Intel® C++ Compiler Classic Developer Guide and Reference

ID 767249
Date 12/16/2022
Public

A newer version of this document is available. Customers should click here to go to the newest version.

Document Table of Contents

<span class='option'>_mm_i32gather_ps, _mm256_i32gather_ps</span>

Gathers 2/4 packed single-precision floating point values from memory referenced by the given base address, dword indices, and scale. The corresponding Intel® AVX2 instruction is VGATHERDPS.

Syntax

extern __m128 _mm_mask_i32gather_ps(float const * base, __m128i vindex, const int scale);

extern __m256 _mm256_mask_i32gather_ps(float const * base, __m256i vindex, const int scale);

Arguments

base

the base address used to reference the loaded FP elements.

vindex

the vector of dword indices used to reference the loaded FP elements.

scale

The compilation time literal constant, which is used as the vector indices scale to address the loaded elements. Possible values are one of the following: 1, 2, 4, 8.

Description

The intrinsics load 2/4 packed single-precision floating-point values from memory using dword indices.

Below is the pseudo-code for the intrinsics:

_mm_i32gather_ps():

result[31:0] = mem[base+vindex[31:0]*scale];
result[63:32] = mem[base+vindex[63:32]*scale];
result[95:64] = mem[base+vindex[95:64]*scale];
result127:96] = mem[base+vindex[127:96]*scale];

_mm256_i32gather_ps():

result[31:0] = mem[base+vindex[31:0]*scale];
result[63:32] = mem[base+vindex[63:32]*scale];
result[95:64] = mem[base+vindex[95:64]*scale];
result127:96] = mem[base+vindex[127:96]*scale];
result[159:128] = mem[base+vindex[159:128]*scale];
result[191:160] = mem[base+vindex[191:160]*scale];
result[223:192] = mem[base+vindex[223:192]*scale];
result[255:224] = mem[base+vindex[255:224]*scale];
Returns

A 128/256-bit vector with unconditionally gathered single-precision FP values.