Skip to content

[AMDGPU][True16][MC] true16 for v_cmpx_xx_f16 #123419

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 2 commits into from
Jan 27, 2025
Merged

[AMDGPU][True16][MC] true16 for v_cmpx_xx_f16 #123419

merged 2 commits into from
Jan 27, 2025

Conversation

broxigarchen
Copy link
Contributor

@broxigarchen broxigarchen commented Jan 17, 2025 •
edited
Loading

A bulk commit of true16 support for v_cmpx_xx_f16 instructions including:

v_cmpx_f_f16
v_cmpx_le_f16
v_cmpx_gt_f16
v_cmpx_lg_f16
v_cmpx_ge_f16
v_cmpx_o_f16
v_cmpx_u_f16
v_cmpx_nge_f16
v_cmpx_nlg_f16
v_cmpx_ngt_f16
v_cmpx_nle_f16
v_cmpx_neq_f16
v_cmpx_nlt_f16
v_cmpx_t_f16

v_cmpx_eq_f16 is not in this patch and will be added in the following patch

@broxigarchen broxigarchen changed the title true16 for v_cmpx_xx_f16 [AMDGPU][True16][MC] true16 for v_cmpx_xx_f16 Jan 17, 2025
@broxigarchen broxigarchen force-pushed the main-merge-true16-vopc-mc-more-instruction-8 branch from 31953bd to a078660 Compare January 20, 2025 05:52
@broxigarchen broxigarchen marked this pull request as ready for review January 20, 2025 05:52
@llvmbot llvmbot added backend:AMDGPU mc Machine (object) code labels Jan 20, 2025
@llvmbot
Copy link
Member

llvmbot commented Jan 20, 2025 •
edited
Loading

@llvm/pr-subscribers-mc

@llvm/pr-subscribers-backend-amdgpu

Author: Brox Chen (broxigarchen)

Changes

A bulk commit of true16 support for v_cmpx_xx_f16 instructions including:

v_cmpx_f_f16
v_cmpx_eq_f16
v_cmpx_le_f16
v_cmpx_gt_f16
v_cmpx_lg_f16
v_cmpx_ge_f16
v_cmpx_o_f16
v_cmpx_u_f16
v_cmpx_nge_f16
v_cmpx_nlg_f16
v_cmpx_ngt_f16
v_cmpx_nle_f16
v_cmpx_neq_f16
v_cmpx_nlt_f16
v_cmpx_t_f16

Patch is 1.65 MiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/123419.diff

33 Files Affected:

  • (modified) llvm/lib/Target/AMDGPU/VOPCInstructions.td (+15-15)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16_from_vopcx.s (+546-420)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp8_from_vopcx.s (+246-120)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vop3_from_vopcx.s (+144-60)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vopc_t16_err.s (+1-1)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vopc_t16_promote.s (+1-1)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vopcx.s (+669-444)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vopcx_dpp16.s (+549-414)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vopcx_dpp8.s (+221-86)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vopcx_t16_err.s (+417-165)
  • (modified) llvm/test/MC/AMDGPU/gfx11_asm_vopcx_t16_promote.s (+450-180)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3cx.s (+120-48)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3cx_dpp16.s (+492-384)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vop3cx_dpp8.s (+252-144)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vopc_t16_err.s (+1-1)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vopc_t16_promote.s (+1-1)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vopcx.s (+504-360)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vopcx_dpp16.s (+408-336)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vopcx_dpp8.s (+144-72)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vopcx_t16_err.s (+360-144)
  • (modified) llvm/test/MC/AMDGPU/gfx12_asm_vopcx_t16_promote.s (+360-144)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3_dpp16_from_vopcx.txt (+588-210)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3_dpp8_from_vopcx.txt (+288-60)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vop3_from_vopcx.txt (+172-30)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vopcx.txt (+730-225)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vopcx_dpp16.txt (+588-210)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vopcx_dpp8.txt (+312-30)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3cx.txt (+159-26)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3cx_dpp16.txt ()
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vop3cx_dpp8.txt ()
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vopcx.txt (+582-195)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vopcx_dpp16.txt (+460-182)
  • (modified) llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vopcx_dpp8.txt (+148-26)
diff --git a/llvm/lib/Target/AMDGPU/VOPCInstructions.td b/llvm/lib/Target/AMDGPU/VOPCInstructions.td
index bba8aa570d2b58..ebc2d622bb16a3 100644
--- a/llvm/lib/Target/AMDGPU/VOPCInstructions.td
+++ b/llvm/lib/Target/AMDGPU/VOPCInstructions.td
@@ -1924,22 +1924,22 @@ defm V_CMP_CLASS_F16_fake16 : VOPC_Real_t16_gfx11_gfx12<0x07d, "v_cmp_class_f16"
 defm V_CMP_CLASS_F32     : VOPC_Real_gfx11_gfx12<0x07e>;
 defm V_CMP_CLASS_F64     : VOPC_Real_gfx11_gfx12<0x07f>;
 
-defm V_CMPX_F_F16_fake16     : VOPCX_Real_t16_gfx11<0x080, "v_cmpx_f_f16">;
+defm V_CMPX_F_F16            : VOPCX_Real_t16_and_fake16_gfx11<0x080, "v_cmpx_f_f16">;
 defm V_CMPX_LT_F16           : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x081, "v_cmpx_lt_f16">;
-defm V_CMPX_EQ_F16_fake16    : VOPCX_Real_t16_gfx11_gfx12<0x082, "v_cmpx_eq_f16">;
-defm V_CMPX_LE_F16_fake16    : VOPCX_Real_t16_gfx11_gfx12<0x083, "v_cmpx_le_f16">;
-defm V_CMPX_GT_F16_fake16    : VOPCX_Real_t16_gfx11_gfx12<0x084, "v_cmpx_gt_f16">;
-defm V_CMPX_LG_F16_fake16    : VOPCX_Real_t16_gfx11_gfx12<0x085, "v_cmpx_lg_f16">;
-defm V_CMPX_GE_F16_fake16    : VOPCX_Real_t16_gfx11_gfx12<0x086, "v_cmpx_ge_f16">;
-defm V_CMPX_O_F16_fake16     : VOPCX_Real_t16_gfx11_gfx12<0x087, "v_cmpx_o_f16">;
-defm V_CMPX_U_F16_fake16     : VOPCX_Real_t16_gfx11_gfx12<0x088, "v_cmpx_u_f16">;
-defm V_CMPX_NGE_F16_fake16   : VOPCX_Real_t16_gfx11_gfx12<0x089, "v_cmpx_nge_f16">;
-defm V_CMPX_NLG_F16_fake16   : VOPCX_Real_t16_gfx11_gfx12<0x08a, "v_cmpx_nlg_f16">;
-defm V_CMPX_NGT_F16_fake16   : VOPCX_Real_t16_gfx11_gfx12<0x08b, "v_cmpx_ngt_f16">;
-defm V_CMPX_NLE_F16_fake16   : VOPCX_Real_t16_gfx11_gfx12<0x08c, "v_cmpx_nle_f16">;
-defm V_CMPX_NEQ_F16_fake16   : VOPCX_Real_t16_gfx11_gfx12<0x08d, "v_cmpx_neq_f16">;
-defm V_CMPX_NLT_F16_fake16   : VOPCX_Real_t16_gfx11_gfx12<0x08e, "v_cmpx_nlt_f16">;
-defm V_CMPX_T_F16_fake16     : VOPCX_Real_with_name_gfx11<0x08f, "V_CMPX_TRU_F16_fake16", "v_cmpx_t_f16", "v_cmpx_tru_f16">;
+defm V_CMPX_EQ_F16           : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x082, "v_cmpx_eq_f16">;
+defm V_CMPX_LE_F16           : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x083, "v_cmpx_le_f16">;
+defm V_CMPX_GT_F16           : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x084, "v_cmpx_gt_f16">;
+defm V_CMPX_LG_F16           : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x085, "v_cmpx_lg_f16">;
+defm V_CMPX_GE_F16           : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x086, "v_cmpx_ge_f16">;
+defm V_CMPX_O_F16            : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x087, "v_cmpx_o_f16">;
+defm V_CMPX_U_F16            : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x088, "v_cmpx_u_f16">;
+defm V_CMPX_NGE_F16          : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x089, "v_cmpx_nge_f16">;
+defm V_CMPX_NLG_F16          : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x08a, "v_cmpx_nlg_f16">;
+defm V_CMPX_NGT_F16          : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x08b, "v_cmpx_ngt_f16">;
+defm V_CMPX_NLE_F16          : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x08c, "v_cmpx_nle_f16">;
+defm V_CMPX_NEQ_F16          : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x08d, "v_cmpx_neq_f16">;
+defm V_CMPX_NLT_F16          : VOPCX_Real_t16_and_fake16_gfx11_gfx12<0x08e, "v_cmpx_nlt_f16">;
+defm V_CMPX_T_F16            : VOPCX_Real_t16_and_fake16_gfx11<0x08f, "v_cmpx_t_f16", "V_CMPX_TRU_F16", "v_cmpx_tru_f16">;
 
 defm V_CMPX_F_F32     : VOPCX_Real_gfx11<0x090>;
 defm V_CMPX_LT_F32    : VOPCX_Real_gfx11_gfx12<0x091>;
diff --git a/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16_from_vopcx.s b/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16_from_vopcx.s
index 60ec94446235ed..15ddb6374821c7 100644
--- a/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16_from_vopcx.s
+++ b/llvm/test/MC/AMDGPU/gfx11_asm_vop3_dpp16_from_vopcx.s
@@ -339,47 +339,56 @@ v_cmpx_eq_u32_e64_dpp v1, v2 row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1
 v_cmpx_eq_u32_e64_dpp v255, v255 row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
 // GFX11: v_cmpx_eq_u32_e64_dpp v255, v255 row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0x7e,0x00,0xca,0xd4,0xfa,0xfe,0x03,0x00,0xff,0x6f,0x05,0x30]
 
-v_cmpx_f_f16_e64_dpp v1, v2 quad_perm:[3,2,1,0]
-// GFX11: v_cmpx_f_f16_e64_dpp v1, v2 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x1b,0x00,0xff]
+v_cmpx_f_f16_e64_dpp v1.l, v2.l quad_perm:[3,2,1,0]
+// GFX11: v_cmpx_f_f16_e64_dpp v1.l, v2.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x1b,0x00,0xff]
 
-v_cmpx_f_f16_e64_dpp v1, v2 quad_perm:[0,1,2,3]
-// GFX11: v_cmpx_f_f16_e64_dpp v1, v2 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0xe4,0x00,0xff]
+v_cmpx_f_f16_e64_dpp v1.l, v2.l quad_perm:[0,1,2,3]
+// GFX11: v_cmpx_f_f16_e64_dpp v1.l, v2.l quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0xe4,0x00,0xff]
 
-v_cmpx_f_f16_e64_dpp v1, v2 row_mirror
-// GFX11: v_cmpx_f_f16_e64_dpp v1, v2 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x40,0x01,0xff]
+v_cmpx_f_f16_e64_dpp v1.l, v2.l row_mirror
+// GFX11: v_cmpx_f_f16_e64_dpp v1.l, v2.l row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x40,0x01,0xff]
 
-v_cmpx_f_f16_e64_dpp v1, v2 row_half_mirror
-// GFX11: v_cmpx_f_f16_e64_dpp v1, v2 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x41,0x01,0xff]
+v_cmpx_f_f16_e64_dpp v1.l, v2.l row_half_mirror
+// GFX11: v_cmpx_f_f16_e64_dpp v1.l, v2.l row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x41,0x01,0xff]
 
-v_cmpx_f_f16_e64_dpp v1, v2 row_shl:1
-// GFX11: v_cmpx_f_f16_e64_dpp v1, v2 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x01,0x01,0xff]
+v_cmpx_f_f16_e64_dpp v1.l, v2.l row_shl:1
+// GFX11: v_cmpx_f_f16_e64_dpp v1.l, v2.l row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x01,0x01,0xff]
 
-v_cmpx_f_f16_e64_dpp v1, v2 row_shl:15
-// GFX11: v_cmpx_f_f16_e64_dpp v1, v2 row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x0f,0x01,0xff]
+v_cmpx_f_f16_e64_dpp v1.l, v2.l row_shl:15
+// GFX11: v_cmpx_f_f16_e64_dpp v1.l, v2.l row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x0f,0x01,0xff]
 
-v_cmpx_f_f16_e64_dpp v1, v2 row_shr:1
-// GFX11: v_cmpx_f_f16_e64_dpp v1, v2 row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x11,0x01,0xff]
+v_cmpx_f_f16_e64_dpp v1.l, v2.l row_shr:1
+// GFX11: v_cmpx_f_f16_e64_dpp v1.l, v2.l row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x11,0x01,0xff]
 
-v_cmpx_f_f16_e64_dpp v1, v2 row_shr:15
-// GFX11: v_cmpx_f_f16_e64_dpp v1, v2 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x1f,0x01,0xff]
+v_cmpx_f_f16_e64_dpp v1.l, v2.l row_shr:15
+// GFX11: v_cmpx_f_f16_e64_dpp v1.l, v2.l row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x1f,0x01,0xff]
 
-v_cmpx_f_f16_e64_dpp v1, v2 row_ror:1
-// GFX11: v_cmpx_f_f16_e64_dpp v1, v2 row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x21,0x01,0xff]
+v_cmpx_f_f16_e64_dpp v1.l, v2.l row_ror:1
+// GFX11: v_cmpx_f_f16_e64_dpp v1.l, v2.l row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x21,0x01,0xff]
 
-v_cmpx_f_f16_e64_dpp v1, v2 row_ror:15
-// GFX11: v_cmpx_f_f16_e64_dpp v1, v2 row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x2f,0x01,0xff]
+v_cmpx_f_f16_e64_dpp v1.l, v2.l row_ror:15
+// GFX11: v_cmpx_f_f16_e64_dpp v1.l, v2.l row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x2f,0x01,0xff]
 
-v_cmpx_f_f16_e64_dpp v1, v2 row_share:0 row_mask:0xf bank_mask:0xf
-// GFX11: v_cmpx_f_f16_e64_dpp v1, v2 row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x50,0x01,0xff]
+v_cmpx_f_f16_e64_dpp v1.l, v2.l row_share:0 row_mask:0xf bank_mask:0xf
+// GFX11: v_cmpx_f_f16_e64_dpp v1.l, v2.l row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x80,0xd4,0xfa,0x04,0x02,0x00,0x01,0x50,0x01,0xff]
 
-v_cmpx_f_f16_e64_dpp |v1|, -v2 row_share:15 row_mask:0x0 bank_mask:0x1
-// GFX11: v_cmpx_f_f16_e64_dpp |v1|, -v2 row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0x7e,0x01,0x80,0xd4,0xfa,0x04,0x02,0x40,0x01,0x5f,0x01,0x01]
+v_cmpx_f_f16_e64_dpp |v1.l|, -v2.l row_share:15 row_mask:0x0 bank_mask:0x1
+// GFX11: v_cmpx_f_f16_e64_dpp |v1.l|, -v2.l row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0x7e,0x01,0x80,0xd4,0xfa,0x04,0x02,0x40,0x01,0x5f,0x01,0x01]
 
-v_cmpx_f_f16_e64_dpp -v1, |v2| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
-// GFX11: v_cmpx_f_f16_e64_dpp -v1, |v2| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0x7e,0x02,0x80,0xd4,0xfa,0x04,0x02,0x20,0x01,0x60,0x09,0x13]
+v_cmpx_f_f16_e64_dpp -v1.l, |v2.l| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
+// GFX11: v_cmpx_f_f16_e64_dpp -v1.l, |v2.l| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0x7e,0x02,0x80,0xd4,0xfa,0x04,0x02,0x20,0x01,0x60,0x09,0x13]
 
-v_cmpx_f_f16_e64_dpp -|v255|, -|v255| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
-// GFX11: v_cmpx_f_f16_e64_dpp -|v255|, -|v255| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0x7e,0x83,0x80,0xd4,0xfa,0xfe,0x03,0x60,0xff,0x6f,0x05,0x30]
+v_cmpx_f_f16_e64_dpp -|v255.l|, -|v255.l| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
+// GFX11: v_cmpx_f_f16_e64_dpp -|v255.l|, -|v255.l| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0x7e,0x83,0x80,0xd4,0xfa,0xfe,0x03,0x60,0xff,0x6f,0x05,0x30]
+
+v_cmpx_f_f16_e64_dpp |v1.h|, -v2.h row_share:15 row_mask:0x0 bank_mask:0x1
+// GFX11: v_cmpx_f_f16_e64_dpp |v1.h|, -v2.h op_sel:[1,1] row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0x7e,0x19,0x80,0xd4,0xfa,0x04,0x02,0x40,0x01,0x5f,0x01,0x01]
+
+v_cmpx_f_f16_e64_dpp -v1.h, |v2.l| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
+// GFX11: v_cmpx_f_f16_e64_dpp -v1.h, |v2.l| op_sel:[1,0] row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0x7e,0x0a,0x80,0xd4,0xfa,0x04,0x02,0x20,0x01,0x60,0x09,0x13]
+
+v_cmpx_f_f16_e64_dpp -|v255.l|, -|v255.h| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
+// GFX11: v_cmpx_f_f16_e64_dpp -|v255.l|, -|v255.h| op_sel:[0,1] clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0x7e,0x93,0x80,0xd4,0xfa,0xfe,0x03,0x60,0xff,0x6f,0x05,0x30]
 
 v_cmpx_f_f32_e64_dpp v1, v2 quad_perm:[3,2,1,0]
 // GFX11: v_cmpx_f_f32_e64_dpp v1, v2 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x90,0xd4,0xfa,0x04,0x02,0x00,0x01,0x1b,0x00,0xff]
@@ -507,47 +516,56 @@ v_cmpx_f_u32_e64_dpp v1, v2 row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1
 v_cmpx_f_u32_e64_dpp v255, v255 row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
 // GFX11: v_cmpx_f_u32_e64_dpp v255, v255 row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0x7e,0x00,0xc8,0xd4,0xfa,0xfe,0x03,0x00,0xff,0x6f,0x05,0x30]
 
-v_cmpx_ge_f16_e64_dpp v1, v2 quad_perm:[3,2,1,0]
-// GFX11: v_cmpx_ge_f16_e64_dpp v1, v2 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x1b,0x00,0xff]
+v_cmpx_ge_f16_e64_dpp v1.l, v2.l quad_perm:[3,2,1,0]
+// GFX11: v_cmpx_ge_f16_e64_dpp v1.l, v2.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x1b,0x00,0xff]
+
+v_cmpx_ge_f16_e64_dpp v1.l, v2.l quad_perm:[0,1,2,3]
+// GFX11: v_cmpx_ge_f16_e64_dpp v1.l, v2.l quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0xe4,0x00,0xff]
+
+v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_mirror
+// GFX11: v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x40,0x01,0xff]
+
+v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_half_mirror
+// GFX11: v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x41,0x01,0xff]
 
-v_cmpx_ge_f16_e64_dpp v1, v2 quad_perm:[0,1,2,3]
-// GFX11: v_cmpx_ge_f16_e64_dpp v1, v2 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0xe4,0x00,0xff]
+v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_shl:1
+// GFX11: v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x01,0x01,0xff]
 
-v_cmpx_ge_f16_e64_dpp v1, v2 row_mirror
-// GFX11: v_cmpx_ge_f16_e64_dpp v1, v2 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x40,0x01,0xff]
+v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_shl:15
+// GFX11: v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x0f,0x01,0xff]
 
-v_cmpx_ge_f16_e64_dpp v1, v2 row_half_mirror
-// GFX11: v_cmpx_ge_f16_e64_dpp v1, v2 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x41,0x01,0xff]
+v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_shr:1
+// GFX11: v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x11,0x01,0xff]
 
-v_cmpx_ge_f16_e64_dpp v1, v2 row_shl:1
-// GFX11: v_cmpx_ge_f16_e64_dpp v1, v2 row_shl:1 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x01,0x01,0xff]
+v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_shr:15
+// GFX11: v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x1f,0x01,0xff]
 
-v_cmpx_ge_f16_e64_dpp v1, v2 row_shl:15
-// GFX11: v_cmpx_ge_f16_e64_dpp v1, v2 row_shl:15 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x0f,0x01,0xff]
+v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_ror:1
+// GFX11: v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x21,0x01,0xff]
 
-v_cmpx_ge_f16_e64_dpp v1, v2 row_shr:1
-// GFX11: v_cmpx_ge_f16_e64_dpp v1, v2 row_shr:1 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x11,0x01,0xff]
+v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_ror:15
+// GFX11: v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x2f,0x01,0xff]
 
-v_cmpx_ge_f16_e64_dpp v1, v2 row_shr:15
-// GFX11: v_cmpx_ge_f16_e64_dpp v1, v2 row_shr:15 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x1f,0x01,0xff]
+v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_share:0 row_mask:0xf bank_mask:0xf
+// GFX11: v_cmpx_ge_f16_e64_dpp v1.l, v2.l row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x50,0x01,0xff]
 
-v_cmpx_ge_f16_e64_dpp v1, v2 row_ror:1
-// GFX11: v_cmpx_ge_f16_e64_dpp v1, v2 row_ror:1 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x21,0x01,0xff]
+v_cmpx_ge_f16_e64_dpp |v1.l|, -v2.l row_share:15 row_mask:0x0 bank_mask:0x1
+// GFX11: v_cmpx_ge_f16_e64_dpp |v1.l|, -v2.l row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0x7e,0x01,0x86,0xd4,0xfa,0x04,0x02,0x40,0x01,0x5f,0x01,0x01]
 
-v_cmpx_ge_f16_e64_dpp v1, v2 row_ror:15
-// GFX11: v_cmpx_ge_f16_e64_dpp v1, v2 row_ror:15 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x2f,0x01,0xff]
+v_cmpx_ge_f16_e64_dpp -v1.l, |v2.l| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
+// GFX11: v_cmpx_ge_f16_e64_dpp -v1.l, |v2.l| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0x7e,0x02,0x86,0xd4,0xfa,0x04,0x02,0x20,0x01,0x60,0x09,0x13]
 
-v_cmpx_ge_f16_e64_dpp v1, v2 row_share:0 row_mask:0xf bank_mask:0xf
-// GFX11: v_cmpx_ge_f16_e64_dpp v1, v2 row_share:0 row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x86,0xd4,0xfa,0x04,0x02,0x00,0x01,0x50,0x01,0xff]
+v_cmpx_ge_f16_e64_dpp -|v255.l|, -|v255.l| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
+// GFX11: v_cmpx_ge_f16_e64_dpp -|v255.l|, -|v255.l| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0x7e,0x83,0x86,0xd4,0xfa,0xfe,0x03,0x60,0xff,0x6f,0x05,0x30]
 
-v_cmpx_ge_f16_e64_dpp |v1|, -v2 row_share:15 row_mask:0x0 bank_mask:0x1
-// GFX11: v_cmpx_ge_f16_e64_dpp |v1|, -v2 row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0x7e,0x01,0x86,0xd4,0xfa,0x04,0x02,0x40,0x01,0x5f,0x01,0x01]
+v_cmpx_ge_f16_e64_dpp |v1.h|, -v2.h row_share:15 row_mask:0x0 bank_mask:0x1
+// GFX11: v_cmpx_ge_f16_e64_dpp |v1.h|, -v2.h op_sel:[1,1] row_share:15 row_mask:0x0 bank_mask:0x1 ; encoding: [0x7e,0x19,0x86,0xd4,0xfa,0x04,0x02,0x40,0x01,0x5f,0x01,0x01]
 
-v_cmpx_ge_f16_e64_dpp -v1, |v2| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
-// GFX11: v_cmpx_ge_f16_e64_dpp -v1, |v2| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0x7e,0x02,0x86,0xd4,0xfa,0x04,0x02,0x20,0x01,0x60,0x09,0x13]
+v_cmpx_ge_f16_e64_dpp -v1.h, |v2.l| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
+// GFX11: v_cmpx_ge_f16_e64_dpp -v1.h, |v2.l| op_sel:[1,0] row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 ; encoding: [0x7e,0x0a,0x86,0xd4,0xfa,0x04,0x02,0x20,0x01,0x60,0x09,0x13]
 
-v_cmpx_ge_f16_e64_dpp -|v255|, -|v255| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
-// GFX11: v_cmpx_ge_f16_e64_dpp -|v255|, -|v255| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0x7e,0x83,0x86,0xd4,0xfa,0xfe,0x03,0x60,0xff,0x6f,0x05,0x30]
+v_cmpx_ge_f16_e64_dpp -|v255.l|, -|v255.h| clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
+// GFX11: v_cmpx_ge_f16_e64_dpp -|v255.l|, -|v255.h| op_sel:[0,1] clamp row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0x7e,0x93,0x86,0xd4,0xfa,0xfe,0x03,0x60,0xff,0x6f,0x05,0x30]
 
 v_cmpx_ge_f32_e64_dpp v1, v2 quad_perm:[3,2,1,0]
 // GFX11: v_cmpx_ge_f32_e64_dpp v1, v2 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x96,0xd4,0xfa,0x04,0x02,0x00,0x01,0x1b,0x00,0xff]
@@ -759,47 +777,56 @@ v_cmpx_ge_u32_e64_dpp v1, v2 row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1
 v_cmpx_ge_u32_e64_dpp v255, v255 row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
 // GFX11: v_cmpx_ge_u32_e64_dpp v255, v255 row_xmask:15 row_mask:0x3 bank_mask:0x0 fi:1 ; encoding: [0x7e,0x00,0xce,0xd4,0xfa,0xfe,0x03,0x00,0xff,0x6f,0x05,0x30]
 
-v_cmpx_gt_f16_e64_dpp v1, v2 quad_perm:[3,2,1,0]
-// GFX11: v_cmpx_gt_f16_e64_dpp v1, v2 quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x84,0xd4,0xfa,0x04,0x02,0x00,0x01,0x1b,0x00,0xff]
+v_cmpx_gt_f16_e64_dpp v1.l, v2.l quad_perm:[3,2,1,0]
+// GFX11: v_cmpx_gt_f16_e64_dpp v1.l, v2.l quad_perm:[3,2,1,0] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x84,0xd4,0xfa,0x04,0x02,0x00,0x01,0x1b,0x00,0xff]
 
-v_cmpx_gt_f16_e64_dpp v1, v2 quad_perm:[0,1,2,3]
-// GFX11: v_cmpx_gt_f16_e64_dpp v1, v2 quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x84,0xd4,0xfa,0x04,0x02,0x00,0x01,0xe4,0x00,0xff]
+v_cmpx_gt_f16_e64_dpp v1.l, v2.l quad_perm:[0,1,2,3]
+// GFX11: v_cmpx_gt_f16_e64_dpp v1.l, v2.l quad_perm:[0,1,2,3] row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x84,0xd4,0xfa,0x04,0x02,0x00,0x01,0xe4,0x00,0xff]
 
-v_cmpx_gt_f16_e64_dpp v1, v2 row_mirror
-// GFX11: v_cmpx_gt_f16_e64_dpp v1, v2 row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x84,0xd4,0xfa,0x04,0x02,0x00,0x01,0x40,0x01,0xff]
+v_cmpx_gt_f16_e64_dpp v1.l, v2.l row_mirror
+// GFX11: v_cmpx_gt_f16_e64_dpp v1.l, v2.l row_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x84,0xd4,0xfa,0x04,0x02,0x00,0x01,0x40,0x01,0xff]
 
-v_cmpx_gt_f16_e64_dpp v1, v2 row_half_mirror
-// GFX11: v_cmpx_gt_f16_e64_dpp v1, v2 row_half_mirror row_mask:0xf bank_mask:0xf ; encoding: [0x7e,0x00,0x84,0xd4,0xfa,0x04,0x02,0x00,0x01,0x41,0x01,0xff]
+v_cmpx_gt_f16_e64_dpp v1.l, v2.l row_half_mirror
+// GFX11: v_cmpx_gt_f16_e64_dpp...
[truncated]

Sisyph
Sisyph reviewed Jan 20, 2025
View reviewed changes
llvm/test/MC/AMDGPU/gfx11_asm_vop3_from_vopcx.s
@@ -518,11 +518,11 @@ v_cmpx_eq_u64_e64 src_scc, exec
v_cmpx_eq_u64_e64 0xaf123456, vcc
Copy link
Contributor

Choose a reason for hiding this comment

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

Missing any tests for v_cmpx_eq_f16_e64 .l and .h in this file and several others. In one file I see .l but no .h tests for v_cmpx_eq_f16_e64.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Thanks Joe for pointing this out! I will double check

Copy link
Contributor Author

@broxigarchen broxigarchen Jan 22, 2025 •
edited
Loading

Choose a reason for hiding this comment

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

It seems both v_cmp_eq_f16(from previous patch) and v_cmpx_eq_f16(from this patch) are missing. Removed v_cmpx_eq_f16 from this patch and will open another patch to merge the missing two

Copy link
Contributor

Choose a reason for hiding this comment

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

That sounds fine for asm. I see in the disasm tests we are missing tests for .h registers on v_cmpx_eq_f16. What is the plan for those?

Copy link
Contributor Author

@broxigarchen broxigarchen Jan 22, 2025 •
edited
Loading

Choose a reason for hiding this comment

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

Both asm,disasm test and t16 instruction definition for v_cmpx_eq_f16 are removed from this patch and I will add them in the next patch

Copy link
Contributor Author

@broxigarchen broxigarchen Jan 23, 2025 •
edited
Loading

Choose a reason for hiding this comment

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

Another patch for v_cmpx_eq_f16 is created here #124038

In fact, after fixing the script it seems v_cmp_eq_f16 is merged correctly so we don't need a patch for that anymore.

@broxigarchen broxigarchen force-pushed the main-merge-true16-vopc-mc-more-instruction-8 branch from a078660 to 6301c5b Compare January 22, 2025 17:32
@broxigarchen broxigarchen force-pushed the main-merge-true16-vopc-mc-more-instruction-8 branch from 6301c5b to decabc3 Compare January 22, 2025 19:50
@broxigarchen broxigarchen requested a review from Sisyph January 24, 2025 23:21
@broxigarchen
Copy link
Contributor Author

rebased and resolved conflicts

@broxigarchen broxigarchen merged commit 62340ff into llvm:main Jan 27, 2025
6 of 8 checks passed
kosarev
kosarev reviewed Jan 31, 2025
View reviewed changes
llvm/test/MC/Disassembler/AMDGPU/gfx11_dasm_vopcx_dpp8.txt
Comment on lines +140 to +141
# GFX12-REAL16: v_cmpx_f_f16 v127.l, v127.l dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0xe9,0xfe,0x00,0x7d,0x7f,0x77,0x39,0x05]
# GFX12-FAKE16: v_cmpx_f_f16 v127, v127 dpp8:[7,6,5,4,3,2,1,0] ; encoding: [0xe9,0xfe,0x00,0x7d,0x7f,0x77,0x39,0x05]
Copy link
Collaborator

Choose a reason for hiding this comment

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

@broxigarchen This shouldn't be here.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Right. Seems I somehow still missed this.

I will create a new PR and cleanup the test altogether

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@kosarev kosarev kosarev left review comments

@arsenm arsenm arsenm approved these changes

@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.
5 participants
@broxigarchen @llvmbot @arsenm @kosarev @Sisyph