xref: /llvm-project/llvm/docs/AMDGPU/gfx8_vdst_829fc5.rst (revision 564d47db9eb031ae1269f66311f6cb0c47153d1d)
1..
2    **************************************************
3    *                                                *
4    *   Automatically generated file, do not edit!   *
5    *                                                *
6    **************************************************
7
8.. _amdgpu_synid_gfx8_vdst_829fc5:
9
10vdst
11====
12
13Image data to be loaded by an *image_gather4* instruction.
14
15*Size:* 4 data elements by default. Each data element occupies either 32 bits or 16 bits, depending on :ref:`d16<amdgpu_synid_d16>`.
16
17:ref:`d16<amdgpu_synid_d16>` and :ref:`tfe<amdgpu_synid_tfe>` affect operand size as follows:
18
19* :ref:`d16<amdgpu_synid_d16>` has different meanings for GFX8.0 and GFX8.1:
20
21  * For GFX8.0, this modifier does not affect the size of data elements in registers. Values in registers are stored in low 16 bits, high 16 bits are unused. There is no packing.
22  * Starting from GFX8.1, this modifier specifies that values in registers are packed; each value occupies 16 bits.
23
24* :ref:`tfe<amdgpu_synid_tfe>` adds 1 dword if specified.
25
26*Operands:* :ref:`v<amdgpu_synid_v>`
27