[ARM] Fold VCMP into VPT
[llvm-core.git] / docs / AMDGPU / gfx9_dst_mimg_gather4.rst
blob3bb6f30811f15c5703b0e41ea4c3913c3f8b72da
1 ..
2     **************************************************
3     *                                                *
4     *   Automatically generated file, do not edit!   *
5     *                                                *
6     **************************************************
8 .. _amdgpu_synid9_dst_mimg_gather4:
10 vdst
11 ===========================
13 Image data to load by an *image_gather4* instruction.
15 *Size:* 4 data elements by default. Each data element occupies either 32 bits or 16 bits depending on :ref:`d16<amdgpu_synid_d16>`.
17 :ref:`d16<amdgpu_synid_d16>` and :ref:`tfe<amdgpu_synid_tfe>` affect operand size as follows:
19 * :ref:`d16<amdgpu_synid_d16>` specifies that data elements in registers are packed; each value occupies 16 bits.
20 * :ref:`tfe<amdgpu_synid_tfe>` adds one dword if specified.
22 *Operands:* :ref:`v<amdgpu_synid_v>`