forked from NRZCode/ia32-64
88 lines
4.3 KiB
HTML
88 lines
4.3 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>VDIVSH
|
|||
|
|
— Divide Scalar FP16 Values</title></head><body><header><nav><ul><li><a href='index.html'>Index</a></li><li>December 2023</li></ul></nav></header><h1>VDIVSH
|
|||
|
|
— Divide Scalar FP16 Values</h1>
|
|||
|
|
|
|||
|
|
<table>
|
|||
|
|
<tr>
|
|||
|
|
<th> Instruction En Bit Mode Flag
|
|||
|
|
Support Instruction En Bit Mode Flag
|
|||
|
|
Support 64/32 CPUID Feature Instruction En Bit Mode Flag CPUID Feature Instruction En Bit Mode Flag Op/ 64/32 CPUID Feature Instruction En Bit Mode Flag 64/32 CPUID Feature Instruction En Bit Mode Flag CPUID Feature Instruction En Bit Mode Flag Op/ 64/32 CPUID Feature </th>
|
|||
|
|
<th></th>
|
|||
|
|
<th>Support</th>
|
|||
|
|
<th></th>
|
|||
|
|
<th>Description</th></tr>
|
|||
|
|
<tr>
|
|||
|
|
<td>EVEX.LLIG.F3.MAP5.W0 5E /r VDIVSH xmm1{k1}{z}, xmm2, xmm3/m16 {er}</td>
|
|||
|
|
<td>A</td>
|
|||
|
|
<td>V/V</td>
|
|||
|
|
<td>AVX512-FP16</td>
|
|||
|
|
<td>Divide low FP16 value in xmm2 by low FP16 value in xmm3/m16, and store the result in xmm1 subject to writemask k1. Bits 127:16 of xmm2 are copied to xmm1[127:16].</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</th>
|
|||
|
|
<th>Operand 1</th>
|
|||
|
|
<th>Operand 2</th>
|
|||
|
|
<th>Operand 3</th>
|
|||
|
|
<th>Operand 4</th></tr>
|
|||
|
|
<tr>
|
|||
|
|
<td>A</td>
|
|||
|
|
<td>Scalar</td>
|
|||
|
|
<td>ModRM:reg (w)</td>
|
|||
|
|
<td>VEX.vvvv (r)</td>
|
|||
|
|
<td>ModRM:r/m (r)</td>
|
|||
|
|
<td>N/A</td></tr></table>
|
|||
|
|
<h3 id="description">Description<a class="anchor" href="#description">
|
|||
|
|
¶
|
|||
|
|
</a></h3>
|
|||
|
|
<p>This instruction divides the low FP16 value from the first source operand by the corresponding value in the second source operand, storing the FP16 result in the destination operand. Bits 127:16 of the destination operand are copied from the corresponding bits of the first source operand. Bits MAXVL-1:128 of the destination operand are zeroed. The low FP16 element of the destination is updated according to the writemask.</p>
|
|||
|
|
<h3 id="operation">Operation<a class="anchor" href="#operation">
|
|||
|
|
¶
|
|||
|
|
</a></h3>
|
|||
|
|
<h4 id="vdivsh--evex-encoded-versions-">VDIVSH (EVEX Encoded Versions)<a class="anchor" href="#vdivsh--evex-encoded-versions-">
|
|||
|
|
¶
|
|||
|
|
</a></h4>
|
|||
|
|
<pre>IF EVEX.b = 1 and SRC2 is a register:
|
|||
|
|
SET_RM(EVEX.RC)
|
|||
|
|
ELSE
|
|||
|
|
SET_RM(MXCSR.RC)
|
|||
|
|
IF k1[0] OR *no writemask*:
|
|||
|
|
DEST.fp16[0] := SRC1.fp16[0] / SRC2.fp16[0]
|
|||
|
|
ELSE IF *zeroing*:
|
|||
|
|
DEST.fp16[0] := 0
|
|||
|
|
// else dest.fp16[0] remains unchanged
|
|||
|
|
DEST[127:16] := SRC1[127:16]
|
|||
|
|
DEST[MAXVL-1:128] := 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>VDIVSH __m128h _mm_div_round_sh (__m128h a, __m128h b, int rounding);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VDIVSH __m128h _mm_mask_div_round_sh (__m128h src, __mmask8 k, __m128h a, __m128h b, int rounding);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VDIVSH __m128h _mm_maskz_div_round_sh (__mmask8 k, __m128h a, __m128h b, int rounding);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VDIVSH __m128h _mm_div_sh (__m128h a, __m128h b);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VDIVSH __m128h _mm_mask_div_sh (__m128h src, __mmask8 k, __m128h a, __m128h b);
|
|||
|
|
</pre>
|
|||
|
|
<pre>VDIVSH __m128h _mm_maskz_div_sh (__mmask8 k, __m128h a, __m128h b);
|
|||
|
|
</pre>
|
|||
|
|
<h3 class="exceptions" id="simd-floating-point-exceptions">SIMD Floating-Point Exceptions<a class="anchor" href="#simd-floating-point-exceptions">
|
|||
|
|
¶
|
|||
|
|
</a></h3>
|
|||
|
|
<p>Invalid, Underflow, Overflow, Precision, Denormal, Zero.</p>
|
|||
|
|
<h3 class="exceptions" id="other-exceptions">Other Exceptions<a class="anchor" href="#other-exceptions">
|
|||
|
|
¶
|
|||
|
|
</a></h3>
|
|||
|
|
<p>EVEX-encoded instructions, see <span class="not-imported">Table 2-47</span>, “Type E3 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>
|