Run DCE after a LoopFlatten test to reduce spurious output [nfc]
[llvm-project.git] / llvm / docs / AMDGPU / gfx1030_vaddr_c5ab43.rst
blobbe28c306dd54260ef9fc1c6da42b9d5abf48209e
1 ..
2     **************************************************
3     *                                                *
4     *   Automatically generated file, do not edit!   *
5     *                                                *
6     **************************************************
8 .. _amdgpu_synid_gfx1030_vaddr_c5ab43:
10 vaddr
11 =====
13 Image address which includes from one to four dimensional coordinates and other data used to locate a position in the image.
15 This operand may be specified using either :ref:`standard VGPR syntax<amdgpu_synid_v>` or special :ref:`NSA VGPR syntax<amdgpu_synid_nsa>`.
17 *Size:* 8-12 dwords. Actual size depends on opcode and :ref:`a16<amdgpu_synid_a16>`.
21   Examples:
23   .. parsed-literal::
25     image_bvh_intersect_ray   v[4:7], v[9:16], s[4:7]
26     image_bvh64_intersect_ray v[5:8], v[1:12], s[8:11]
27     image_bvh_intersect_ray   v[39:42], [v5, v4, v2, v1, v7, v3, v0, v6], s[12:15] a16
28     image_bvh64_intersect_ray v[39:42], [v50, v46, v23, v17, v16, v15, v21, v20, v19, v37, v40, v42], s[12:15]
30 *Operands:* :ref:`v<amdgpu_synid_v>`