ia32-64/x86/vcvtpd2ph.html
2025-07-08 02:23:29 -03:00

120 lines
6 KiB
HTML
Raw Permalink Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

<!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>VCVTPD2PH
— Convert Packed Double Precision FP Values to Packed FP16 Values</title></head><body><header><nav><ul><li><a href='index.html'>Index</a></li><li>December 2023</li></ul></nav></header><h1>VCVTPD2PH
— Convert Packed Double Precision FP Values to Packed 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.128.66.MAP5.W1 5A /r VCVTPD2PH xmm1{k1}{z}, xmm2/m128/m64bcst</td>
<td>A</td>
<td>V/V</td>
<td>AVX512-FP16 AVX512VL</td>
<td>Convert two packed double precision floating-point values in xmm2/m128/m64bcst to two packed FP16 values, and store the result in xmm1 subject to writemask k1.</td></tr>
<tr>
<td>EVEX.256.66.MAP5.W1 5A /r VCVTPD2PH xmm1{k1}{z}, ymm2/m256/m64bcst</td>
<td>A</td>
<td>V/V</td>
<td>AVX512-FP16 AVX512VL</td>
<td>Convert four packed double precision floating-point values in ymm2/m256/m64bcst to four packed FP16 values, and store the result in xmm1 subject to writemask k1.</td></tr>
<tr>
<td>EVEX.512.66.MAP5.W1 5A /r VCVTPD2PH xmm1{k1}{z}, zmm2/m512/m64bcst {er}</td>
<td>A</td>
<td>V/V</td>
<td>AVX512-FP16</td>
<td>Convert eight packed double precision floating-point values in zmm2/m512/m64bcst to eight packed FP16 values, and store the result in ymm1 subject to writemask 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</th>
<th>Operand 1</th>
<th>Operand 2</th>
<th>Operand 3</th>
<th>Operand 4</th></tr>
<tr>
<td>A</td>
<td>Full</td>
<td>ModRM:reg (w)</td>
<td>ModRM:r/m (r)</td>
<td>N/A</td>
<td>N/A</td></tr></table>
<h3 id="description">Description<a class="anchor" href="#description">
</a></h3>
<p>This instruction converts two, four, or eight packed double precision floating-point values in the source operand (second operand) to two, four, or eight packed FP16 values in the destination operand (first operand). When a conversion is inexact, the value returned is rounded according to the rounding control bits in the MXCSR register or the embedded rounding control bits.</p>
<p>EVEX encoded versions: The source operand is a ZMM/YMM/XMM register, a 512/256/128-bit memory location, or a 512/256/128-bit vector broadcasts from a 64-bit memory location. The destination operand is a XMM register conditionally updated with writemask k1. The upper bits (MAXVL-1:128/64/32) of the corresponding destination are zeroed.</p>
<p>EVEX.vvvv are reserved and must be 1111b otherwise instructions will #UD.</p>
<p>This instruction uses MXCSR.DAZ for handling FP64 inputs. FP16 outputs can be normal or denormal, and are not conditionally flushed to zero.</p>
<h3 id="operation">Operation<a class="anchor" href="#operation">
</a></h3>
<h4 id="vcvtpd2ph-dest--src">VCVTPD2PH DEST, SRC<a class="anchor" href="#vcvtpd2ph-dest--src">
</a></h4>
<pre>VL = 128, 256 or 512
KL := VL / 64
IF *SRC is a register* and (VL = 512) AND (EVEX.b = 1):
SET_RM(EVEX.RC)
ELSE:
SET_RM(MXCSR.RC)
FOR j := 0 TO KL-1:
IF k1[j] OR *no writemask*:
IF *SRC is memory* and EVEX.b = 1:
tsrc := SRC.double[0]
ELSE
tsrc := SRC.double[j]
DEST.fp16[j] := Convert_fp64_to_fp16(tsrc)
ELSE IF *zeroing*:
DEST.fp16[j] := 0
// else dest.fp16[j] remains unchanged
DEST[MAXVL-1:VL/4] := 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>VCVTPD2PH __m128h _mm512_cvt_roundpd_ph (__m512d a, int rounding);
</pre>
<pre>VCVTPD2PH __m128h _mm512_mask_cvt_roundpd_ph (__m128h src, __mmask8 k, __m512d a, int rounding);
</pre>
<pre>VCVTPD2PH __m128h _mm512_maskz_cvt_roundpd_ph (__mmask8 k, __m512d a, int rounding);
</pre>
<pre>VCVTPD2PH __m128h _mm_cvtpd_ph (__m128d a);
</pre>
<pre>VCVTPD2PH __m128h _mm_mask_cvtpd_ph (__m128h src, __mmask8 k, __m128d a);
</pre>
<pre>VCVTPD2PH __m128h _mm_maskz_cvtpd_ph (__mmask8 k, __m128d a);
</pre>
<pre>VCVTPD2PH __m128h _mm256_cvtpd_ph (__m256d a);
</pre>
<pre>VCVTPD2PH __m128h _mm256_mask_cvtpd_ph (__m128h src, __mmask8 k, __m256d a);
</pre>
<pre>VCVTPD2PH __m128h _mm256_maskz_cvtpd_ph (__mmask8 k, __m256d a);
</pre>
<pre>VCVTPD2PH __m128h _mm512_cvtpd_ph (__m512d a);
</pre>
<pre>VCVTPD2PH __m128h _mm512_mask_cvtpd_ph (__m128h src, __mmask8 k, __m512d a);
</pre>
<pre>VCVTPD2PH __m128h _mm512_maskz_cvtpd_ph (__mmask8 k, __m512d a);
</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.</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 2-46</span>, “Type E2 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 Developers Manual</a> for anything serious.
</p></footer></body></html>