129 lines
6 KiB
HTML
129 lines
6 KiB
HTML
|
<!DOCTYPE html>
|
|||
|
|
<html xmlns="http://www.w3.org/1999/xhtml" xmlns:svg="http://www.w3.org/2000/svg" xmlns:x86="http://www.felixcloutier.com/x86"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><link rel="stylesheet" type="text/css" href="style.css"></link><title>VPCOMPRESSQ
|
|||
|
|
— Store Sparse Packed Quadword Integer Values Into Dense Memory/Register</title></head><body><header><nav><ul><li><a href='index.html'>Index</a></li><li>December 2023</li></ul></nav></header><h1>VPCOMPRESSQ
|
|||
|
|
— Store Sparse Packed Quadword Integer Values Into Dense Memory/Register</h1>
|
|||
|
|
|
|||
|
|
<table>
|
|||
|
|
<tr>
|
|||
|
|
<th>Opcode/Instruction</th>
|
|||
|
|
<th>Op/En</th>
|
|||
|
|
<th>64/32 bit Mode Support</th>
|
|||
|
|
<th>CPUID Feature Flag</th>
|
|||
|
|
<th>Description</th></tr>
|
|||
|
|
<tr>
|
|||
|
|
<td>EVEX.128.66.0F38.W1 8B /r VPCOMPRESSQ xmm1/m128 {k1}{z}, xmm2</td>
|
|||
|
|
<td>A</td>
|
|||
|
|
<td>V/V</td>
|
|||
|
|
<td>AVX512VL AVX512F</td>
|
|||
|
|
<td>Compress packed quadword integer values from xmm2 to xmm1/m128 using control mask k1.</td></tr>
|
|||
|
|
<tr>
|
|||
|
|
<td>EVEX.256.66.0F38.W1 8B /r VPCOMPRESSQ ymm1/m256 {k1}{z}, ymm2</td>
|
|||
|
|
<td>A</td>
|
|||
|
|
<td>V/V</td>
|
|||
|
|
<td>AVX512VL AVX512F</td>
|
|||
|
|
<td>Compress packed quadword integer values from ymm2 to ymm1/m256 using control mask k1.</td></tr>
|
|||
|
|
<tr>
|
|||
|
|
<td>EVEX.512.66.0F38.W1 8B /r VPCOMPRESSQ zmm1/m512 {k1}{z}, zmm2</td>
|
|||
|
|
<td>A</td>
|
|||
|
|
<td>V/V</td>
|
|||
|
|
<td>AVX512F</td>
|
|||
|
|
<td>Compress packed quadword integer values from zmm2 to zmm1/m512 using control mask k1.</td></tr></table>
|
|||
|
|
<h2 id="instruction-operand-encoding">Instruction Operand Encoding<a class="anchor" href="#instruction-operand-encoding">
|
|||
|
|
¶
|
|||
|
|
</a></h2>
|
|||
|
|
<table>
|
|||
|
|
<tr>
|
|||
|
|
<th>Op/En</th>
|
|||
|
|
<th>Tuple Type</th>
|
|||
|
|
<th>Operand 1</th>
|
|||
|
|
<th>Operand 2</th>
|
|||
|
|
<th>Operand 3</th>
|
|||
|
|
<th>Operand 4</th></tr>
|
|||
|
|
<tr>
|
|||
|
|
<td>A</td>
|
|||
|
|
<td>Tuple1 Scalar</td>
|
|||
|
|
<td>ModRM:r/m (w)</td>
|
|||
|
|
<td>ModRM:reg (r)</td>
|
|||
|
|
<td>N/A</td>
|
|||
|
|
<td>N/A</td></tr></table>
|
|||
|
|
<h3 id="description">Description<a class="anchor" href="#description">
|
|||
|
|
¶
|
|||
|
|
</a></h3>
|
|||
|
|
<p>Compress (stores) up to 8/4/2 quadword integer values from the source operand (second operand) to the destination operand (first operand). The source operand is a ZMM/YMM/XMM register, the destination operand can be a ZMM/YMM/XMM register or a 512/256/128-bit memory location.</p>
|
|||
|
|
<p>The opmask register k1 selects the active elements (partial vector or possibly non-contiguous if less than 8 active elements) from the source operand to compress into a contiguous vector. The contiguous vector is written to the destination starting from the low element of the destination operand.</p>
|
|||
|
|
<p>Memory destination version: Only the contiguous vector is written to the destination memory location. EVEX.z must be zero.</p>
|
|||
|
|
<p>Register destination version: If the vector length of the contiguous vector is less than that of the input vector in the source operand, the upper bits of the destination register are unmodified if EVEX.z is not set, otherwise the upper bits are zeroed.</p>
|
|||
|
|
<p>Note: EVEX.vvvv is reserved and must be 1111b otherwise instructions will #UD.</p>
|
|||
|
|
<p>Note that the compressed displacement assumes a pre-scaling (N) corresponding to the size of one single element instead of the size of the full vector.</p>
|
|||
|
|
<h3 id="operation">Operation<a class="anchor" href="#operation">
|
|||
|
|
¶
|
|||
|
|
</a></h3>
|
|||
|
|
<h4 id="vpcompressq--evex-encoded-versions--store-form">VPCOMPRESSQ (EVEX encoded versions) store form<a class="anchor" href="#vpcompressq--evex-encoded-versions--store-form">
|
|||
|
|
¶
|
|||
|
|
</a></h4>
|
|||
|
|
<pre>(KL, VL) = (2, 128), (4, 256), (8, 512)
|
|||
|
|
SIZE := 64
|
|||
|
|
k := 0
|
|||
|
|
FOR j := 0 TO KL-1
|
|||
|
|
i := j * 64
|
|||
|
|
IF k1[j] OR *no controlmask*
|
|||
|
|
THEN
|
|||
|
|
DEST[k+SIZE-1:k] := SRC[i+63:i]
|
|||
|
|
k := k + SIZE
|
|||
|
|
FI;
|
|||
|
|
ENFOR
|
|||
|
|
</pre>
|
|||
|
|
<h4 id="vpcompressq--evex-encoded-versions--reg-reg-form">VPCOMPRESSQ (EVEX encoded versions) reg-reg form<a class="anchor" href="#vpcompressq--evex-encoded-versions--reg-reg-form">
|
|||
|
|
¶
|
|||
|
|
</a></h4>
|
|||
|
|
<pre>(KL, VL) = (2, 128), (4, 256), (8, 512)
|
|||
|
|
SIZE := 64
|
|||
|
|
k := 0
|
|||
|
|
FOR j := 0 TO KL-1
|
|||
|
|
i := j * 64
|
|||
|
|
IF k1[j] OR *no controlmask*
|
|||
|
|
THEN
|
|||
|
|
DEST[k+SIZE-1:k] := SRC[i+63:i]
|
|||
|
|
k := k + SIZE
|
|||
|
|
FI;
|
|||
|
|
ENDFOR
|
|||
|
|
IF *merging-masking*
|
|||
|
|
THEN *DEST[VL-1:k] remains unchanged*
|
|||
|
|
ELSE DEST[VL-1:k] := 0
|
|||
|
|
FI
|
|||
|
|
DEST[MAXVL-1:VL] := 0
|
|||
|
|
</pre>
|
|||
|
|
<h3 id="intel-c-c++-compiler-intrinsic-equivalent">Intel C/C++ Compiler Intrinsic Equivalent<a class="anchor" href="#intel-c-c++-compiler-intrinsic-equivalent">
|
|||
|
|
¶
|
|||
|
|
</a></h3>
|
|||
|
|
<pre>VPCOMPRESSQ __m512i _mm512_mask_compress_epi64(__m512i s, __mmask8 c, __m512i a);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VPCOMPRESSQ __m512i _mm512_maskz_compress_epi64( __mmask8 c, __m512i a);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VPCOMPRESSQ void _mm512_mask_compressstoreu_epi64(void * a, __mmask8 c, __m512i s);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VPCOMPRESSQ __m256i _mm256_mask_compress_epi64(__m256i s, __mmask8 c, __m256i a);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VPCOMPRESSQ __m256i _mm256_maskz_compress_epi64( __mmask8 c, __m256i a);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VPCOMPRESSQ void _mm256_mask_compressstoreu_epi64(void * a, __mmask8 c, __m256i s);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VPCOMPRESSQ __m128i _mm_mask_compress_epi64(__m128i s, __mmask8 c, __m128i a);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VPCOMPRESSQ __m128i _mm_maskz_compress_epi64( __mmask8 c, __m128i a);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VPCOMPRESSQ void _mm_mask_compressstoreu_epi64(void * a, __mmask8 c, __m128i s);
|
|||
|
|
</pre>
|
|||
|
|
<h3 class="exceptions" id="simd-floating-point-exceptions">SIMD Floating-Point Exceptions<a class="anchor" href="#simd-floating-point-exceptions">
|
|||
|
|
¶
|
|||
|
|
</a></h3>
|
|||
|
|
<p>None</p>
|
|||
|
|
<h3 class="exceptions" id="other-exceptions">Other Exceptions<a class="anchor" href="#other-exceptions">
|
|||
|
|
¶
|
|||
|
|
</a></h3>
|
|||
|
|
<p>EVEX-encoded instruction, see Exceptions Type E4.nb in <span class="not-imported">Table 2-49</span>, “Type E4 Class Exception Conditions.”</p><footer><p>
|
|||
|
|
This UNOFFICIAL, mechanically-separated, non-verified reference is provided for convenience, but it may be
|
|||
|
|
inc<span style="opacity: 0.2">omp</span>lete or b<sub>r</sub>oke<sub>n</sub> in various obvious or non-obvious
|
|||
|
|
ways. Refer to <a href="https://software.intel.com/en-us/download/intel-64-and-ia-32-architectures-sdm-combined-volumes-1-2a-2b-2c-2d-3a-3b-3c-3d-and-4">Intel® 64 and IA-32 Architectures Software Developer’s Manual</a> for anything serious.
|
|||
|
|
</p></footer></body></html>
|