89 lines
4.4 KiB
HTML
89 lines
4.4 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>VCVTSS2SH
|
||
— Convert Low FP32 Value to an FP16 Value</title></head><body><header><nav><ul><li><a href='index.html'>Index</a></li><li>December 2023</li></ul></nav></header><h1>VCVTSS2SH
|
||
— Convert Low FP32 Value to an FP16 Value</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.NP.MAP5.W0 1D /r VCVTSS2SH xmm1{k1}{z}, xmm2, xmm3/m32 {er}</td>
|
||
<td>A</td>
|
||
<td>V/V</td>
|
||
<td>AVX512-FP16</td>
|
||
<td>Convert low FP32 value in xmm3/m32 to an FP16 value and store in the low element of xmm1 subject to writemask k1. Bits 127:16 from 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 converts the low FP32 value in the second source operand to a FP16 value in the low element of the destination operand.</p>
|
||
<p>When the conversion is inexact, the value returned is rounded according to the rounding control bits in the MXCSR register.</p>
|
||
<p>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="vcvtss2sh-dest--src1--src2">VCVTSS2SH dest, src1, src2<a class="anchor" href="#vcvtss2sh-dest--src1--src2">
|
||
¶
|
||
</a></h4>
|
||
<pre>IF *SRC2 is a register* and (EVEX.b = 1):
|
||
SET_RM(EVEX.RC)
|
||
ELSE:
|
||
SET_RM(MXCSR.RC)
|
||
IF k1[0] OR *no writemask*:
|
||
DEST.fp16[0] := Convert_fp32_to_fp16(SRC2.fp32[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>VCVTSS2SH __m128h _mm_cvt_roundss_sh (__m128h a, __m128 b, const int rounding);
|
||
</pre>
|
||
<pre>VCVTSS2SH __m128h _mm_mask_cvt_roundss_sh (__m128h src, __mmask8 k, __m128h a, __m128 b, const int rounding);
|
||
</pre>
|
||
<pre>VCVTSS2SH __m128h _mm_maskz_cvt_roundss_sh (__mmask8 k, __m128h a, __m128 b, const int rounding);
|
||
</pre>
|
||
<pre>VCVTSS2SH __m128h _mm_cvtss_sh (__m128h a, __m128 b);
|
||
</pre>
|
||
<pre>VCVTSS2SH __m128h _mm_mask_cvtss_sh (__m128h src, __mmask8 k, __m128h a, __m128 b);
|
||
</pre>
|
||
<pre>VCVTSS2SH __m128h _mm_maskz_cvtss_sh (__mmask8 k, __m128h a, __m128 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.</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>
|