ia32-64/x86/kshiftrw.kshiftrb.kshiftrq.kshiftrd.html

118 lines
4.4 KiB
HTML
Raw Permalink Normal View History

2025-07-08 02:23:29 -03:00
<!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>KSHIFTRW/KSHIFTRB/KSHIFTRQ/KSHIFTRD
— Shift Right Mask Registers</title></head><body><header><nav><ul><li><a href='index.html'>Index</a></li><li>December 2023</li></ul></nav></header><h1>KSHIFTRW/KSHIFTRB/KSHIFTRQ/KSHIFTRD
— Shift Right Mask Registers</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>VEX.L0.66.0F3A.W1 30 /r KSHIFTRW k1, k2, imm8</td>
<td>RRI</td>
<td>V/V</td>
<td>AVX512F</td>
<td>Shift right 16 bits in k2 by immediate and write result in k1.</td></tr>
<tr>
<td>VEX.L0.66.0F3A.W0 30 /r KSHIFTRB k1, k2, imm8</td>
<td>RRI</td>
<td>V/V</td>
<td>AVX512DQ</td>
<td>Shift right 8 bits in k2 by immediate and write result in k1.</td></tr>
<tr>
<td>VEX.L0.66.0F3A.W1 31 /r KSHIFTRQ k1, k2, imm8</td>
<td>RRI</td>
<td>V/V</td>
<td>AVX512BW</td>
<td>Shift right 64 bits in k2 by immediate and write result in k1.</td></tr>
<tr>
<td>VEX.L0.66.0F3A.W0 31 /r KSHIFTRD k1, k2, imm8</td>
<td>RRI</td>
<td>V/V</td>
<td>AVX512BW</td>
<td>Shift right 32 bits in k2 by immediate and write result in 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>Operand 1</th>
<th>Operand 2</th>
<th>Operand 3</th></tr>
<tr>
<td>RRI</td>
<td>ModRM:reg (w)</td>
<td>ModRM:r/m (r, ModRM:[7:6] must be 11b)</td>
<td>imm8</td></tr></table>
<h2 id="description">Description<a class="anchor" href="#description">
</a></h2>
<p>Shifts 8/16/32/64 bits in the second operand (source operand) right by the count specified in immediate and place the least significant 8/16/32/64 bits of the result in the destination operand. The higher bits of the destination are zero-extended. The destination is set to zero if the count value is greater than 7 (for byte shift), 15 (for word shift), 31 (for doubleword shift) or 63 (for quadword shift).</p>
<h2 id="operation">Operation<a class="anchor" href="#operation">
</a></h2>
<h3 id="kshiftrw">KSHIFTRW<a class="anchor" href="#kshiftrw">
</a></h3>
<pre>COUNT := imm8[7:0]
DEST[MAX_KL-1:0] := 0
IF COUNT &lt;=15
THEN DEST[15:0] := SRC1[15:0] &gt;&gt; COUNT;
FI;
</pre>
<h3 id="kshiftrb">KSHIFTRB<a class="anchor" href="#kshiftrb">
</a></h3>
<pre>COUNT := imm8[7:0]
DEST[MAX_KL-1:0] := 0
IF COUNT &lt;=7
THEN DEST[7:0] := SRC1[7:0] &gt;&gt; COUNT;
FI;
</pre>
<h3 id="kshiftrq">KSHIFTRQ<a class="anchor" href="#kshiftrq">
</a></h3>
<pre>COUNT := imm8[7:0]
DEST[MAX_KL-1:0] := 0
IF COUNT &lt;=63
THEN DEST[63:0] := SRC1[63:0] &gt;&gt; COUNT;
FI;
</pre>
<h3 id="kshiftrd">KSHIFTRD<a class="anchor" href="#kshiftrd">
</a></h3>
<pre>COUNT := imm8[7:0]
DEST[MAX_KL-1:0] := 0
IF COUNT &lt;=31
THEN DEST[31:0] := SRC1[31:0] &gt;&gt; COUNT;
FI;
</pre>
<h2 id="intel-c-c++-compiler-intrinsic-equivalent">Intel C/C++ Compiler Intrinsic Equivalent<a class="anchor" href="#intel-c-c++-compiler-intrinsic-equivalent">
</a></h2>
<pre>Compiler auto generates KSHIFTRW when needed.
</pre>
<h2 id="flags-affected">Flags Affected<a class="anchor" href="#flags-affected">
</a></h2>
<p>None.</p>
<h2 class="exceptions" id="simd-floating-point-exceptions">SIMD Floating-Point Exceptions<a class="anchor" href="#simd-floating-point-exceptions">
</a></h2>
<p>None.</p>
<h2 class="exceptions" id="other-exceptions">Other Exceptions<a class="anchor" href="#other-exceptions">
</a></h2>
<p>See <span class="not-imported">Table 2-63</span>, “TYPE K20 Exception Definition (VEX-Encoded OpMask Instructions w/o Memory Arg).”</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 Developers Manual</a> for anything serious.
</p></footer></body></html>