[llvm] [AMDGPU][GFX12] Add tests for {image/flat}_atomic_pk (PR #78683)
Mariusz Sikora via llvm-commits
llvm-commits at lists.llvm.org
Fri Jan 19 00:59:04 PST 2024
https://github.com/mariusz-sikora-at-amd created https://github.com/llvm/llvm-project/pull/78683
None
>From d93d98a363244b9dd972c6d21cfcd757c088ba1b Mon Sep 17 00:00:00 2001
From: Mariusz Sikora <mariusz.sikora at amd.com>
Date: Fri, 19 Jan 2024 09:45:02 +0100
Subject: [PATCH] [AMDGPU][GFX12] Add tests for {image/flat}_atomic_pk
---
llvm/test/MC/AMDGPU/gfx12_asm_vflat.s | 42 +++++++++++++++
llvm/test/MC/AMDGPU/gfx12_asm_vimage.s | 54 +++++++++++++++++++
.../Disassembler/AMDGPU/gfx12_dasm_vflat.txt | 42 +++++++++++++++
.../Disassembler/AMDGPU/gfx12_dasm_vimage.txt | 54 +++++++++++++++++++
4 files changed, 192 insertions(+)
diff --git a/llvm/test/MC/AMDGPU/gfx12_asm_vflat.s b/llvm/test/MC/AMDGPU/gfx12_asm_vflat.s
index bac3655d19ec69..c814b42c127552 100644
--- a/llvm/test/MC/AMDGPU/gfx12_asm_vflat.s
+++ b/llvm/test/MC/AMDGPU/gfx12_asm_vflat.s
@@ -432,6 +432,48 @@ flat_atomic_xor_b64 v[1:2], v[0:1], v[2:3] offset:-64 th:TH_ATOMIC_RETURN
flat_atomic_xor_b64 v[1:2], v[0:1], v[2:3] offset:64 th:TH_ATOMIC_RETURN
// GFX12: encoding: [0x7c,0xc0,0x12,0xec,0x01,0x00,0x10,0x01,0x00,0x40,0x00,0x00]
+flat_atomic_pk_add_f16 v1, v[0:1], v2 offset:-8000000 th:TH_ATOMIC_RETURN
+// GFX12: encoding: [0x7c,0x40,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x00,0xee,0x85]
+
+flat_atomic_pk_add_f16 v1, v[0:1], v2 offset:8000000 th:TH_ATOMIC_RETURN
+// GFX12: encoding: [0x7c,0x40,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x00,0x12,0x7a]
+
+flat_atomic_pk_add_f16 v1, v[0:1], v2 offset:-64 th:TH_ATOMIC_RETURN
+// GFX12: encoding: [0x7c,0x40,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0xc0,0xff,0xff]
+
+flat_atomic_pk_add_f16 v1, v[0:1], v2 offset:64 th:TH_ATOMIC_RETURN
+// GFX12: encoding: [0x7c,0x40,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x40,0x00,0x00]
+
+flat_atomic_pk_add_f16 v[0:1], v2 offset:-64
+// GFX12: encoding: [0x7c,0x40,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0xc0,0xff,0xff]
+
+flat_atomic_pk_add_f16 v[0:1], v2 offset:64
+// GFX12: encoding: [0x7c,0x40,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0x40,0x00,0x00]
+
+flat_atomic_pk_add_f16 v[0:1], v2
+// GFX12: encoding: [0x7c,0x40,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0x00,0x00,0x00]
+
+flat_atomic_pk_add_bf16 v1, v[0:1], v2 offset:-8000000 th:TH_ATOMIC_RETURN
+// GFX12: encoding: [0x7c,0x80,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x00,0xee,0x85]
+
+flat_atomic_pk_add_bf16 v1, v[0:1], v2 offset:8000000 th:TH_ATOMIC_RETURN
+// GFX12: encoding: [0x7c,0x80,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x00,0x12,0x7a]
+
+flat_atomic_pk_add_bf16 v1, v[0:1], v2 offset:-64 th:TH_ATOMIC_RETURN
+// GFX12: encoding: [0x7c,0x80,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0xc0,0xff,0xff]
+
+flat_atomic_pk_add_bf16 v1, v[0:1], v2 offset:64 th:TH_ATOMIC_RETURN
+// GFX12: encoding: [0x7c,0x80,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x40,0x00,0x00]
+
+flat_atomic_pk_add_bf16 v[0:1], v2 offset:-64
+// GFX12: encoding: [0x7c,0x80,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0xc0,0xff,0xff]
+
+flat_atomic_pk_add_bf16 v[0:1], v2 offset:64
+// GFX12: encoding: [0x7c,0x80,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0x40,0x00,0x00]
+
+flat_atomic_pk_add_bf16 v[0:1], v2
+// GFX12: encoding: [0x7c,0x80,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0x00,0x00,0x00]
+
flat_load_b128 v[1:4], v[0:1] offset:-64
// GFX12: encoding: [0x7c,0xc0,0x05,0xec,0x01,0x00,0x00,0x00,0x00,0xc0,0xff,0xff]
diff --git a/llvm/test/MC/AMDGPU/gfx12_asm_vimage.s b/llvm/test/MC/AMDGPU/gfx12_asm_vimage.s
index c06fb614b88b10..0d7237922c3ebb 100644
--- a/llvm/test/MC/AMDGPU/gfx12_asm_vimage.s
+++ b/llvm/test/MC/AMDGPU/gfx12_asm_vimage.s
@@ -1018,6 +1018,60 @@ image_atomic_max_flt v[254:255], [v4, v5, v6, v7], s[96:103] dmask:0x3 dim:SQ_RS
image_atomic_max_flt v0, v0, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_NT
// GFX12: encoding: [0x00,0x40,0x61,0xd0,0x00,0x00,0x20,0x00,0x00,0x00,0x00,0x00]
+image_atomic_pk_add_f16 v0, v0, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D
+// GFX12: encoding: [0x00,0x80,0x61,0xd0,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00]
+
+image_atomic_pk_add_f16 v1, [v2, v3], s[4:11] dmask:0x1 dim:SQ_RSRC_IMG_2D
+// GFX12: encoding: [0x01,0x80,0x61,0xd0,0x01,0x08,0x00,0x00,0x02,0x03,0x00,0x00]
+
+image_atomic_pk_add_f16 v4, [v4, v5, v6], s[8:15] dmask:0x1 dim:SQ_RSRC_IMG_3D
+// GFX12: encoding: [0x02,0x80,0x61,0xd0,0x04,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+
+image_atomic_pk_add_f16 v255, [v4, v5, v6], s[8:15] dmask:0x1 dim:SQ_RSRC_IMG_CUBE
+// GFX12: encoding: [0x03,0x80,0x61,0xd0,0xff,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+
+image_atomic_pk_add_f16 v[0:1], [v4, v5], s[8:15] dmask:0x3 dim:SQ_RSRC_IMG_1D_ARRAY
+// GFX12: encoding: [0x04,0x80,0xe1,0xd0,0x00,0x10,0x00,0x00,0x04,0x05,0x00,0x00]
+
+image_atomic_pk_add_f16 v[1:2], [v4, v5, v6], s[8:15] dmask:0x3 dim:SQ_RSRC_IMG_2D_ARRAY
+// GFX12: encoding: [0x05,0x80,0xe1,0xd0,0x01,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+
+image_atomic_pk_add_f16 v[3:4], [v4, v5, v6], s[8:15] dmask:0x3 dim:SQ_RSRC_IMG_2D_MSAA
+// GFX12: encoding: [0x06,0x80,0xe1,0xd0,0x03,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+
+image_atomic_pk_add_f16 v[254:255], [v4, v5, v6, v7], s[96:103] dmask:0x3 dim:SQ_RSRC_IMG_2D_MSAA_ARRAY
+// GFX12: encoding: [0x07,0x80,0xe1,0xd0,0xfe,0xc0,0x00,0x00,0x04,0x05,0x06,0x07]
+
+image_atomic_pk_add_f16 v0, v0, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_NT
+// GFX12: encoding: [0x00,0x80,0x61,0xd0,0x00,0x00,0x20,0x00,0x00,0x00,0x00,0x00]
+
+image_atomic_pk_add_bf16 v0, v0, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D
+// GFX12: encoding: [0x00,0xc0,0x61,0xd0,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00]
+
+image_atomic_pk_add_bf16 v1, [v2, v3], s[4:11] dmask:0x1 dim:SQ_RSRC_IMG_2D
+// GFX12: encoding: [0x01,0xc0,0x61,0xd0,0x01,0x08,0x00,0x00,0x02,0x03,0x00,0x00]
+
+image_atomic_pk_add_bf16 v4, [v4, v5, v6], s[8:15] dmask:0x1 dim:SQ_RSRC_IMG_3D
+// GFX12: encoding: [0x02,0xc0,0x61,0xd0,0x04,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+
+image_atomic_pk_add_bf16 v255, [v4, v5, v6], s[8:15] dmask:0x1 dim:SQ_RSRC_IMG_CUBE
+// GFX12: encoding: [0x03,0xc0,0x61,0xd0,0xff,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+
+image_atomic_pk_add_bf16 v[0:1], [v4, v5], s[8:15] dmask:0x3 dim:SQ_RSRC_IMG_1D_ARRAY
+// GFX12: encoding: [0x04,0xc0,0xe1,0xd0,0x00,0x10,0x00,0x00,0x04,0x05,0x00,0x00]
+
+image_atomic_pk_add_bf16 v[1:2], [v4, v5, v6], s[8:15] dmask:0x3 dim:SQ_RSRC_IMG_2D_ARRAY
+// GFX12: encoding: [0x05,0xc0,0xe1,0xd0,0x01,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+
+image_atomic_pk_add_bf16 v[3:4], [v4, v5, v6], s[8:15] dmask:0x3 dim:SQ_RSRC_IMG_2D_MSAA
+// GFX12: encoding: [0x06,0xc0,0xe1,0xd0,0x03,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+
+image_atomic_pk_add_bf16 v[254:255], [v4, v5, v6, v7], s[96:103] dmask:0x3 dim:SQ_RSRC_IMG_2D_MSAA_ARRAY
+// GFX12: encoding: [0x07,0xc0,0xe1,0xd0,0xfe,0xc0,0x00,0x00,0x04,0x05,0x06,0x07]
+
+image_atomic_pk_add_bf16 v0, v0, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_NT
+// GFX12: encoding: [0x00,0xc0,0x61,0xd0,0x00,0x00,0x20,0x00,0x00,0x00,0x00,0x00]
+
image_bvh_intersect_ray v[4:7], [v9, v10, v[11:13], v[14:16], v[17:19]], s[4:7]
// GFX12: encoding: [0x10,0x40,0xc6,0xd3,0x04,0x08,0x00,0x11,0x09,0x0a,0x0b,0x0e]
diff --git a/llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vflat.txt b/llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vflat.txt
index b579297235d85e..181f2bd646069e 100644
--- a/llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vflat.txt
+++ b/llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vflat.txt
@@ -240,6 +240,48 @@
# GFX12: flat_atomic_xor_b64 v[1:2], v[0:1], v[2:3] offset:64 th:TH_ATOMIC_RETURN ; encoding: [0x7c,0xc0,0x12,0xec,0x01,0x00,0x10,0x01,0x00,0x40,0x00,0x00]
0x7c,0xc0,0x12,0xec,0x01,0x00,0x10,0x01,0x00,0x40,0x00,0x00
+# GFX12: flat_atomic_pk_add_f16 v1, v[0:1], v2 offset:-8000000 th:TH_ATOMIC_RETURN ; encoding: [0x7c,0x40,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x00,0xee,0x85]
+0x7c,0x40,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x00,0xee,0x85
+
+# GFX12: flat_atomic_pk_add_f16 v1, v[0:1], v2 offset:8000000 th:TH_ATOMIC_RETURN ; encoding: [0x7c,0x40,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x00,0x12,0x7a]
+0x7c,0x40,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x00,0x12,0x7a
+
+# GFX12: flat_atomic_pk_add_f16 v1, v[0:1], v2 offset:-64 th:TH_ATOMIC_RETURN ; encoding: [0x7c,0x40,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0xc0,0xff,0xff]
+0x7c,0x40,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0xc0,0xff,0xff
+
+# GFX12: flat_atomic_pk_add_f16 v1, v[0:1], v2 offset:64 th:TH_ATOMIC_RETURN ; encoding: [0x7c,0x40,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x40,0x00,0x00]
+0x7c,0x40,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x40,0x00,0x00
+
+# GFX12: flat_atomic_pk_add_f16 v[0:1], v2 offset:-64 ; encoding: [0x7c,0x40,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0xc0,0xff,0xff]
+0x7c,0x40,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0xc0,0xff,0xff
+
+# GFX12: flat_atomic_pk_add_f16 v[0:1], v2 offset:64 ; encoding: [0x7c,0x40,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0x40,0x00,0x00]
+0x7c,0x40,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0x40,0x00,0x00
+
+# GFX12: flat_atomic_pk_add_f16 v[0:1], v2 ; encoding: [0x7c,0x40,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0x00,0x00,0x00]
+0x7c,0x40,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0x00,0x00,0x00
+
+# GFX12: flat_atomic_pk_add_bf16 v1, v[0:1], v2 offset:-8000000 th:TH_ATOMIC_RETURN ; encoding: [0x7c,0x80,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x00,0xee,0x85]
+0x7c,0x80,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x00,0xee,0x85
+
+# GFX12: flat_atomic_pk_add_bf16 v1, v[0:1], v2 offset:8000000 th:TH_ATOMIC_RETURN ; encoding: [0x7c,0x80,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x00,0x12,0x7a]
+0x7c,0x80,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x00,0x12,0x7a
+
+# GFX12: flat_atomic_pk_add_bf16 v1, v[0:1], v2 offset:-64 th:TH_ATOMIC_RETURN ; encoding: [0x7c,0x80,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0xc0,0xff,0xff]
+0x7c,0x80,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0xc0,0xff,0xff
+
+# GFX12: flat_atomic_pk_add_bf16 v1, v[0:1], v2 offset:64 th:TH_ATOMIC_RETURN ; encoding: [0x7c,0x80,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x40,0x00,0x00]
+0x7c,0x80,0x16,0xec,0x01,0x00,0x10,0x01,0x00,0x40,0x00,0x00
+
+# GFX12: flat_atomic_pk_add_bf16 v[0:1], v2 offset:-64 ; encoding: [0x7c,0x80,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0xc0,0xff,0xff]
+0x7c,0x80,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0xc0,0xff,0xff
+
+# GFX12: flat_atomic_pk_add_bf16 v[0:1], v2 offset:64 ; encoding: [0x7c,0x80,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0x40,0x00,0x00]
+0x7c,0x80,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0x40,0x00,0x00
+
+# GFX12: flat_atomic_pk_add_bf16 v[0:1], v2 ; encoding: [0x7c,0x80,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0x00,0x00,0x00]
+0x7c,0x80,0x16,0xec,0x00,0x00,0x00,0x01,0x00,0x00,0x00,0x00
+
# GFX12: flat_load_b128 v[1:4], v[0:1] offset:64 ; encoding: [0x7c,0xc0,0x05,0xec,0x01,0x00,0x00,0x00,0x00,0x40,0x00,0x00]
0x7c,0xc0,0x05,0xec,0x01,0x00,0x00,0x00,0x00,0x40,0x00,0x00
diff --git a/llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vimage.txt b/llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vimage.txt
index bb09f602c0ac32..fc5e04dd440861 100644
--- a/llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vimage.txt
+++ b/llvm/test/MC/Disassembler/AMDGPU/gfx12_dasm_vimage.txt
@@ -1018,6 +1018,60 @@
# GFX12: image_atomic_max_flt v0, v0, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_NT ; encoding: [0x00,0x40,0x61,0xd0,0x00,0x00,0x20,0x00,0x00,0x00,0x00,0x00]
0x00,0x40,0x61,0xd0,0x00,0x00,0x20,0x00,0x00,0x00,0x00,0x00
+# GFX12: image_atomic_pk_add_f16 v0, v0, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D ; encoding: [0x00,0x80,0x61,0xd0,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00]
+0x00,0x80,0x61,0xd0,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00
+
+# GFX12: image_atomic_pk_add_f16 v1, [v2, v3], s[4:11] dmask:0x1 dim:SQ_RSRC_IMG_2D ; encoding: [0x01,0x80,0x61,0xd0,0x01,0x08,0x00,0x00,0x02,0x03,0x00,0x00]
+0x01,0x80,0x61,0xd0,0x01,0x08,0x00,0x00,0x02,0x03,0x00,0x00
+
+# GFX12: image_atomic_pk_add_f16 v4, [v4, v5, v6], s[8:15] dmask:0x1 dim:SQ_RSRC_IMG_3D ; encoding: [0x02,0x80,0x61,0xd0,0x04,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+0x02,0x80,0x61,0xd0,0x04,0x10,0x00,0x00,0x04,0x05,0x06,0x00
+
+# GFX12: image_atomic_pk_add_f16 v255, [v4, v5, v6], s[8:15] dmask:0x1 dim:SQ_RSRC_IMG_CUBE ; encoding: [0x03,0x80,0x61,0xd0,0xff,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+0x03,0x80,0x61,0xd0,0xff,0x10,0x00,0x00,0x04,0x05,0x06,0x00
+
+# GFX12: image_atomic_pk_add_f16 v[0:1], [v4, v5], s[8:15] dmask:0x3 dim:SQ_RSRC_IMG_1D_ARRAY ; encoding: [0x04,0x80,0xe1,0xd0,0x00,0x10,0x00,0x00,0x04,0x05,0x00,0x00]
+0x04,0x80,0xe1,0xd0,0x00,0x10,0x00,0x00,0x04,0x05,0x00,0x00
+
+# GFX12: image_atomic_pk_add_f16 v[1:2], [v4, v5, v6], s[8:15] dmask:0x3 dim:SQ_RSRC_IMG_2D_ARRAY ; encoding: [0x05,0x80,0xe1,0xd0,0x01,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+0x05,0x80,0xe1,0xd0,0x01,0x10,0x00,0x00,0x04,0x05,0x06,0x00
+
+# GFX12: image_atomic_pk_add_f16 v[3:4], [v4, v5, v6], s[8:15] dmask:0x3 dim:SQ_RSRC_IMG_2D_MSAA ; encoding: [0x06,0x80,0xe1,0xd0,0x03,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+0x06,0x80,0xe1,0xd0,0x03,0x10,0x00,0x00,0x04,0x05,0x06,0x00
+
+# GFX12: image_atomic_pk_add_f16 v[254:255], [v4, v5, v6, v7], s[96:103] dmask:0x3 dim:SQ_RSRC_IMG_2D_MSAA_ARRAY ; encoding: [0x07,0x80,0xe1,0xd0,0xfe,0xc0,0x00,0x00,0x04,0x05,0x06,0x07]
+0x07,0x80,0xe1,0xd0,0xfe,0xc0,0x00,0x00,0x04,0x05,0x06,0x07
+
+# GFX12: image_atomic_pk_add_f16 v0, v0, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_NT ; encoding: [0x00,0x80,0x61,0xd0,0x00,0x00,0x20,0x00,0x00,0x00,0x00,0x00]
+0x00,0x80,0x61,0xd0,0x00,0x00,0x20,0x00,0x00,0x00,0x00,0x00
+
+# GFX12: image_atomic_pk_add_bf16 v0, v0, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D ; encoding: [0x00,0xc0,0x61,0xd0,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00]
+0x00,0xc0,0x61,0xd0,0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00
+
+# GFX12: image_atomic_pk_add_bf16 v1, [v2, v3], s[4:11] dmask:0x1 dim:SQ_RSRC_IMG_2D ; encoding: [0x01,0xc0,0x61,0xd0,0x01,0x08,0x00,0x00,0x02,0x03,0x00,0x00]
+0x01,0xc0,0x61,0xd0,0x01,0x08,0x00,0x00,0x02,0x03,0x00,0x00
+
+# GFX12: image_atomic_pk_add_bf16 v4, [v4, v5, v6], s[8:15] dmask:0x1 dim:SQ_RSRC_IMG_3D ; encoding: [0x02,0xc0,0x61,0xd0,0x04,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+0x02,0xc0,0x61,0xd0,0x04,0x10,0x00,0x00,0x04,0x05,0x06,0x00
+
+# GFX12: image_atomic_pk_add_bf16 v255, [v4, v5, v6], s[8:15] dmask:0x1 dim:SQ_RSRC_IMG_CUBE ; encoding: [0x03,0xc0,0x61,0xd0,0xff,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+0x03,0xc0,0x61,0xd0,0xff,0x10,0x00,0x00,0x04,0x05,0x06,0x00
+
+# GFX12: image_atomic_pk_add_bf16 v[0:1], [v4, v5], s[8:15] dmask:0x3 dim:SQ_RSRC_IMG_1D_ARRAY ; encoding: [0x04,0xc0,0xe1,0xd0,0x00,0x10,0x00,0x00,0x04,0x05,0x00,0x00]
+0x04,0xc0,0xe1,0xd0,0x00,0x10,0x00,0x00,0x04,0x05,0x00,0x00
+
+# GFX12: image_atomic_pk_add_bf16 v[1:2], [v4, v5, v6], s[8:15] dmask:0x3 dim:SQ_RSRC_IMG_2D_ARRAY ; encoding: [0x05,0xc0,0xe1,0xd0,0x01,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+0x05,0xc0,0xe1,0xd0,0x01,0x10,0x00,0x00,0x04,0x05,0x06,0x00
+
+# GFX12: image_atomic_pk_add_bf16 v[3:4], [v4, v5, v6], s[8:15] dmask:0x3 dim:SQ_RSRC_IMG_2D_MSAA ; encoding: [0x06,0xc0,0xe1,0xd0,0x03,0x10,0x00,0x00,0x04,0x05,0x06,0x00]
+0x06,0xc0,0xe1,0xd0,0x03,0x10,0x00,0x00,0x04,0x05,0x06,0x00
+
+# GFX12: image_atomic_pk_add_bf16 v[254:255], [v4, v5, v6, v7], s[96:103] dmask:0x3 dim:SQ_RSRC_IMG_2D_MSAA_ARRAY ; encoding: [0x07,0xc0,0xe1,0xd0,0xfe,0xc0,0x00,0x00,0x04,0x05,0x06,0x07]
+0x07,0xc0,0xe1,0xd0,0xfe,0xc0,0x00,0x00,0x04,0x05,0x06,0x07
+
+# GFX12: image_atomic_pk_add_bf16 v0, v0, s[0:7] dmask:0x1 dim:SQ_RSRC_IMG_1D th:TH_ATOMIC_NT ; encoding: [0x00,0xc0,0x61,0xd0,0x00,0x00,0x20,0x00,0x00,0x00,0x00,0x00]
+0x00,0xc0,0x61,0xd0,0x00,0x00,0x20,0x00,0x00,0x00,0x00,0x00
+
# GFX12: image_bvh_intersect_ray v[4:7], [v9, v10, v[11:13], v[14:16], v[17:19]], s[4:7] ; encoding: [0x10,0x40,0xc6,0xd3,0x04,0x08,0x00,0x11,0x09,0x0a,0x0b,0x0e]
0x10,0x40,0xc6,0xd3,0x04,0x08,0x00,0x11,0x09,0x0a,0x0b,0x0e
More information about the llvm-commits
mailing list