc_lo quad_perm:[3,2,1,0] row_mask:0x0 b
 v_fmac_f32 v5, v1, v2 quad_perm:[3,2,1,0] row_mask:0x0 bank_mask:0x0
 // GFX11: encoding: [0xfa,0x04,0x0a,0x56,0x01,0x1b,0x00,0x00]
 
+v_add_f16 v5, v1, v2 quad_perm:[3,2,1,0] row_mask:0x0 bank_mask:0x0
+// GFX11: encoding: [0xfa,0x04,0x0a,0x64,0x01,0x1b,0x00,0x00]
+
 v_sub_f16 v5, v1, v2 quad_perm:[3,2,1,0] row_mask:0x0 bank_mask:0x0
 // GFX11: encoding: [0xfa,0x04,0x0a,0x66,0x01,0x1b,0x00,0x00]
 
@@ -570,6 +573,9 @@ v_xnor_b32 v5, v1, v2 quad_perm:[3,2,1,0] row_mask:0x0 bank_mask:0x0 fi:1
 v_fmac_f32 v5, v1, v2 quad_perm:[3,2,1,0] row_mask:0x0 bank_mask:0x0 fi:1
 // GFX11: encoding: [0xfa,0x04,0x0a,0x56,0x01,0x1b,0x04,0x00]
 
+v_add_f16 v5, v1, v2 quad_perm:[3,2,1,0] row_mask:0x0 bank_mask:0x0 fi:1
+// GFX11: encoding: [0xfa,0x04,0x0a,0x64,0x01,0x1b,0x04,0x00]
+
 v_sub_f16 v5, v1, v2 quad_perm:[3,2,1,0] row_mask:0x0 bank_mask:0x0 fi:1
 // GFX11: encoding: [0xfa,0x04,0x0a,0x66,0x01,0x1b,0x04,0x00]
 
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_dpp8.s b/llvm/test/MC/AMDGPU/gfx11_asm_dpp8.s
index 8569828..4275113 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_dpp8.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_dpp8.s
@@ -228,6 +228,9 @@ v_xor_b32 v5, v1, v2 dpp8:[0,1,2,3,4,5,6,7]
 v_xnor_b32 v5, v1, v2 dpp8:[0,1,2,3,4,5,6,7]
 // GFX11: encoding: [0xe9,0x04,0x0a,0x3c,0x01,0x88,0xc6,0xfa]
 
+v_add_f16 v5, v1, v2 dpp8:[0,1,2,3,4,5,6,7]
+// GFX11: encoding: [0xe9,0x04,0x0a,0x64,0x01,0x88,0xc6,0xfa]
+
 v_sub_f16 v5, v1, v2 dpp8:[0,1,2,3,4,5,6,7]
 // GFX11: encoding: [0xe9,0x04,0x0a,0x66,0x01,0x88,0xc6,0xfa]
 
@@ -477,6 +480,9 @@ v_xor_b32 v5, v1, v2 dpp8:[0,1,2,3,4,5,6,7] fi:1
 v_xnor_b32 v5, v1, v2 dpp8:[0,1,2,3,4,5,6,7] fi:1
 // GFX11: encoding: [0xea,0x04,0x0a,0x3c,0x01,0x88,0xc6,0xfa]
 
+v_add_f16 v5, v1, v2 dpp8:[0,1,2,3,4,5,6,7] fi:1
+// GFX11: encoding: [0xea,0x04,0x0a,0x64,0x01,0x88,0xc6,0xfa]
+
 v_sub_f16 v5, v1, v2 dpp8:[0,1,2,3,4,5,6,7] fi:1
 // GFX11: encoding: [0xea,0x04,0x0a,0x66,0x01,0x88,0xc6,0xfa]
 
diff --git a/llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_all.txt b/llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_all.txt
index 9470a53..56b373a 100644
--- a/llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_all.txt
+++ b/llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_all.txt
@@ -46334,6 +46334,12 @@
 # W64: v_add_co_ci_u32_dpp v5, vcc, v1, v2, vcc quad_perm:[3,2,1,0] row_mask:0x0 bank_mask:0x0 ; encoding: [0xfa,0x04,0x0a,0x40,0x01,0x1b,0x00,0x00]
 0xfa,0x04,0x0a,0x40,0x01,0x1b,0x00,0x00
 
+# GFX11: v_add_f16_dpp v5, v1, v2 quad_perm:[3,2,1,0] row_mask:0x0 bank_mask:0x0 ; encoding: [0xfa,0x04,0x0a,0x64,0x01,0x1b,0x00,0x00]
+0xfa,0x04,0x0a,0x64,0x01,0x1b,0x00,0x00
+
+# GFX11: v_add_f16_dpp v5, v1, v2 quad_perm:[3,2,1,0] row_mask:0x0 bank_mask:0x0 fi:1 ; encoding: [0xfa,0x04,0x0a,0x64,0x01,0x1b,0x04,0x00]
+0xfa,0x04,0x0a,0x64,0x01,0x1b,0x04,0x00
+
 # GFX11: v_add_f32_dpp v5, v1, v2 quad_perm:[3,2,1,0] row_mask:0x0 bank_mask:0x0 ; encoding: [0xfa,0x04,0x0a,0x06,0x01,0x1b,0x00,0x00]
 0xfa,0x04,0x0a,0x06,0x01,0x1b,0x00,0x00
 
@@ -46951,6 +46957,12 @@
 # GFX11: v_xor_b32_dpp v5, v1, v2 quad_perm:[3,2,1,0] row_mask:0x0 bank_mask:0x0 fi:1 ; encoding: [0xfa,0x04,0x0a,0x3a,0x01,0x1b,0x04,0x00]
 0xfa,0x04,0x0a,0x3a,0x01,0x1b,0x04,0x00
 
+# GFX11: v_add_f16_dpp v5, v1, v2 dpp8:[0,1,2,3,4,5,6,7] ; encoding: [0xe9,0x04,0x0a,0x64,0x01,0x88,0xc6,0xfa]
+0xe9,0x04,0x0a,0x64,0x01,0x88,0xc6,0xfa
+
+# GFX11: v_add_f16_dpp v5, v1, v2 dpp8:[0,1,2,3,4,5,6,7] fi:1 ; encoding: [0xea,0x04,0x0a,0x64,0x01,0x88,0xc6,0xfa]
+0xea,0x04,0x0a,0x64,0x01,0x88,0xc6,0xfa
+
 # GFX11: v_add_f32_dpp v5, v1, v2 dpp8:[0,1,2,3,4,5,6,7] ; encoding: [0xe9,0x04,0x0a,0x06,0x01,0x88,0xc6,0xfa]
 0xe9,0x04,0x0a,0x06,0x01,0x88,0xc6,0xfa