[AMDGPU][True16][MC] true16 for v_div_fixup_f16 by broxigarchen · Pull Request #119613 · llvm/llvm-project

broxigarchen · 2024-12-11T20:54:59Z

Support true16 format for v_div_fixup_f16 in MC.

llvmbot · 2024-12-12T16:10:14Z

@llvm/pr-subscribers-backend-amdgpu

Author: Brox Chen (broxigarchen)

Changes

Support true16 format for v_div_fixup_f16 in MC.

Patch is 168.62 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/119613.diff

14 Files Affected:

(modified) llvm/lib/Target/AMDGPU/SIInstrInfo.cpp (+1)
(modified) llvm/lib/Target/AMDGPU/VOP3Instructions.td (+2-3)
(modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3.s (+56-32)
(modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s (+74-38)
(modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp8.s (+67-34)
(modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3.s (+42-30)
(modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3_dpp16.s (+57-42)
(modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3_dpp8.s (+53-38)
(modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3.txt (+88-16)
(modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3_dpp16.txt (+98-14)
(modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3_dpp8.txt (+98-14)
(modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3.txt (+84-15)
(modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3_dpp16.txt (+90-15)
(modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3_dpp8.txt (+90-15)

diff --git a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
index 4a94d690297949..05dd8e9cf530ee 100644
--- a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
@@ -9284,6 +9284,7 @@ static bool isRenamedInGFX9(int Opcode) {
     GENERATE_RENAMED_GFX9_CASES(AMDGPU::V_SUB_U32)
   //
   case AMDGPU::V_DIV_FIXUP_F16_gfx9_e64:
+  case AMDGPU::V_DIV_FIXUP_F16_gfx9_fake16_e64:
   case AMDGPU::V_FMA_F16_gfx9_e64:
   case AMDGPU::V_INTERP_P2_F16:
   case AMDGPU::V_MAD_F16_e64:
diff --git a/llvm/lib/Target/AMDGPU/VOP3Instructions.td b/llvm/lib/Target/AMDGPU/VOP3Instructions.td
index 8a9f8aa3d16d3a..9fd8bf2c8aa786 100644
--- a/llvm/lib/Target/AMDGPU/VOP3Instructions.td
+++ b/llvm/lib/Target/AMDGPU/VOP3Instructions.td
@@ -339,8 +339,7 @@ let FPDPRounding = 1 in {
   } // End Predicates = [Has16BitInsts, isGFX8Only]
 
   let SubtargetPredicate = isGFX9Plus in {
-    defm V_DIV_FIXUP_F16_gfx9 : VOP3Inst <"v_div_fixup_f16_gfx9",
-                                          VOP3_Profile<VOP_F16_F16_F16_F16, VOP3_OPSEL>, AMDGPUdiv_fixup>;
+    defm V_DIV_FIXUP_F16_gfx9 : VOP3Inst_t16 <"v_div_fixup_f16_gfx9", VOP_F16_F16_F16_F16, AMDGPUdiv_fixup>;
     defm V_FMA_F16_gfx9 : VOP3Inst <"v_fma_f16_gfx9", VOP3_Profile<VOP_F16_F16_F16_F16, VOP3_OPSEL>, any_fma>;
   } // End SubtargetPredicate = isGFX9Plus
 } // End FPDPRounding = 1
@@ -1717,7 +1716,7 @@ defm V_MED3_F16            : VOP3_Realtriple_gfx11<0x24f>;
 defm V_MED3_I16            : VOP3_Realtriple_gfx11_gfx12<0x250>;
 defm V_MED3_U16            : VOP3_Realtriple_gfx11_gfx12<0x251>;
 defm V_MAD_I16             : VOP3_Realtriple_with_name_gfx11_gfx12<0x253, "V_MAD_I16_gfx9", "v_mad_i16">;
-defm V_DIV_FIXUP_F16       : VOP3_Realtriple_with_name_gfx11_gfx12<0x254, "V_DIV_FIXUP_F16_gfx9", "v_div_fixup_f16">;
+defm V_DIV_FIXUP_F16       : VOP3_Realtriple_t16_and_fake16_gfx11_gfx12<0x254, "v_div_fixup_f16", "V_DIV_FIXUP_F16_gfx9">;
 defm V_ADD3_U32            : VOP3_Realtriple_gfx11_gfx12<0x255>;
 defm V_LSHL_OR_B32         : VOP3_Realtriple_gfx11_gfx12<0x256>;
 defm V_AND_OR_B32          : VOP3_Realtriple_gfx11_gfx12<0x257>;
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop3.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop3.s
index b649bab532f262..78d226a71025ed 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop3.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop3.s
@@ -1574,53 +1574,77 @@ v_cvt_pk_norm_u16_f32 v5, -src_scc, |vcc_lo|
 v_cvt_pk_norm_u16_f32 v255, -|0xaf123456|, -|vcc_hi|
 // GFX11: v_cvt_pk_norm_u16_f32 v255, -|0xaf123456|, -|vcc_hi| ; encoding: [0xff,0x03,0x22,0xd7,0xff,0xd6,0x00,0x60,0x56,0x34,0x12,0xaf]
 
-v_div_fixup_f16 v5, v1, v2, s3
-// GFX11: v_div_fixup_f16 v5, v1, v2, s3          ; encoding: [0x05,0x00,0x54,0xd6,0x01,0x05,0x0e,0x00]
+v_div_fixup_f16 v5.l, v1.l, v2.l, s3
+// GFX11: v_div_fixup_f16 v5.l, v1.l, v2.l, s3    ; encoding: [0x05,0x00,0x54,0xd6,0x01,0x05,0x0e,0x00]
 
-v_div_fixup_f16 v5, v255, s2, s105
-// GFX11: v_div_fixup_f16 v5, v255, s2, s105      ; encoding: [0x05,0x00,0x54,0xd6,0xff,0x05,0xa4,0x01]
+v_div_fixup_f16 v5.l, v255.l, s2, s105
+// GFX11: v_div_fixup_f16 v5.l, v255.l, s2, s105  ; encoding: [0x05,0x00,0x54,0xd6,0xff,0x05,0xa4,0x01]
 
-v_div_fixup_f16 v5, s1, v255, exec_hi
-// GFX11: v_div_fixup_f16 v5, s1, v255, exec_hi   ; encoding: [0x05,0x00,0x54,0xd6,0x01,0xfe,0xff,0x01]
+v_div_fixup_f16 v5.l, s1, v255.l, exec_hi
+// GFX11: v_div_fixup_f16 v5.l, s1, v255.l, exec_hi ; encoding: [0x05,0x00,0x54,0xd6,0x01,0xfe,0xff,0x01]
 
-v_div_fixup_f16 v5, s105, s105, exec_lo
-// GFX11: v_div_fixup_f16 v5, s105, s105, exec_lo ; encoding: [0x05,0x00,0x54,0xd6,0x69,0xd2,0xf8,0x01]
+v_div_fixup_f16 v5.l, s105, s105, exec_lo
+// GFX11: v_div_fixup_f16 v5.l, s105, s105, exec_lo ; encoding: [0x05,0x00,0x54,0xd6,0x69,0xd2,0xf8,0x01]
 
-v_div_fixup_f16 v5, vcc_lo, ttmp15, v3
-// GFX11: v_div_fixup_f16 v5, vcc_lo, ttmp15, v3  ; encoding: [0x05,0x00,0x54,0xd6,0x6a,0xf6,0x0c,0x04]
+v_div_fixup_f16 v5.l, vcc_lo, ttmp15, v3.l
+// GFX11: v_div_fixup_f16 v5.l, vcc_lo, ttmp15, v3.l ; encoding: [0x05,0x00,0x54,0xd6,0x6a,0xf6,0x0c,0x04]
 
-v_div_fixup_f16 v5, vcc_hi, 0xfe0b, v255
-// GFX11: v_div_fixup_f16 v5, vcc_hi, 0xfe0b, v255 ; encoding: [0x05,0x00,0x54,0xd6,0x6b,0xfe,0xfd,0x07,0x0b,0xfe,0x00,0x00]
+v_div_fixup_f16 v5.l, vcc_hi, 0xfe0b, v255.l
+// GFX11: v_div_fixup_f16 v5.l, vcc_hi, 0xfe0b, v255.l ; encoding: [0x05,0x00,0x54,0xd6,0x6b,0xfe,0xfd,0x07,0x0b,0xfe,0x00,0x00]
 
-v_div_fixup_f16 v5, -|ttmp15|, -|src_scc|, -|ttmp15|
-// GFX11: v_div_fixup_f16 v5, -|ttmp15|, -|src_scc|, -|ttmp15| ; encoding: [0x05,0x07,0x54,0xd6,0x7b,0xfa,0xed,0xe1]
+v_div_fixup_f16 v5.l, -|ttmp15|, -|src_scc|, -|ttmp15|
+// GFX11: v_div_fixup_f16 v5.l, -|ttmp15|, -|src_scc|, -|ttmp15| ; encoding: [0x05,0x07,0x54,0xd6,0x7b,0xfa,0xed,0xe1]
 
-v_div_fixup_f16 v5, m0, 0.5, m0
-// GFX11: v_div_fixup_f16 v5, m0, 0.5, m0         ; encoding: [0x05,0x00,0x54,0xd6,0x7d,0xe0,0xf5,0x01]
+v_div_fixup_f16 v5.l, m0, 0.5, m0
+// GFX11: v_div_fixup_f16 v5.l, m0, 0.5, m0       ; encoding: [0x05,0x00,0x54,0xd6,0x7d,0xe0,0xf5,0x01]
 
-v_div_fixup_f16 v5, |exec_lo|, -1, vcc_hi
-// GFX11: v_div_fixup_f16 v5, |exec_lo|, -1, vcc_hi ; encoding: [0x05,0x01,0x54,0xd6,0x7e,0x82,0xad,0x01]
+v_div_fixup_f16 v5.l, |exec_lo|, -1, vcc_hi
+// GFX11: v_div_fixup_f16 v5.l, |exec_lo|, -1, vcc_hi ; encoding: [0x05,0x01,0x54,0xd6,0x7e,0x82,0xad,0x01]
 
-v_div_fixup_f16 v5, -|exec_hi|, null, -|vcc_lo| op_sel:[1,1,1,1]
-// GFX11: v_div_fixup_f16 v5, -|exec_hi|, null, -|vcc_lo| op_sel:[1,1,1,1] ; encoding: [0x05,0x7d,0x54,0xd6,0x7f,0xf8,0xa8,0xa1]
+v_div_fixup_f16 v5.h, -|exec_hi|, null, -|vcc_lo| op_sel:[1,1,1,1]
+// GFX11: v_div_fixup_f16 v5.h, -|exec_hi|, null, -|vcc_lo| op_sel:[1,1,1,1] ; encoding: [0x05,0x7d,0x54,0xd6,0x7f,0xf8,0xa8,0xa1]
 
-v_div_fixup_f16 v5, null, exec_lo, -|0xfe0b| op_sel:[0,0,0,0]
-// GFX11: v_div_fixup_f16 v5, null, exec_lo, -|0xfe0b| ; encoding: [0x05,0x04,0x54,0xd6,0x7c,0xfc,0xfc,0x83,0x0b,0xfe,0x00,0x00]
+v_div_fixup_f16 v5.l, null, exec_lo, -|0xfe0b|
+// GFX11: v_div_fixup_f16 v5.l, null, exec_lo, -|0xfe0b| ; encoding: [0x05,0x04,0x54,0xd6,0x7c,0xfc,0xfc,0x83,0x0b,0xfe,0x00,0x00]
 
-v_div_fixup_f16 v5, -1, -|exec_hi|, -|src_scc| op_sel:[1,0,0,0]
-// GFX11: v_div_fixup_f16 v5, -1, -|exec_hi|, -|src_scc| op_sel:[1,0,0,0] ; encoding: [0x05,0x0e,0x54,0xd6,0xc1,0xfe,0xf4,0xc3]
+v_div_fixup_f16 v5.l, -1, -|exec_hi|, -|src_scc| op_sel:[1,0,0,0]
+// GFX11: v_div_fixup_f16 v5.l, -1, -|exec_hi|, -|src_scc| op_sel:[1,0,0,0] ; encoding: [0x05,0x0e,0x54,0xd6,0xc1,0xfe,0xf4,0xc3]
 
-v_div_fixup_f16 v5, 0.5, -m0, 0.5 op_sel:[0,1,0,0]
-// GFX11: v_div_fixup_f16 v5, 0.5, -m0, 0.5 op_sel:[0,1,0,0] ; encoding: [0x05,0x10,0x54,0xd6,0xf0,0xfa,0xc0,0x43]
+v_div_fixup_f16 v5.l, 0.5, -m0, 0.5 op_sel:[0,1,0,0]
+// GFX11: v_div_fixup_f16 v5.l, 0.5, -m0, 0.5 op_sel:[0,1,0,0] ; encoding: [0x05,0x10,0x54,0xd6,0xf0,0xfa,0xc0,0x43]
 
-v_div_fixup_f16 v5, -src_scc, |vcc_lo|, -1 op_sel:[0,0,1,0]
-// GFX11: v_div_fixup_f16 v5, -src_scc, |vcc_lo|, -1 op_sel:[0,0,1,0] ; encoding: [0x05,0x22,0x54,0xd6,0xfd,0xd4,0x04,0x23]
+v_div_fixup_f16 v5.l, -src_scc, |vcc_lo|, -1 op_sel:[0,0,1,0]
+// GFX11: v_div_fixup_f16 v5.l, -src_scc, |vcc_lo|, -1 op_sel:[0,0,1,0] ; encoding: [0x05,0x22,0x54,0xd6,0xfd,0xd4,0x04,0x23]
 
-v_div_fixup_f16 v255, -|0xfe0b|, -|vcc_hi|, null op_sel:[0,0,0,1] clamp
-// GFX11: v_div_fixup_f16 v255, -|0xfe0b|, -|vcc_hi|, null op_sel:[0,0,0,1] clamp ; encoding: [0xff,0xc3,0x54,0xd6,0xff,0xd6,0xf0,0x61,0x0b,0xfe,0x00,0x00]
+v_div_fixup_f16 v255.h, -|0xfe0b|, -|vcc_hi|, null clamp
+// GFX11: v_div_fixup_f16 v255.h, -|0xfe0b|, -|vcc_hi|, null op_sel:[0,0,0,1] clamp ; encoding: [0xff,0xc3,0x54,0xd6,0xff,0xd6,0xf0,0x61,0x0b,0xfe,0x00,0x00]
 
-v_div_fixup_f16 v5, 0.5, -m0, 0.5 op_sel:[0,1,0,0] mul:2
-// GFX11: v_div_fixup_f16 v5, 0.5, -m0, 0.5 op_sel:[0,1,0,0] mul:2 ; encoding: [0x05,0x10,0x54,0xd6,0xf0,0xfa,0xc0,0x4b]
+v_div_fixup_f16 v5.l, 0.5, -m0, 0.5 op_sel:[0,1,0,0] mul:2
+// GFX11: v_div_fixup_f16 v5.l, 0.5, -m0, 0.5 op_sel:[0,1,0,0] mul:2 ; encoding: [0x05,0x10,0x54,0xd6,0xf0,0xfa,0xc0,0x4b]
+
+v_div_fixup_f16 v5.l, v255.h, s2, s105
+// GFX11: v_div_fixup_f16 v5.l, v255.h, s2, s105 op_sel:[1,0,0,0] ; encoding: [0x05,0x08,0x54,0xd6,0xff,0x05,0xa4,0x01]
+
+v_div_fixup_f16 v5.l, s1, v255.h, exec_hi
+// GFX11: v_div_fixup_f16 v5.l, s1, v255.h, exec_hi op_sel:[0,1,0,0] ; encoding: [0x05,0x10,0x54,0xd6,0x01,0xfe,0xff,0x01]
+
+v_div_fixup_f16 v5.l, vcc_hi, 0xfe0b, v255.h
+// GFX11: v_div_fixup_f16 v5.l, vcc_hi, 0xfe0b, v255.h op_sel:[0,0,1,0] ; encoding: [0x05,0x20,0x54,0xd6,0x6b,0xfe,0xfd,0x07,0x0b,0xfe,0x00,0x00]
+
+v_div_fixup_f16 v5.l, -|exec_hi|, null, -|vcc_lo|
+// GFX11: v_div_fixup_f16 v5.l, -|exec_hi|, null, -|vcc_lo| ; encoding: [0x05,0x05,0x54,0xd6,0x7f,0xf8,0xa8,0xa1]
+
+v_div_fixup_f16 v5.l, -1, -|exec_hi|, -|src_scc|
+// GFX11: v_div_fixup_f16 v5.l, -1, -|exec_hi|, -|src_scc| ; encoding: [0x05,0x06,0x54,0xd6,0xc1,0xfe,0xf4,0xc3]
+
+v_div_fixup_f16 v5.l, 0.5, -m0, 0.5
+// GFX11: v_div_fixup_f16 v5.l, 0.5, -m0, 0.5     ; encoding: [0x05,0x00,0x54,0xd6,0xf0,0xfa,0xc0,0x43]
+
+v_div_fixup_f16 v5.l, -src_scc, |vcc_lo|, -1
+// GFX11: v_div_fixup_f16 v5.l, -src_scc, |vcc_lo|, -1 ; encoding: [0x05,0x02,0x54,0xd6,0xfd,0xd4,0x04,0x23]
+
+v_div_fixup_f16 v5.l, 0.5, -m0, 0.5 mul:2
+// GFX11: v_div_fixup_f16 v5.l, 0.5, -m0, 0.5 mul:2 ; encoding: [0x05,0x00,0x54,0xd6,0xf0,0xfa,0xc0,0x4b]
 
 v_div_fixup_f32 v5, v1, v2, s3
 // GFX11: v_div_fixup_f32 v5, v1, v2, s3          ; encoding: [0x05,0x00,0x27,0xd6,0x01,0x05,0x0e,0x00]
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s
index e6f868d2b40e7e..718f8ce2f21ac4 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s
@@ -1406,47 +1406,83 @@ v_cvt_pk_norm_u16_f32_e64_dpp v5, -v1, |v2| row_xmask:0 row_mask:0x1 bank_mask:0
 v_cvt_pk_norm_u16_f32_e64_dpp v255, -|v255|, -|v255| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
 // GFX11: v_cvt_pk_norm_u16_f32_e64_dpp v255, -|v255|, -|v255| row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xff,0x03,0x22,0xd7,0xfa,0xfe,0x03,0x60,0xff,0x6f,0x05,0x30]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, v3 quad_perm:[3,2,1,0]
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, v3 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x1b,0x00,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v3.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v3.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x1b,0x00,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, v3 quad_perm:[0,1,2,3]
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, v3 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0xe4,0x00,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v3.l quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v3.l quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0xe4,0x00,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, v3 row_mirror
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, v3 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x40,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v3.l row_mirror row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v3.l row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x40,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, v255 row_half_mirror
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, v255 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xfe,0x07,0x01,0x41,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v255.l row_half_mirror row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v255.l row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xfe,0x07,0x01,0x41,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, s105 row_shl:1
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, s105 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xa6,0x01,0x01,0x01,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, s105 row_shl:1 row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, s105 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xa6,0x01,0x01,0x01,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, vcc_hi row_shl:15
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, vcc_hi row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xae,0x01,0x01,0x0f,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, vcc_hi row_shl:15
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, vcc_hi row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xae,0x01,0x01,0x0f,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, vcc_lo row_shr:1
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, vcc_lo row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xaa,0x01,0x01,0x11,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, vcc_lo row_shr:1
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, vcc_lo row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xaa,0x01,0x01,0x11,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, |v1|, v2, -ttmp15 row_shr:15
-// GFX11: v_div_fixup_f16_e64_dpp v5, |v1|, v2, -ttmp15 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x01,0x54,0xd6,0xfa,0x04,0xee,0x81,0x01,0x1f,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, |v1.l|, v2.l, -ttmp15 row_shr:15 row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, |v1.l|, v2.l, -ttmp15 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x01,0x54,0xd6,0xfa,0x04,0xee,0x81,0x01,0x1f,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, -|v2|, exec_hi row_ror:1
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, -|v2|, exec_hi row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x02,0x54,0xd6,0xfa,0x04,0xfe,0x41,0x01,0x21,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, -|v2.l|, exec_hi row_ror:1 row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, -|v2.l|, exec_hi row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x02,0x54,0xd6,0xfa,0x04,0xfe,0x41,0x01,0x21,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, -v1, v2, |exec_lo| row_ror:15
-// GFX11: v_div_fixup_f16_e64_dpp v5, -v1, v2, |exec_lo| row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x04,0x54,0xd6,0xfa,0x04,0xfa,0x21,0x01,0x2f,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, -v1.l, v2.l, |exec_lo| row_ror:15 row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, -v1.l, v2.l, |exec_lo| row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x04,0x54,0xd6,0xfa,0x04,0xfa,0x21,0x01,0x2f,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, -|v1|, -|v2|, null row_share:0 row_mask:0xf bank_mask:0xf
-// GFX11: v_div_fixup_f16_e64_dpp v5, -|v1|, -|v2|, null row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x03,0x54,0xd6,0xfa,0x04,0xf2,0x61,0x01,0x50,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, -|v2.l|, null row_share:0 row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, -|v2.l|, null row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x03,0x54,0xd6,0xfa,0x04,0xf2,0x61,0x01,0x50,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, -|v1|, v2, -|-1| row_share:15 row_mask:0x0 bank_mask:0x1
-// GFX11: v_div_fixup_f16_e64_dpp v5, -|v1|, v2, -|-1| row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0x05,0x05,0x54,0xd6,0xfa,0x04,0x06,0xa3,0x01,0x5f,0x01,0x01]
+v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, v2.l, -|-1| row_share:15 row_mask:0x0 bank_mask:0x1
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, v2.l, -|-1| row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0x05,0x05,0x54,0xd6,0xfa,0x04,0x06,0xa3,0x01,0x5f,0x01,0x01]
 
-v_div_fixup_f16_e64_dpp v5, v1, -|v2|, -|0.5| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, -|v2|, -|0.5| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0x05,0x06,0x54,0xd6,0xfa,0x04,0xc2,0xc3,0x01,0x60,0x09,0x13]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, -|v2.l|, -|0.5| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, -|v2.l|, -|0.5| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0x05,0x06,0x54,0xd6,0xfa,0x04,0xc2,0xc3,0x01,0x60,0x09,0x13]
 
-v_div_fixup_f16_e64_dpp v255, -|v255|, -|v255|, -|src_scc| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
-// GFX11: v_div_fixup_f16_e64_dpp v255, -|v255|, -|v255|, -|src_scc| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xff,0x87,0x54,0xd6,0xfa,0xfe,0xf7,0xe3,0xff,0x6f,0x05,0x30]
+v_div_fixup_f16_e64_dpp v255.l, -|v255.l|, -|v255.l|, -|src_scc| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1
+// GFX11: v_div_fixup_f16_e64_dpp v255.l, -|v255.l|, -|v255.l|, -|src_scc| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xff,0x87,0x54,0xd6,0xfa,0xfe,0xf7,0xe3,0xff,0x6f,0x05,0x30]
+
+v_div_fixup_f16_e64_dpp v5.h, v1.h, v2.h, v3.h quad_perm:[3,2,1,0]
+// GFX11: v_div_fixup_f16_e64_dpp v5.h, v1.h, v2.h, v3.h op_sel:[1,1,1,1] quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x78,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x1b,0x00,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v255.h quad_perm:[0,1,2,3]
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v255.h op_sel:[0,0,1,0] quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x20,0x54,0xd6,0xfa,0x04,0xfe,0x07,0x01,0xe4,0x00,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, s3 row_mirror
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, s3 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x00,0x01,0x40,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, s105 row_half_mirror
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, s105 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xa6,0x01,0x01,0x41,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, ttmp15 row_shl:1
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, ttmp15 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xee,0x01,0x01,0x01,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, -|m0| row_shr:15
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, -|m0| row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x04,0x54,0xd6,0xfa,0x04,0xf6,0x81,0x01,0x1f,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, v2.l, -|exec_hi| row_ror:1
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, v2.l, -|exec_hi| row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x05,0x54,0xd6,0xfa,0x04,0xfe,0xa1,0x01,0x21,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, v1.l, -|v2.l|, -|exec_lo| row_ror:15
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, -|v2.l|, -|exec_lo| row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x06,0x54,0xd6,0xfa,0x04,0xfa,0xc1,0x01,0x2f,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, |v1.l|, -v2.l, null row_share:0 row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, |v1.l|, -v2.l, null row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x01,0x54,0xd6,0xfa,0x04,0xf2,0x41,0x01,0x50,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, -v1.h, |v2.l|, -1 row_share:15 row_mask:0x0 bank_mask:0x1
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, -v1.h, |v2.l|, -1 op_sel:[1,0,0,0] row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0x05,0x0a,0x54,0xd6,0xfa,0x04,0x06,0x23,0x01,0x5f,0x01,0x01]
+
+v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, -|v2.h|, 0.5 row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, -|v2.h|, 0.5 op_sel:[0,1,0,0] row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0x05,0x13,0x54,0xd6,0xfa,0x04,0xc2,0x63,0x01,0x60,0x09,0x13]
+
+v_div_fixup_f16_e64_dpp v255.h, -|v255.l|, -|v255.l|, -|src_scc| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
+// GFX11: v_div_fixup_f16_e64_dpp v255.h, -|v255.l|, -|v255.l|, -|src_scc| op_sel:[0,0,0,1] clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xff,0xc7,0x54,0xd6,0xfa,0xfe,0xf7,0xe3,0xff,0x6f...
[truncated]

llvmbot · 2024-12-12T16:10:15Z

@llvm/pr-subscribers-mc

Author: Brox Chen (broxigarchen)

Changes

Support true16 format for v_div_fixup_f16 in MC.

Patch is 168.62 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/119613.diff

14 Files Affected:

(modified) llvm/lib/Target/AMDGPU/SIInstrInfo.cpp (+1)
(modified) llvm/lib/Target/AMDGPU/VOP3Instructions.td (+2-3)
(modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3.s (+56-32)
(modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s (+74-38)
(modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp8.s (+67-34)
(modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3.s (+42-30)
(modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3_dpp16.s (+57-42)
(modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3_dpp8.s (+53-38)
(modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3.txt (+88-16)
(modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3_dpp16.txt (+98-14)
(modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3_dpp8.txt (+98-14)
(modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3.txt (+84-15)
(modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3_dpp16.txt (+90-15)
(modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3_dpp8.txt (+90-15)

diff --git a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
index 4a94d690297949..05dd8e9cf530ee 100644
--- a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
@@ -9284,6 +9284,7 @@ static bool isRenamedInGFX9(int Opcode) {
     GENERATE_RENAMED_GFX9_CASES(AMDGPU::V_SUB_U32)
   //
   case AMDGPU::V_DIV_FIXUP_F16_gfx9_e64:
+  case AMDGPU::V_DIV_FIXUP_F16_gfx9_fake16_e64:
   case AMDGPU::V_FMA_F16_gfx9_e64:
   case AMDGPU::V_INTERP_P2_F16:
   case AMDGPU::V_MAD_F16_e64:
diff --git a/llvm/lib/Target/AMDGPU/VOP3Instructions.td b/llvm/lib/Target/AMDGPU/VOP3Instructions.td
index 8a9f8aa3d16d3a..9fd8bf2c8aa786 100644
--- a/llvm/lib/Target/AMDGPU/VOP3Instructions.td
+++ b/llvm/lib/Target/AMDGPU/VOP3Instructions.td
@@ -339,8 +339,7 @@ let FPDPRounding = 1 in {
   } // End Predicates = [Has16BitInsts, isGFX8Only]
 
   let SubtargetPredicate = isGFX9Plus in {
-    defm V_DIV_FIXUP_F16_gfx9 : VOP3Inst <"v_div_fixup_f16_gfx9",
-                                          VOP3_Profile<VOP_F16_F16_F16_F16, VOP3_OPSEL>, AMDGPUdiv_fixup>;
+    defm V_DIV_FIXUP_F16_gfx9 : VOP3Inst_t16 <"v_div_fixup_f16_gfx9", VOP_F16_F16_F16_F16, AMDGPUdiv_fixup>;
     defm V_FMA_F16_gfx9 : VOP3Inst <"v_fma_f16_gfx9", VOP3_Profile<VOP_F16_F16_F16_F16, VOP3_OPSEL>, any_fma>;
   } // End SubtargetPredicate = isGFX9Plus
 } // End FPDPRounding = 1
@@ -1717,7 +1716,7 @@ defm V_MED3_F16            : VOP3_Realtriple_gfx11<0x24f>;
 defm V_MED3_I16            : VOP3_Realtriple_gfx11_gfx12<0x250>;
 defm V_MED3_U16            : VOP3_Realtriple_gfx11_gfx12<0x251>;
 defm V_MAD_I16             : VOP3_Realtriple_with_name_gfx11_gfx12<0x253, "V_MAD_I16_gfx9", "v_mad_i16">;
-defm V_DIV_FIXUP_F16       : VOP3_Realtriple_with_name_gfx11_gfx12<0x254, "V_DIV_FIXUP_F16_gfx9", "v_div_fixup_f16">;
+defm V_DIV_FIXUP_F16       : VOP3_Realtriple_t16_and_fake16_gfx11_gfx12<0x254, "v_div_fixup_f16", "V_DIV_FIXUP_F16_gfx9">;
 defm V_ADD3_U32            : VOP3_Realtriple_gfx11_gfx12<0x255>;
 defm V_LSHL_OR_B32         : VOP3_Realtriple_gfx11_gfx12<0x256>;
 defm V_AND_OR_B32          : VOP3_Realtriple_gfx11_gfx12<0x257>;
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop3.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop3.s
index b649bab532f262..78d226a71025ed 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop3.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop3.s
@@ -1574,53 +1574,77 @@ v_cvt_pk_norm_u16_f32 v5, -src_scc, |vcc_lo|
 v_cvt_pk_norm_u16_f32 v255, -|0xaf123456|, -|vcc_hi|
 // GFX11: v_cvt_pk_norm_u16_f32 v255, -|0xaf123456|, -|vcc_hi| ; encoding: [0xff,0x03,0x22,0xd7,0xff,0xd6,0x00,0x60,0x56,0x34,0x12,0xaf]
 
-v_div_fixup_f16 v5, v1, v2, s3
-// GFX11: v_div_fixup_f16 v5, v1, v2, s3          ; encoding: [0x05,0x00,0x54,0xd6,0x01,0x05,0x0e,0x00]
+v_div_fixup_f16 v5.l, v1.l, v2.l, s3
+// GFX11: v_div_fixup_f16 v5.l, v1.l, v2.l, s3    ; encoding: [0x05,0x00,0x54,0xd6,0x01,0x05,0x0e,0x00]
 
-v_div_fixup_f16 v5, v255, s2, s105
-// GFX11: v_div_fixup_f16 v5, v255, s2, s105      ; encoding: [0x05,0x00,0x54,0xd6,0xff,0x05,0xa4,0x01]
+v_div_fixup_f16 v5.l, v255.l, s2, s105
+// GFX11: v_div_fixup_f16 v5.l, v255.l, s2, s105  ; encoding: [0x05,0x00,0x54,0xd6,0xff,0x05,0xa4,0x01]
 
-v_div_fixup_f16 v5, s1, v255, exec_hi
-// GFX11: v_div_fixup_f16 v5, s1, v255, exec_hi   ; encoding: [0x05,0x00,0x54,0xd6,0x01,0xfe,0xff,0x01]
+v_div_fixup_f16 v5.l, s1, v255.l, exec_hi
+// GFX11: v_div_fixup_f16 v5.l, s1, v255.l, exec_hi ; encoding: [0x05,0x00,0x54,0xd6,0x01,0xfe,0xff,0x01]
 
-v_div_fixup_f16 v5, s105, s105, exec_lo
-// GFX11: v_div_fixup_f16 v5, s105, s105, exec_lo ; encoding: [0x05,0x00,0x54,0xd6,0x69,0xd2,0xf8,0x01]
+v_div_fixup_f16 v5.l, s105, s105, exec_lo
+// GFX11: v_div_fixup_f16 v5.l, s105, s105, exec_lo ; encoding: [0x05,0x00,0x54,0xd6,0x69,0xd2,0xf8,0x01]
 
-v_div_fixup_f16 v5, vcc_lo, ttmp15, v3
-// GFX11: v_div_fixup_f16 v5, vcc_lo, ttmp15, v3  ; encoding: [0x05,0x00,0x54,0xd6,0x6a,0xf6,0x0c,0x04]
+v_div_fixup_f16 v5.l, vcc_lo, ttmp15, v3.l
+// GFX11: v_div_fixup_f16 v5.l, vcc_lo, ttmp15, v3.l ; encoding: [0x05,0x00,0x54,0xd6,0x6a,0xf6,0x0c,0x04]
 
-v_div_fixup_f16 v5, vcc_hi, 0xfe0b, v255
-// GFX11: v_div_fixup_f16 v5, vcc_hi, 0xfe0b, v255 ; encoding: [0x05,0x00,0x54,0xd6,0x6b,0xfe,0xfd,0x07,0x0b,0xfe,0x00,0x00]
+v_div_fixup_f16 v5.l, vcc_hi, 0xfe0b, v255.l
+// GFX11: v_div_fixup_f16 v5.l, vcc_hi, 0xfe0b, v255.l ; encoding: [0x05,0x00,0x54,0xd6,0x6b,0xfe,0xfd,0x07,0x0b,0xfe,0x00,0x00]
 
-v_div_fixup_f16 v5, -|ttmp15|, -|src_scc|, -|ttmp15|
-// GFX11: v_div_fixup_f16 v5, -|ttmp15|, -|src_scc|, -|ttmp15| ; encoding: [0x05,0x07,0x54,0xd6,0x7b,0xfa,0xed,0xe1]
+v_div_fixup_f16 v5.l, -|ttmp15|, -|src_scc|, -|ttmp15|
+// GFX11: v_div_fixup_f16 v5.l, -|ttmp15|, -|src_scc|, -|ttmp15| ; encoding: [0x05,0x07,0x54,0xd6,0x7b,0xfa,0xed,0xe1]
 
-v_div_fixup_f16 v5, m0, 0.5, m0
-// GFX11: v_div_fixup_f16 v5, m0, 0.5, m0         ; encoding: [0x05,0x00,0x54,0xd6,0x7d,0xe0,0xf5,0x01]
+v_div_fixup_f16 v5.l, m0, 0.5, m0
+// GFX11: v_div_fixup_f16 v5.l, m0, 0.5, m0       ; encoding: [0x05,0x00,0x54,0xd6,0x7d,0xe0,0xf5,0x01]
 
-v_div_fixup_f16 v5, |exec_lo|, -1, vcc_hi
-// GFX11: v_div_fixup_f16 v5, |exec_lo|, -1, vcc_hi ; encoding: [0x05,0x01,0x54,0xd6,0x7e,0x82,0xad,0x01]
+v_div_fixup_f16 v5.l, |exec_lo|, -1, vcc_hi
+// GFX11: v_div_fixup_f16 v5.l, |exec_lo|, -1, vcc_hi ; encoding: [0x05,0x01,0x54,0xd6,0x7e,0x82,0xad,0x01]
 
-v_div_fixup_f16 v5, -|exec_hi|, null, -|vcc_lo| op_sel:[1,1,1,1]
-// GFX11: v_div_fixup_f16 v5, -|exec_hi|, null, -|vcc_lo| op_sel:[1,1,1,1] ; encoding: [0x05,0x7d,0x54,0xd6,0x7f,0xf8,0xa8,0xa1]
+v_div_fixup_f16 v5.h, -|exec_hi|, null, -|vcc_lo| op_sel:[1,1,1,1]
+// GFX11: v_div_fixup_f16 v5.h, -|exec_hi|, null, -|vcc_lo| op_sel:[1,1,1,1] ; encoding: [0x05,0x7d,0x54,0xd6,0x7f,0xf8,0xa8,0xa1]
 
-v_div_fixup_f16 v5, null, exec_lo, -|0xfe0b| op_sel:[0,0,0,0]
-// GFX11: v_div_fixup_f16 v5, null, exec_lo, -|0xfe0b| ; encoding: [0x05,0x04,0x54,0xd6,0x7c,0xfc,0xfc,0x83,0x0b,0xfe,0x00,0x00]
+v_div_fixup_f16 v5.l, null, exec_lo, -|0xfe0b|
+// GFX11: v_div_fixup_f16 v5.l, null, exec_lo, -|0xfe0b| ; encoding: [0x05,0x04,0x54,0xd6,0x7c,0xfc,0xfc,0x83,0x0b,0xfe,0x00,0x00]
 
-v_div_fixup_f16 v5, -1, -|exec_hi|, -|src_scc| op_sel:[1,0,0,0]
-// GFX11: v_div_fixup_f16 v5, -1, -|exec_hi|, -|src_scc| op_sel:[1,0,0,0] ; encoding: [0x05,0x0e,0x54,0xd6,0xc1,0xfe,0xf4,0xc3]
+v_div_fixup_f16 v5.l, -1, -|exec_hi|, -|src_scc| op_sel:[1,0,0,0]
+// GFX11: v_div_fixup_f16 v5.l, -1, -|exec_hi|, -|src_scc| op_sel:[1,0,0,0] ; encoding: [0x05,0x0e,0x54,0xd6,0xc1,0xfe,0xf4,0xc3]
 
-v_div_fixup_f16 v5, 0.5, -m0, 0.5 op_sel:[0,1,0,0]
-// GFX11: v_div_fixup_f16 v5, 0.5, -m0, 0.5 op_sel:[0,1,0,0] ; encoding: [0x05,0x10,0x54,0xd6,0xf0,0xfa,0xc0,0x43]
+v_div_fixup_f16 v5.l, 0.5, -m0, 0.5 op_sel:[0,1,0,0]
+// GFX11: v_div_fixup_f16 v5.l, 0.5, -m0, 0.5 op_sel:[0,1,0,0] ; encoding: [0x05,0x10,0x54,0xd6,0xf0,0xfa,0xc0,0x43]
 
-v_div_fixup_f16 v5, -src_scc, |vcc_lo|, -1 op_sel:[0,0,1,0]
-// GFX11: v_div_fixup_f16 v5, -src_scc, |vcc_lo|, -1 op_sel:[0,0,1,0] ; encoding: [0x05,0x22,0x54,0xd6,0xfd,0xd4,0x04,0x23]
+v_div_fixup_f16 v5.l, -src_scc, |vcc_lo|, -1 op_sel:[0,0,1,0]
+// GFX11: v_div_fixup_f16 v5.l, -src_scc, |vcc_lo|, -1 op_sel:[0,0,1,0] ; encoding: [0x05,0x22,0x54,0xd6,0xfd,0xd4,0x04,0x23]
 
-v_div_fixup_f16 v255, -|0xfe0b|, -|vcc_hi|, null op_sel:[0,0,0,1] clamp
-// GFX11: v_div_fixup_f16 v255, -|0xfe0b|, -|vcc_hi|, null op_sel:[0,0,0,1] clamp ; encoding: [0xff,0xc3,0x54,0xd6,0xff,0xd6,0xf0,0x61,0x0b,0xfe,0x00,0x00]
+v_div_fixup_f16 v255.h, -|0xfe0b|, -|vcc_hi|, null clamp
+// GFX11: v_div_fixup_f16 v255.h, -|0xfe0b|, -|vcc_hi|, null op_sel:[0,0,0,1] clamp ; encoding: [0xff,0xc3,0x54,0xd6,0xff,0xd6,0xf0,0x61,0x0b,0xfe,0x00,0x00]
 
-v_div_fixup_f16 v5, 0.5, -m0, 0.5 op_sel:[0,1,0,0] mul:2
-// GFX11: v_div_fixup_f16 v5, 0.5, -m0, 0.5 op_sel:[0,1,0,0] mul:2 ; encoding: [0x05,0x10,0x54,0xd6,0xf0,0xfa,0xc0,0x4b]
+v_div_fixup_f16 v5.l, 0.5, -m0, 0.5 op_sel:[0,1,0,0] mul:2
+// GFX11: v_div_fixup_f16 v5.l, 0.5, -m0, 0.5 op_sel:[0,1,0,0] mul:2 ; encoding: [0x05,0x10,0x54,0xd6,0xf0,0xfa,0xc0,0x4b]
+
+v_div_fixup_f16 v5.l, v255.h, s2, s105
+// GFX11: v_div_fixup_f16 v5.l, v255.h, s2, s105 op_sel:[1,0,0,0] ; encoding: [0x05,0x08,0x54,0xd6,0xff,0x05,0xa4,0x01]
+
+v_div_fixup_f16 v5.l, s1, v255.h, exec_hi
+// GFX11: v_div_fixup_f16 v5.l, s1, v255.h, exec_hi op_sel:[0,1,0,0] ; encoding: [0x05,0x10,0x54,0xd6,0x01,0xfe,0xff,0x01]
+
+v_div_fixup_f16 v5.l, vcc_hi, 0xfe0b, v255.h
+// GFX11: v_div_fixup_f16 v5.l, vcc_hi, 0xfe0b, v255.h op_sel:[0,0,1,0] ; encoding: [0x05,0x20,0x54,0xd6,0x6b,0xfe,0xfd,0x07,0x0b,0xfe,0x00,0x00]
+
+v_div_fixup_f16 v5.l, -|exec_hi|, null, -|vcc_lo|
+// GFX11: v_div_fixup_f16 v5.l, -|exec_hi|, null, -|vcc_lo| ; encoding: [0x05,0x05,0x54,0xd6,0x7f,0xf8,0xa8,0xa1]
+
+v_div_fixup_f16 v5.l, -1, -|exec_hi|, -|src_scc|
+// GFX11: v_div_fixup_f16 v5.l, -1, -|exec_hi|, -|src_scc| ; encoding: [0x05,0x06,0x54,0xd6,0xc1,0xfe,0xf4,0xc3]
+
+v_div_fixup_f16 v5.l, 0.5, -m0, 0.5
+// GFX11: v_div_fixup_f16 v5.l, 0.5, -m0, 0.5     ; encoding: [0x05,0x00,0x54,0xd6,0xf0,0xfa,0xc0,0x43]
+
+v_div_fixup_f16 v5.l, -src_scc, |vcc_lo|, -1
+// GFX11: v_div_fixup_f16 v5.l, -src_scc, |vcc_lo|, -1 ; encoding: [0x05,0x02,0x54,0xd6,0xfd,0xd4,0x04,0x23]
+
+v_div_fixup_f16 v5.l, 0.5, -m0, 0.5 mul:2
+// GFX11: v_div_fixup_f16 v5.l, 0.5, -m0, 0.5 mul:2 ; encoding: [0x05,0x00,0x54,0xd6,0xf0,0xfa,0xc0,0x4b]
 
 v_div_fixup_f32 v5, v1, v2, s3
 // GFX11: v_div_fixup_f32 v5, v1, v2, s3          ; encoding: [0x05,0x00,0x27,0xd6,0x01,0x05,0x0e,0x00]
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s
index e6f868d2b40e7e..718f8ce2f21ac4 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16.s
@@ -1406,47 +1406,83 @@ v_cvt_pk_norm_u16_f32_e64_dpp v5, -v1, |v2| row_xmask:0 row_mask:0x1 bank_mask:0
 v_cvt_pk_norm_u16_f32_e64_dpp v255, -|v255|, -|v255| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
 // GFX11: v_cvt_pk_norm_u16_f32_e64_dpp v255, -|v255|, -|v255| row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xff,0x03,0x22,0xd7,0xfa,0xfe,0x03,0x60,0xff,0x6f,0x05,0x30]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, v3 quad_perm:[3,2,1,0]
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, v3 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x1b,0x00,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v3.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v3.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x1b,0x00,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, v3 quad_perm:[0,1,2,3]
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, v3 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0xe4,0x00,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v3.l quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v3.l quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0xe4,0x00,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, v3 row_mirror
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, v3 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x40,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v3.l row_mirror row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v3.l row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x40,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, v255 row_half_mirror
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, v255 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xfe,0x07,0x01,0x41,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v255.l row_half_mirror row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v255.l row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xfe,0x07,0x01,0x41,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, s105 row_shl:1
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, s105 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xa6,0x01,0x01,0x01,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, s105 row_shl:1 row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, s105 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xa6,0x01,0x01,0x01,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, vcc_hi row_shl:15
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, vcc_hi row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xae,0x01,0x01,0x0f,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, vcc_hi row_shl:15
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, vcc_hi row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xae,0x01,0x01,0x0f,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, v2, vcc_lo row_shr:1
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, v2, vcc_lo row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xaa,0x01,0x01,0x11,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, vcc_lo row_shr:1
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, vcc_lo row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xaa,0x01,0x01,0x11,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, |v1|, v2, -ttmp15 row_shr:15
-// GFX11: v_div_fixup_f16_e64_dpp v5, |v1|, v2, -ttmp15 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x01,0x54,0xd6,0xfa,0x04,0xee,0x81,0x01,0x1f,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, |v1.l|, v2.l, -ttmp15 row_shr:15 row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, |v1.l|, v2.l, -ttmp15 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x01,0x54,0xd6,0xfa,0x04,0xee,0x81,0x01,0x1f,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, v1, -|v2|, exec_hi row_ror:1
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, -|v2|, exec_hi row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x02,0x54,0xd6,0xfa,0x04,0xfe,0x41,0x01,0x21,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, -|v2.l|, exec_hi row_ror:1 row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, -|v2.l|, exec_hi row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x02,0x54,0xd6,0xfa,0x04,0xfe,0x41,0x01,0x21,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, -v1, v2, |exec_lo| row_ror:15
-// GFX11: v_div_fixup_f16_e64_dpp v5, -v1, v2, |exec_lo| row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x04,0x54,0xd6,0xfa,0x04,0xfa,0x21,0x01,0x2f,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, -v1.l, v2.l, |exec_lo| row_ror:15 row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, -v1.l, v2.l, |exec_lo| row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x04,0x54,0xd6,0xfa,0x04,0xfa,0x21,0x01,0x2f,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, -|v1|, -|v2|, null row_share:0 row_mask:0xf bank_mask:0xf
-// GFX11: v_div_fixup_f16_e64_dpp v5, -|v1|, -|v2|, null row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x03,0x54,0xd6,0xfa,0x04,0xf2,0x61,0x01,0x50,0x01,0xff]
+v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, -|v2.l|, null row_share:0 row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, -|v2.l|, null row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x03,0x54,0xd6,0xfa,0x04,0xf2,0x61,0x01,0x50,0x01,0xff]
 
-v_div_fixup_f16_e64_dpp v5, -|v1|, v2, -|-1| row_share:15 row_mask:0x0 bank_mask:0x1
-// GFX11: v_div_fixup_f16_e64_dpp v5, -|v1|, v2, -|-1| row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0x05,0x05,0x54,0xd6,0xfa,0x04,0x06,0xa3,0x01,0x5f,0x01,0x01]
+v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, v2.l, -|-1| row_share:15 row_mask:0x0 bank_mask:0x1
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, v2.l, -|-1| row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0x05,0x05,0x54,0xd6,0xfa,0x04,0x06,0xa3,0x01,0x5f,0x01,0x01]
 
-v_div_fixup_f16_e64_dpp v5, v1, -|v2|, -|0.5| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
-// GFX11: v_div_fixup_f16_e64_dpp v5, v1, -|v2|, -|0.5| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0x05,0x06,0x54,0xd6,0xfa,0x04,0xc2,0xc3,0x01,0x60,0x09,0x13]
+v_div_fixup_f16_e64_dpp v5.l, v1.l, -|v2.l|, -|0.5| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, -|v2.l|, -|0.5| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0x05,0x06,0x54,0xd6,0xfa,0x04,0xc2,0xc3,0x01,0x60,0x09,0x13]
 
-v_div_fixup_f16_e64_dpp v255, -|v255|, -|v255|, -|src_scc| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
-// GFX11: v_div_fixup_f16_e64_dpp v255, -|v255|, -|v255|, -|src_scc| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xff,0x87,0x54,0xd6,0xfa,0xfe,0xf7,0xe3,0xff,0x6f,0x05,0x30]
+v_div_fixup_f16_e64_dpp v255.l, -|v255.l|, -|v255.l|, -|src_scc| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1
+// GFX11: v_div_fixup_f16_e64_dpp v255.l, -|v255.l|, -|v255.l|, -|src_scc| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xff,0x87,0x54,0xd6,0xfa,0xfe,0xf7,0xe3,0xff,0x6f,0x05,0x30]
+
+v_div_fixup_f16_e64_dpp v5.h, v1.h, v2.h, v3.h quad_perm:[3,2,1,0]
+// GFX11: v_div_fixup_f16_e64_dpp v5.h, v1.h, v2.h, v3.h op_sel:[1,1,1,1] quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x78,0x54,0xd6,0xfa,0x04,0x0e,0x04,0x01,0x1b,0x00,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v255.h quad_perm:[0,1,2,3]
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, v255.h op_sel:[0,0,1,0] quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x20,0x54,0xd6,0xfa,0x04,0xfe,0x07,0x01,0xe4,0x00,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, s3 row_mirror
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, s3 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0x0e,0x00,0x01,0x40,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, s105 row_half_mirror
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, s105 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xa6,0x01,0x01,0x41,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, ttmp15 row_shl:1
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, ttmp15 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x00,0x54,0xd6,0xfa,0x04,0xee,0x01,0x01,0x01,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, -|m0| row_shr:15
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, v2.l, -|m0| row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x04,0x54,0xd6,0xfa,0x04,0xf6,0x81,0x01,0x1f,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, v2.l, -|exec_hi| row_ror:1
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, v2.l, -|exec_hi| row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x05,0x54,0xd6,0xfa,0x04,0xfe,0xa1,0x01,0x21,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, v1.l, -|v2.l|, -|exec_lo| row_ror:15
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, v1.l, -|v2.l|, -|exec_lo| row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x06,0x54,0xd6,0xfa,0x04,0xfa,0xc1,0x01,0x2f,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, |v1.l|, -v2.l, null row_share:0 row_mask:0xf bank_mask:0xf
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, |v1.l|, -v2.l, null row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0x05,0x01,0x54,0xd6,0xfa,0x04,0xf2,0x41,0x01,0x50,0x01,0xff]
+
+v_div_fixup_f16_e64_dpp v5.l, -v1.h, |v2.l|, -1 row_share:15 row_mask:0x0 bank_mask:0x1
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, -v1.h, |v2.l|, -1 op_sel:[1,0,0,0] row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0x05,0x0a,0x54,0xd6,0xfa,0x04,0x06,0x23,0x01,0x5f,0x01,0x01]
+
+v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, -|v2.h|, 0.5 row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
+// GFX11: v_div_fixup_f16_e64_dpp v5.l, -|v1.l|, -|v2.h|, 0.5 op_sel:[0,1,0,0] row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0x05,0x13,0x54,0xd6,0xfa,0x04,0xc2,0x63,0x01,0x60,0x09,0x13]
+
+v_div_fixup_f16_e64_dpp v255.h, -|v255.l|, -|v255.l|, -|src_scc| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
+// GFX11: v_div_fixup_f16_e64_dpp v255.h, -|v255.l|, -|v255.l|, -|src_scc| op_sel:[0,0,0,1] clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0xff,0xc7,0x54,0xd6,0xfa,0xfe,0xf7,0xe3,0xff,0x6f...
[truncated]

broxigarchen · 2024-12-18T18:27:30Z

resolved conflicts

Sisyph · 2024-12-19T17:58:44Z

    GENERATE_RENAMED_GFX9_CASES(AMDGPU::V_SUB_U32)
  //
  case AMDGPU::V_DIV_FIXUP_F16_gfx9_e64:
+  case AMDGPU::V_DIV_FIXUP_F16_gfx9_fake16_e64:


This looks slightly weird. The fake16 instruction didn't exist until gfx11. But it depends on how isRenamedInGFX9 is used. I'm not sure if it is necessary, harmless, or harmful.

True16 for v_div_fixup_f16 in MC

ea6c36f

broxigarchen force-pushed the main-merge-true16-vop3-mc-more-instructions-6 branch from 4f45969 to ea6c36f Compare December 11, 2024 22:34

broxigarchen marked this pull request as ready for review December 12, 2024 16:09

llvmbot added backend:AMDGPU llvm:mc Machine (object) code labels Dec 12, 2024

broxigarchen changed the title ~~[AMDGPU][True16][MC] true16 for v_div_fixup_f16 True16~~ [AMDGPU][True16][MC] true16 for v_div_fixup_f16 Dec 12, 2024

arsenm approved these changes Dec 13, 2024

View reviewed changes

Merge branch 'main' into main-merge-true16-vop3-mc-more-instructions-6

74c9cfa

broxigarchen merged commit e10b12e into llvm:main Dec 18, 2024

Sisyph reviewed Dec 19, 2024

View reviewed changes

broxigarchen requested a review from kosarev January 10, 2025 15:22

This was referenced Jun 2, 2025

[MTE] [NFC] use vector to collect globals to tag (#120283) #142329

Closed

[MTE] [NFC] use vector to collect globals to tag (#120283) #142330

Draft

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[AMDGPU][True16][MC] true16 for v_div_fixup_f16#119613

[AMDGPU][True16][MC] true16 for v_div_fixup_f16#119613
broxigarchen merged 2 commits into
llvm:mainfrom
broxigarchen:main-merge-true16-vop3-mc-more-instructions-6

broxigarchen commented Dec 11, 2024 •

edited

Loading

Uh oh!

llvmbot commented Dec 12, 2024

Uh oh!

llvmbot commented Dec 12, 2024

Uh oh!

broxigarchen commented Dec 18, 2024

Uh oh!

Sisyph Dec 19, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

broxigarchen commented Dec 11, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

llvmbot commented Dec 12, 2024

Uh oh!

llvmbot commented Dec 12, 2024

Uh oh!

broxigarchen commented Dec 18, 2024

Uh oh!

Sisyph Dec 19, 2024

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

broxigarchen commented Dec 11, 2024 •

edited

Loading