Skip to content

[AMDGPU][True16][MC][CodeGen] true16 for v_alignbyte_b32 #125706

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 5 commits into from
Feb 7, 2025
Merged

[AMDGPU][True16][MC][CodeGen] true16 for v_alignbyte_b32 #125706

merged 5 commits into from
Feb 7, 2025

Conversation

broxigarchen
Copy link
Contributor

Support true16 format for v_alignbyte_b32 in MC and CodeGen

@llvmbot llvmbot added backend:AMDGPU mc Machine (object) code labels Feb 4, 2025
@llvmbot
Copy link
Member

llvmbot commented Feb 4, 2025 •
edited
Loading

@llvm/pr-subscribers-backend-amdgpu

@llvm/pr-subscribers-mc

Author: Brox Chen (broxigarchen)

Changes

Support true16 format for v_alignbyte_b32 in MC and CodeGen

Patch is 50.54 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/125706.diff

14 Files Affected:

  • (modified) llvm/lib/Target/AMDGPU/VOP3Instructions.td (+26-1)
  • (modified) llvm/test/CodeGen/AMDGPU/llvm.amdgcn.alignbyte.ll (+116-2)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3.s (+7-4)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s (+30-12)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp8.s (+13-4)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3.s (+3)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3_dpp16.s (+3)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3_dpp8.s (+3)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3.txt (+14-2)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3_dpp16.txt (+26-5)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3_dpp8.txt (+14-2)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3.txt (+14-2)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3_dpp16.txt (+30-6)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3_dpp8.txt (+18-3)
diff --git a/llvm/lib/Target/AMDGPU/VOP3Instructions.td b/llvm/lib/Target/AMDGPU/VOP3Instructions.td
index ce73e0ca361d9b..afafc2ecccfafe 100644
--- a/llvm/lib/Target/AMDGPU/VOP3Instructions.td
+++ b/llvm/lib/Target/AMDGPU/VOP3Instructions.td
@@ -212,7 +212,13 @@ defm V_BFE_U32 : VOP3Inst <"v_bfe_u32", VOP3_Profile<VOP_I32_I32_I32_I32>, AMDGP
 defm V_BFE_I32 : VOP3Inst <"v_bfe_i32", VOP3_Profile<VOP_I32_I32_I32_I32>, AMDGPUbfe_i32>;
 defm V_BFI_B32 : VOP3Inst <"v_bfi_b32", VOP3_Profile<VOP_I32_I32_I32_I32>, AMDGPUbfi>;
 defm V_ALIGNBIT_B32 : VOP3Inst <"v_alignbit_b32", VOP3_Profile<VOP_I32_I32_I32_I32>, fshr>;
+
+let True16Predicate = NotHasTrue16BitInsts in
 defm V_ALIGNBYTE_B32 : VOP3Inst <"v_alignbyte_b32", VOP3_Profile<VOP_I32_I32_I32_I32>, int_amdgcn_alignbyte>;
+let True16Predicate = UseRealTrue16Insts in
+defm V_ALIGNBYTE_B32_t16 : VOP3Inst <"v_alignbyte_b32_t16", VOP3_Profile_True16<VOP_I32_I32_I32_I16, VOP3_OPSEL>>;
+let True16Predicate = UseFakeTrue16Insts in 
+defm V_ALIGNBYTE_B32_fake16 : VOP3Inst <"v_alignbyte_b32_fake16", VOP3_Profile_Fake16<VOP_I32_I32_I32_I16, VOP3_OPSEL>>;
 
 // XXX - No FPException seems suspect but manual doesn't say it does
 let mayRaiseFPException = 0 in {
@@ -250,6 +256,25 @@ let SchedRW = [WriteDoubleAdd], FPDPRounding = 1 in {
 } // End SchedRW = [WriteDoubleAdd], FPDPRounding = 1
 } // End isReMaterializable = 1
 
+let True16Predicate = UseFakeTrue16Insts in
+def : GCNPat <
+(i32 (int_amdgcn_alignbyte (i32 (VOP3OpSelMods i32:$src0, i32:$src0_modifiers)),
+                           (i32 (VOP3OpSelMods i32:$src1, i32:$src1_modifiers)),
+                           (i32 (VOP3OpSelMods i32:$src2, i32:$src2_modifiers)))),
+(V_ALIGNBYTE_B32_fake16_e64 i32:$src0_modifiers, VSrc_b32:$src0,
+                            i32:$src1_modifiers, VSrc_b32:$src1,
+                            i32:$src2_modifiers, VGPR_32:$src2)
+>;
+
+let True16Predicate = UseRealTrue16Insts in
+def : GCNPat <
+(i32 (int_amdgcn_alignbyte (i32 (VOP3OpSelMods i32:$src0, i32:$src0_modifiers)),
+                           (i32 (VOP3OpSelMods i32:$src1, i32:$src1_modifiers)),
+                           (i32 (VOP3OpSelMods i32:$src2, i32:$src2_modifiers)))),
+(V_ALIGNBYTE_B32_t16_e64 i32:$src0_modifiers, VSrc_b32:$src0,
+                         i32:$src1_modifiers, VSrc_b32:$src1,
+                         i32:$src2_modifiers, (i16 (EXTRACT_SUBREG VGPR_32:$src2, lo16)))
+>;
 
 let mayRaiseFPException = 0 in { // Seems suspicious but manual doesn't say it does.
   let SchedRW = [WriteFloatFMA, WriteSALU] in
@@ -1690,7 +1715,7 @@ defm V_FMA_F32             : VOP3_Realtriple_gfx11_gfx12<0x213>;
 defm V_FMA_F64             : VOP3_Real_Base_gfx11_gfx12<0x214>;
 defm V_LERP_U8             : VOP3_Realtriple_gfx11_gfx12<0x215>;
 defm V_ALIGNBIT_B32        : VOP3_Realtriple_gfx11_gfx12<0x216>;
-defm V_ALIGNBYTE_B32       : VOP3_Realtriple_gfx11_gfx12<0x217>;
+defm V_ALIGNBYTE_B32       : VOP3_Realtriple_t16_and_fake16_gfx11_gfx12<0x217, "v_alignbyte_b32">;
 defm V_MULLIT_F32          : VOP3_Realtriple_gfx11_gfx12<0x218>;
 defm V_MIN3_F32            : VOP3_Realtriple_gfx11<0x219>;
 defm V_MIN3_I32            : VOP3_Realtriple_gfx11_gfx12<0x21a>;
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.alignbyte.ll b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.alignbyte.ll
index 8b16fef915a79d..07421afde7622d 100644
--- a/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.alignbyte.ll
+++ b/llvm/test/CodeGen/AMDGPU/llvm.amdgcn.alignbyte.ll
@@ -1,14 +1,128 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
 ; RUN: llc -mtriple=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -mattr=+real-true16 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX11-TRUE16 %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -mattr=-real-true16 -verify-machineinstrs < %s | FileCheck -check-prefixes=GFX11-FAKE16 %s
 
 declare i32 @llvm.amdgcn.alignbyte(i32, i32, i32) #0
 
-; GCN-LABEL: {{^}}v_alignbyte_b32:
-; GCN: v_alignbyte_b32 {{[vs][0-9]+}}, {{[vs][0-9]+}}, {{[vs][0-9]+}}
 define amdgpu_kernel void @v_alignbyte_b32(ptr addrspace(1) %out, i32 %src1, i32 %src2, i32 %src3) #1 {
+; GCN-LABEL: v_alignbyte_b32:
+; GCN:       ; %bb.0:
+; GCN-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0xb
+; GCN-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0x9
+; GCN-NEXT:    s_mov_b32 s7, 0xf000
+; GCN-NEXT:    s_mov_b32 s6, -1
+; GCN-NEXT:    s_waitcnt lgkmcnt(0)
+; GCN-NEXT:    v_mov_b32_e32 v0, s1
+; GCN-NEXT:    v_mov_b32_e32 v1, s2
+; GCN-NEXT:    v_alignbyte_b32 v0, s0, v0, v1
+; GCN-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; GCN-NEXT:    s_endpgm
+;
+; GFX11-TRUE16-LABEL: v_alignbyte_b32:
+; GFX11-TRUE16:       ; %bb.0:
+; GFX11-TRUE16-NEXT:    s_clause 0x1
+; GFX11-TRUE16-NEXT:    s_load_b128 s[0:3], s[4:5], 0x2c
+; GFX11-TRUE16-NEXT:    s_load_b64 s[4:5], s[4:5], 0x24
+; GFX11-TRUE16-NEXT:    v_mov_b32_e32 v1, 0
+; GFX11-TRUE16-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-TRUE16-NEXT:    v_mov_b16_e32 v0.l, s2
+; GFX11-TRUE16-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX11-TRUE16-NEXT:    v_alignbyte_b32 v0, s0, s1, v0.l
+; GFX11-TRUE16-NEXT:    global_store_b32 v1, v0, s[4:5]
+; GFX11-TRUE16-NEXT:    s_endpgm
+;
+; GFX11-FAKE16-LABEL: v_alignbyte_b32:
+; GFX11-FAKE16:       ; %bb.0:
+; GFX11-FAKE16-NEXT:    s_clause 0x1
+; GFX11-FAKE16-NEXT:    s_load_b128 s[0:3], s[4:5], 0x2c
+; GFX11-FAKE16-NEXT:    s_load_b64 s[4:5], s[4:5], 0x24
+; GFX11-FAKE16-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-FAKE16-NEXT:    v_dual_mov_b32 v1, 0 :: v_dual_mov_b32 v0, s2
+; GFX11-FAKE16-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX11-FAKE16-NEXT:    v_alignbyte_b32 v0, s0, s1, v0
+; GFX11-FAKE16-NEXT:    global_store_b32 v1, v0, s[4:5]
+; GFX11-FAKE16-NEXT:    s_endpgm
   %val = call i32 @llvm.amdgcn.alignbyte(i32 %src1, i32 %src2, i32 %src3) #0
   store i32 %val, ptr addrspace(1) %out
   ret void
 }
 
+define amdgpu_kernel void @v_alignbyte_b32_2(ptr addrspace(1) %out, ptr addrspace(1) %src1, ptr addrspace(1) %src2, i32 %src3) #1 {
+; GCN-LABEL: v_alignbyte_b32_2:
+; GCN:       ; %bb.0:
+; GCN-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
+; GCN-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0xd
+; GCN-NEXT:    s_load_dword s16, s[4:5], 0xf
+; GCN-NEXT:    s_mov_b32 s7, 0xf000
+; GCN-NEXT:    s_mov_b32 s14, 0
+; GCN-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; GCN-NEXT:    v_mov_b32_e32 v1, 0
+; GCN-NEXT:    s_mov_b32 s15, s7
+; GCN-NEXT:    s_mov_b64 s[10:11], s[14:15]
+; GCN-NEXT:    s_waitcnt lgkmcnt(0)
+; GCN-NEXT:    s_mov_b64 s[12:13], s[2:3]
+; GCN-NEXT:    buffer_load_dword v2, v[0:1], s[12:15], 0 addr64 glc
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    buffer_load_dword v0, v[0:1], s[8:11], 0 addr64 glc
+; GCN-NEXT:    s_waitcnt vmcnt(0)
+; GCN-NEXT:    s_mov_b32 s6, -1
+; GCN-NEXT:    s_mov_b32 s4, s0
+; GCN-NEXT:    s_mov_b32 s5, s1
+; GCN-NEXT:    v_alignbyte_b32 v0, v2, v0, s16
+; GCN-NEXT:    buffer_store_dword v0, off, s[4:7], 0
+; GCN-NEXT:    s_endpgm
+;
+; GFX11-TRUE16-LABEL: v_alignbyte_b32_2:
+; GFX11-TRUE16:       ; %bb.0:
+; GFX11-TRUE16-NEXT:    s_clause 0x1
+; GFX11-TRUE16-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX11-TRUE16-NEXT:    s_load_b64 s[6:7], s[4:5], 0x34
+; GFX11-TRUE16-NEXT:    v_dual_mov_b32 v3, 0 :: v_dual_and_b32 v0, 0x3ff, v0
+; GFX11-TRUE16-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX11-TRUE16-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; GFX11-TRUE16-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-TRUE16-NEXT:    global_load_b32 v1, v0, s[2:3] glc dlc
+; GFX11-TRUE16-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-TRUE16-NEXT:    global_load_b32 v2, v0, s[6:7] glc dlc
+; GFX11-TRUE16-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-TRUE16-NEXT:    s_load_b32 s2, s[4:5], 0x3c
+; GFX11-TRUE16-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-TRUE16-NEXT:    v_mov_b16_e32 v0.l, s2
+; GFX11-TRUE16-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX11-TRUE16-NEXT:    v_alignbyte_b32 v0, v1, v2, v0.l
+; GFX11-TRUE16-NEXT:    global_store_b32 v3, v0, s[0:1]
+; GFX11-TRUE16-NEXT:    s_endpgm
+;
+; GFX11-FAKE16-LABEL: v_alignbyte_b32_2:
+; GFX11-FAKE16:       ; %bb.0:
+; GFX11-FAKE16-NEXT:    s_clause 0x1
+; GFX11-FAKE16-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX11-FAKE16-NEXT:    s_load_b64 s[6:7], s[4:5], 0x34
+; GFX11-FAKE16-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX11-FAKE16-NEXT:    v_mov_b32_e32 v2, 0
+; GFX11-FAKE16-NEXT:    s_delay_alu instid0(VALU_DEP_2)
+; GFX11-FAKE16-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; GFX11-FAKE16-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-FAKE16-NEXT:    global_load_b32 v1, v0, s[2:3] glc dlc
+; GFX11-FAKE16-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-FAKE16-NEXT:    global_load_b32 v0, v0, s[6:7] glc dlc
+; GFX11-FAKE16-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-FAKE16-NEXT:    s_load_b32 s2, s[4:5], 0x3c
+; GFX11-FAKE16-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-FAKE16-NEXT:    v_alignbyte_b32 v0, v1, v0, s2
+; GFX11-FAKE16-NEXT:    global_store_b32 v2, v0, s[0:1]
+; GFX11-FAKE16-NEXT:    s_endpgm
+  %tid = call i32 @llvm.amdgcn.workitem.id.x()
+  %a.gep = getelementptr inbounds i32, ptr addrspace(1) %src1, i32 %tid
+  %b.gep = getelementptr inbounds i32, ptr addrspace(1) %src2, i32 %tid
+  %a.val = load volatile i32, ptr addrspace(1) %a.gep
+  %b.val = load volatile i32, ptr addrspace(1) %b.gep
+
+  %val = call i32 @llvm.amdgcn.alignbyte(i32 %a.val, i32 %b.val, i32 %src3) #0
+  store i32 %val, ptr addrspace(1) %out
+  ret void
+}
+
 attributes #0 = { nounwind readnone }
 attributes #1 = { nounwind }
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop3.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop3.s
index e55fbfc6e18c8c..857a1359b00d99 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop3.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop3.s
@@ -461,11 +461,11 @@ v_alignbyte_b32 v5, s1, v255, s3
 v_alignbyte_b32 v5, s105, s105, s105
 // GFX11: v_alignbyte_b32 v5, s105, s105, s105    ; encoding: [0x05,0x00,0x17,0xd6,0x69,0xd2,0xa4,0x01]
 
-v_alignbyte_b32 v5, vcc_lo, ttmp15, v3
-// GFX11: v_alignbyte_b32 v5, vcc_lo, ttmp15, v3  ; encoding: [0x05,0x00,0x17,0xd6,0x6a,0xf6,0x0c,0x04]
+v_alignbyte_b32 v5, vcc_lo, ttmp15, v3.l
+// GFX11: v_alignbyte_b32 v5, vcc_lo, ttmp15, v3.l ; encoding: [0x05,0x00,0x17,0xd6,0x6a,0xf6,0x0c,0x04]
 
-v_alignbyte_b32 v5, vcc_hi, 0xaf123456, v255
-// GFX11: v_alignbyte_b32 v5, vcc_hi, 0xaf123456, v255 ; encoding: [0x05,0x00,0x17,0xd6,0x6b,0xfe,0xfd,0x07,0x56,0x34,0x12,0xaf]
+v_alignbyte_b32 v5, vcc_hi, 0xaf123456, v255.l
+// GFX11: v_alignbyte_b32 v5, vcc_hi, 0xaf123456, v255.l ; encoding: [0x05,0x00,0x17,0xd6,0x6b,0xfe,0xfd,0x07,0x56,0x34,0x12,0xaf]
 
 v_alignbyte_b32 v5, ttmp15, src_scc, ttmp15
 // GFX11: v_alignbyte_b32 v5, ttmp15, src_scc, ttmp15 ; encoding: [0x05,0x00,0x17,0xd6,0x7b,0xfa,0xed,0x01]
@@ -494,6 +494,9 @@ v_alignbyte_b32 v5, src_scc, vcc_lo, -1
 v_alignbyte_b32 v255, 0xaf123456, vcc_hi, null
 // GFX11: v_alignbyte_b32 v255, 0xaf123456, vcc_hi, null ; encoding: [0xff,0x00,0x17,0xd6,0xff,0xd6,0xf0,0x01,0x56,0x34,0x12,0xaf]
 
+v_alignbyte_b32 v5, vcc_hi, 0xaf123456, v255.h
+// GFX11: v_alignbyte_b32 v5, vcc_hi, 0xaf123456, v255.h op_sel:[0,0,1,0] ; encoding: [0x05,0x20,0x17,0xd6,0x6b,0xfe,0xfd,0x07,0x56,0x34,0x12,0xaf]
+
 v_and_b16 v5.l, v1.l, v2.l
 // GFX11: v_and_b16 v5.l, v1.l, v2.l              ; encoding: [0x05,0x00,0x62,0xd7,0x01,0x05,0x02,0x00]
 
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s
index 3a3f6c178cbde2..1864996b260284 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s
@@ -363,22 +363,22 @@ v_alignbit_b32_e64_dpp v5, v1, v2, -1 row_xmask:0 row_mask:0x1 bank_mask:0x3 bou
 v_alignbit_b32_e64_dpp v255, v255, v255, src_scc row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
 // GFX11: v_alignbit_b32_e64_dpp v255, v255, v255, src_scc row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xff,0x00,0x16,0xd6,0xfa,0xfe,0xf7,0x03,0xff,0x6f,0x05,0x30]
 
-v_alignbyte_b32_e64_dpp v5, v1, v2, v3 quad_perm:[3,2,1,0]
-// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v3 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x1b,0x00,0xff]
+v_alignbyte_b32_e64_dpp v5, v1, v2, v3.l quad_perm:[3,2,1,0]
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v3.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x1b,0x00,0xff]
 
-v_alignbyte_b32_e64_dpp v5, v1, v2, v3 quad_perm:[0,1,2,3]
-// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v3 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0x0e,0x04,0x01,0xe4,0x00,0xff]
+v_alignbyte_b32_e64_dpp v5, v1, v2, v3.l quad_perm:[0,1,2,3]
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v3.l quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0x0e,0x04,0x01,0xe4,0x00,0xff]
 
-v_alignbyte_b32_e64_dpp v5, v1, v2, v3 row_mirror
-// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v3 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x40,0x01,0xff]
+v_alignbyte_b32_e64_dpp v5, v1, v2, v3.l row_mirror row_mask:0xf bank_mask:0xf
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v3.l row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x40,0x01,0xff]
 
-v_alignbyte_b32_e64_dpp v5, v1, v2, v3 row_half_mirror
-// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v3 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x41,0x01,0xff]
+v_alignbyte_b32_e64_dpp v5, v1, v2, v3.l row_half_mirror row_mask:0xf bank_mask:0xf
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v3.l row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x41,0x01,0xff]
 
-v_alignbyte_b32_e64_dpp v5, v1, v2, v255 row_shl:1
-// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v255 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0xfe,0x07,0x01,0x01,0x01,0xff]
+v_alignbyte_b32_e64_dpp v5, v1, v2, v255.l row_shl:1 row_mask:0xf bank_mask:0xf
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v255.l row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0xfe,0x07,0x01,0x01,0x01,0xff]
 
-v_alignbyte_b32_e64_dpp v5, v1, v2, s105 row_shl:15
+v_alignbyte_b32_e64_dpp v5, v1, v2, s105 row_shl:15 row_mask:0xf bank_mask:0xf
 // GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, s105 row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0xa6,0x01,0x01,0x0f,0x01,0xff]
 
 v_alignbyte_b32_e64_dpp v5, v1, v2, vcc_hi row_shr:1
@@ -387,7 +387,7 @@ v_alignbyte_b32_e64_dpp v5, v1, v2, vcc_hi row_shr:1
 v_alignbyte_b32_e64_dpp v5, v1, v2, vcc_lo row_shr:15
 // GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, vcc_lo row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0xaa,0x01,0x01,0x1f,0x01,0xff]
 
-v_alignbyte_b32_e64_dpp v5, v1, v2, ttmp15 row_ror:1
+v_alignbyte_b32_e64_dpp v5, v1, v2, ttmp15 row_ror:1 row_mask:0xf bank_mask:0xf
 // GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, ttmp15 row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0xee,0x01,0x01,0x21,0x01,0xff]
 
 v_alignbyte_b32_e64_dpp v5, v1, v2, exec_hi row_ror:15
@@ -405,6 +405,24 @@ v_alignbyte_b32_e64_dpp v5, v1, v2, -1 row_xmask:0 row_mask:0x1 bank_mask:0x3 bo
 v_alignbyte_b32_e64_dpp v255, v255, v255, src_scc row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
 // GFX11: v_alignbyte_b32_e64_dpp v255, v255, v255, src_scc row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xff,0x00,0x17,0xd6,0xfa,0xfe,0xf7,0x03,0xff,0x6f,0x05,0x30]
 
+v_alignbyte_b32_e64_dpp v5, v1, v2, v255.l row_mirror
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v255.l row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0xfe,0x07,0x01,0x40,0x01,0xff]
+
+v_alignbyte_b32_e64_dpp v5, v1, v2, s3 row_half_mirror
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, s3 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0x0e,0x00,0x01,0x41,0x01,0xff]
+
+v_alignbyte_b32_e64_dpp v5, v1, v2, s105 row_shl:1
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, s105 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0xa6,0x01,0x01,0x01,0x01,0xff]
+
+v_alignbyte_b32_e64_dpp v5, v1, v2, ttmp15 row_shl:15
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, ttmp15 row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0xee,0x01,0x01,0x0f,0x01,0xff]
+
+v_alignbyte_b32_e64_dpp v5, v1, v2, m0 row_ror:1
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, m0 row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x17,0xd6,0xfa,0x04,0xf6,0x01,0x01,0x21,0x01,0xff]
+
+v_alignbyte_b32_e64_dpp v5, v1, v2, v255.h row_mirror
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v255.h op_sel:[0,0,1,0] row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x20,0x17,0xd6,0xfa,0x04,0xfe,0x07,0x01,0x40,0x01,0xff]
+
 v_and_b16_e64_dpp v5.l, v1.l, v2.l quad_perm:[3,2,1,0]
 // GFX11: v_and_b16_e64_dpp v5.l, v1.l, v2.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x62,0xd7,0xfa,0x04,0x02,0x00,0x01,0x1b,0x00,0xff]
 
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp8.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp8.s
index 1f9173518d4128..300e5ef22f5aef 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp8.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp8.s
@@ -187,11 +187,11 @@ v_alignbit_b32_e64_dpp v5, v1, v2, -1 dpp8:[7,6,5,4,3,2,1,0] fi:1
 v_alignbit_b32_e64_dpp v255, v255, v255, src_scc dpp8:[0,0,0,0,0,0,0,0] fi:0
 // GFX11: v_alignbit_b32_e64_dpp v255, v255, v255, src_scc dpp8:[0,0,0,0,0,0,0,0] ; encoding: [0xff,0x00,0x16,0xd6,0xe9,0xfe,0xf7,0x03,0xff,0x00,0x00,0x00]
 
-v_alignbyte_b32_e64_dpp v5, v1, v2, v3 dpp8:[7,6,5,4,3,2,1,0]
-// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v3 dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0x05,0x00,0x17,0xd6,0xe9,0x04,0x0e,0x04,0x01,0x77,0x39,0x05]
+v_alignbyte_b32_e64_dpp v5, v1, v2, v3.l dpp8:[7,6,5,4,3,2,1,0]
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v3.l dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0x05,0x00,0x17,0xd6,0xe9,0x04,0x0e,0x04,0x01,0x77,0x39,0x05]
 
-v_alignbyte_b32_e64_dpp v5, v1, v2, v255 dpp8:[7,6,5,4,3,2,1,0]
-// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v255 dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0x05,0x00,0x17,0xd6,0xe9,0x04,0xfe,0x07,0x01,0x77,0x39,0x05]
+v_alignbyte_b32_e64_dpp v5, v1, v2, v255.l dpp8:[7,6,5,4,3,2,1,0]
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v255.l dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0x05,0x00,0x17,0xd6,0xe9,0x04,0xfe,0x07,0x01,0x77,0x39,0x05]
 
 v_alignbyte_b32_e64_dpp v5, v1, v2, s105 dpp8:[7,6,5,4,3,2,1,0]
 // GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, s105 dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0x05,0x00,0x17,0xd6,0xe9,0x04,0xa6,0x01,0x01,0x77,0x39,0x05]
@@ -220,6 +220,15 @@ v_alignbyte_b32_e64_dpp v5, v1, v2, -1 dpp8:[7,6,5,4,3,2,1,0] fi:1
 v_alignbyte_b32_e64_dpp v255, v255, v255, src_scc dpp8:[0,0,0,0,0,0,0,0] fi:0
 // GFX11: v_alignbyte_b32_e64_dpp v255, v255, v255, src_scc dpp8:[0,0,0,0,0,0,0,0] ; encoding: [0xff,0x00,0x17,0xd6,0xe9,0xfe,0xf7,0x03,0xff,0x00,0x00,0x00]
 
+v_alignbyte_b32_e64_dpp v5, v1, v2, s3 dpp8:[7,6,5,4,3,2,1,0]
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, s3 dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0x05,0x00,0x17,0xd6,0xe9,0x04,0x0e,0x00,0x01,0x77,0x39,0x05]
+
+v_alignbyte_b32_e64_dpp v5, v1, v2, m0 dpp8:[7,6,5,4,3,2,1,0]
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, m0 dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0x05,0x00,0x17,0xd6,0xe9,0x04,0xf6,0x01,0x01,0x77,0x39,0x05]
+
+v_alignbyte_b32_e64_dpp v5, v1, v2, v255.h dpp8:[7,6,5,4,3,2,1,0]
+// GFX11: v_alignbyte_b32_e64_dpp v5, v1, v2, v255.h op_sel:[0,0,1,0] dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0x05,0x20,0x17,0xd6,0xe9,0x04,0xfe,0x07,0x01,0x77,0x39,0x05]
+
 v_and_b16_e64_dpp v5.l, v1.l, v2.l dpp8:[7,6,5,4,3,2,1,0]
 // GFX11: v_and_b16_e64_dpp v5.l, v1.l, v2.l dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0x05,0x00,0x62,0xd7,0xe9,0x04,0x02,0x00,0x01,0x77,0x39,0x05]
 
diff --git a/llvm/test/MC/AMDGPU/gfx12_asm_vop3.s b/llvm/test/MC/AMDGPU/gfx12_asm_vop3.s
index 016ad18637bbb8..72d201e060df76 100644
--- a/llvm/test/MC/AMDGPU/gfx12_asm_vop3.s
+++ b/llvm/test/MC/AMDGPU/gfx12_asm_vop3.s
@@ -452,6 +452,9 @@ v_alignbyte_b32 v5, src_scc, vcc_lo, -1
 v_alignbyte_b32 v255, 0xaf123456, vcc_hi, null
 // GFX12: v_alignbyte_b32 v255, 0xaf123456, vcc_hi, null ; encoding: [0xff,0x0...
[truncated]

@broxigarchen
Copy link
Contributor Author

Hi @Sisyph @kosarev @arsenm The previous PR of this patch #119750 is stucked and the new commit is not able be displayed.

Closed and reopen it here.

@kosarev
Copy link
Collaborator

kosarev commented Feb 6, 2025

Same suggestion as in #119750 (comment).

kosarev
kosarev approved these changes Feb 6, 2025
View reviewed changes
Copy link
Collaborator

@kosarev kosarev left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM with a suggestion.

@broxigarchen
Copy link
Contributor Author

Same suggestion as in #119750 (comment).

I see. I think this is better to be done seperately

@broxigarchen broxigarchen merged commit 2b43543 into llvm:main Feb 7, 2025
11 checks passed
Icohedron pushed a commit to Icohedron/llvm-project that referenced this pull request Feb 11, 2025
@broxigarchen @Icohedron
[AMDGPU][True16][MC][CodeGen] true16 for v_alignbyte_b32 (llvm#125706)
e539366 
Support true16 format for v_alignbyte_b32 in MC and CodeGen
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@kosarev kosarev kosarev approved these changes

@arsenm arsenm Awaiting requested review from arsenm

@Sisyph Sisyph Awaiting requested review from Sisyph

Assignees
No one assigned
Labels
backend:AMDGPU mc Machine (object) code
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@broxigarchen @llvmbot @kosarev