[llvm] [AMDGPU][True16][MC] support more VOP3 inst in true16/fake16 format (PR #113603)
Brox Chen via llvm-commits
llvm-commits at lists.llvm.org
Sun Nov 17 21:17:04 PST 2024
================
@@ -5448,90 +5583,90 @@ v_maximum_f32 v5, -v1, |v2| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1
v_maximum_f32 v255, -|v255|, -|v255| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
// GFX12: [0xff,0x03,0x66,0xd7,0xfa,0xfe,0x03,0x60,0xff,0x6f,0x05,0x30]
-v_minimum_f16 v5, v1, v2 quad_perm:[3,2,1,0]
-// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x1b,0x00,0xff]
+v_minimum_f16 v255.l, -|v255.l|, -|v255.l| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
+// GFX12: [0xff,0x03,0x67,0xd7,0xfa,0xfe,0x03,0x60,0xff,0x6f,0x05,0x30]
-v_minimum_f16 v5, v1, v2 quad_perm:[0,1,2,3]
+v_minimum_f16 v5.l, -v1.l, |v2.l| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
+// GFX12: [0x05,0x02,0x67,0xd7,0xfa,0x04,0x02,0x20,0x01,0x60,0x09,0x13]
+
+v_minimum_f16 v5.l, v1.l, v2.l quad_perm:[0,1,2,3]
// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0xe4,0x00,0xff]
-v_minimum_f16 v5, v1, v2 row_mirror
-// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x40,0x01,0xff]
+v_minimum_f16 v5.l, v1.l, v2.l quad_perm:[3,2,1,0]
+// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x1b,0x00,0xff]
-v_minimum_f16 v5, v1, v2 row_half_mirror
+v_minimum_f16 v5.l, v1.l, v2.l row_half_mirror
// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x41,0x01,0xff]
-v_minimum_f16 v5, v1, v2 row_shl:1
+v_minimum_f16 v5.l, v1.l, v2.l row_mirror
+// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x40,0x01,0xff]
+
+v_minimum_f16 v5.l, v1.l, v2.l row_ror:1
+// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x21,0x01,0xff]
+
+v_minimum_f16 v5.l, v1.l, v2.l row_ror:15
+// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x2f,0x01,0xff]
+
+v_minimum_f16 v5.l, v1.l, v2.l row_share:0 row_mask:0xf bank_mask:0xf
+// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x50,0x01,0xff]
+
+v_minimum_f16 v5.l, v1.l, v2.l row_shl:1
// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x01,0x01,0xff]
-v_minimum_f16 v5, v1, v2 row_shl:15
+v_minimum_f16 v5.l, v1.l, v2.l row_shl:15
// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x0f,0x01,0xff]
-v_minimum_f16 v5, v1, v2 row_shr:1
+v_minimum_f16 v5.l, v1.l, v2.l row_shr:1
// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x11,0x01,0xff]
-v_minimum_f16 v5, v1, v2 row_shr:15
+v_minimum_f16 v5.l, v1.l, v2.l row_shr:15
// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x1f,0x01,0xff]
-v_minimum_f16 v5, v1, v2 row_ror:1
-// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x21,0x01,0xff]
-
-v_minimum_f16 v5, v1, v2 row_ror:15
-// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x2f,0x01,0xff]
-
-v_minimum_f16 v5, v1, v2 row_share:0 row_mask:0xf bank_mask:0xf
-// GFX12: [0x05,0x00,0x67,0xd7,0xfa,0x04,0x02,0x00,0x01,0x50,0x01,0xff]
-
-v_minimum_f16 v5, |v1|, -v2 row_share:15 row_mask:0x0 bank_mask:0x1
+v_minimum_f16 v5.l, |v1.l|, -v2.l row_share:15 row_mask:0x0 bank_mask:0x1
// GFX12: [0x05,0x01,0x67,0xd7,0xfa,0x04,0x02,0x40,0x01,0x5f,0x01,0x01]
-v_minimum_f16 v5, -v1, |v2| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
-// GFX12: [0x05,0x02,0x67,0xd7,0xfa,0x04,0x02,0x20,0x01,0x60,0x09,0x13]
+v_maximum_f16 v255.l, -|v255.l|, -|v255.l| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
+// GFX12: [0xff,0x03,0x68,0xd7,0xfa,0xfe,0x03,0x60,0xff,0x6f,0x05,0x30]
-v_minimum_f16 v255, -|v255|, -|v255| row_xmask:15 row_mask:0x3 bank_mask:0x0 bound_ctrl:0 fi:1
-// GFX12: [0xff,0x03,0x67,0xd7,0xfa,0xfe,0x03,0x60,0xff,0x6f,0x05,0x30]
+v_maximum_f16 v5.l, -v1.l, |v2.l| row_xmask:0 row_mask:0x1 bank_mask:0x3 bound_ctrl:1 fi:0
+// GFX12: [0x05,0x02,0x68,0xd7,0xfa,0x04,0x02,0x20,0x01,0x60,0x09,0x13]
+
+v_maximum_f16 v5.l, v1.l, v2.l quad_perm:[0,1,2,3]
+// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0xe4,0x00,0xff]
-v_maximum_f16 v5, v1, v2 quad_perm:[3,2,1,0]
+v_maximum_f16 v5.l, v1.l, v2.l quad_perm:[3,2,1,0]
// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x1b,0x00,0xff]
-v_maximum_f16 v5, v1, v2 quad_perm:[0,1,2,3]
-// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0xe4,0x00,0xff]
+v_maximum_f16 v5.l, v1.l, v2.l row_half_mirror
+// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x41,0x01,0xff]
-v_maximum_f16 v5, v1, v2 row_mirror
+v_maximum_f16 v5.l, v1.l, v2.l row_mirror
// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x40,0x01,0xff]
-v_maximum_f16 v5, v1, v2 row_half_mirror
-// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x41,0x01,0xff]
+v_maximum_f16 v5.l, v1.l, v2.l row_ror:1
+// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x21,0x01,0xff]
+
+v_maximum_f16 v5.l, v1.l, v2.l row_ror:15
+// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x2f,0x01,0xff]
+
+v_maximum_f16 v5.l, v1.l, v2.l row_share:0 row_mask:0xf bank_mask:0xf
+// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x50,0x01,0xff]
-v_maximum_f16 v5, v1, v2 row_shl:1
+v_maximum_f16 v5.l, v1.l, v2.l row_shl:1
// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x01,0x01,0xff]
-v_maximum_f16 v5, v1, v2 row_shl:15
+v_maximum_f16 v5.l, v1.l, v2.l row_shl:15
// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x0f,0x01,0xff]
-v_maximum_f16 v5, v1, v2 row_shr:1
+v_maximum_f16 v5.l, v1.l, v2.l row_shr:1
// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x11,0x01,0xff]
-v_maximum_f16 v5, v1, v2 row_shr:15
+v_maximum_f16 v5.l, v1.l, v2.l row_shr:15
// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x1f,0x01,0xff]
-v_maximum_f16 v5, v1, v2 row_ror:1
-// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x21,0x01,0xff]
-
-v_maximum_f16 v5, v1, v2 row_ror:15
-// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x2f,0x01,0xff]
-
-v_maximum_f16 v5, v1, v2 row_share:0 row_mask:0xf bank_mask:0xf
-// GFX12: [0x05,0x00,0x68,0xd7,0xfa,0x04,0x02,0x00,0x01,0x50,0x01,0xff]
-
-v_maximum_f16 v5, |v1|, -v2 row_share:15 row_mask:0x0 bank_mask:0x1
+v_maximum_f16 v5.l, |v1.l|, -v2.l row_share:15 row_mask:0x0 bank_mask:0x1
----------------
broxigarchen wrote:
added
https://github.com/llvm/llvm-project/pull/113603
More information about the llvm-commits
mailing list