forked from NRZCode/ia32-64
137 lines
6.6 KiB
HTML
137 lines
6.6 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>MAXSD
|
||
— Return Maximum Scalar Double Precision Floating-Point Value</title></head><body><header><nav><ul><li><a href='index.html'>Index</a></li><li>December 2023</li></ul></nav></header><h1>MAXSD
|
||
— Return Maximum Scalar Double Precision Floating-Point Value</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>F2 0F 5F /r MAXSD xmm1, xmm2/m64</td>
|
||
<td>A</td>
|
||
<td>V/V</td>
|
||
<td>SSE2</td>
|
||
<td>Return the maximum scalar double precision floating-point value between xmm2/m64 and xmm1.</td></tr>
|
||
<tr>
|
||
<td>VEX.LIG.F2.0F.WIG 5F /r VMAXSD xmm1, xmm2, xmm3/m64</td>
|
||
<td>B</td>
|
||
<td>V/V</td>
|
||
<td>AVX</td>
|
||
<td>Return the maximum scalar double precision floating-point value between xmm3/m64 and xmm2.</td></tr>
|
||
<tr>
|
||
<td>EVEX.LLIG.F2.0F.W1 5F /r VMAXSD xmm1 {k1}{z}, xmm2, xmm3/m64{sae}</td>
|
||
<td>C</td>
|
||
<td>V/V</td>
|
||
<td>AVX512F</td>
|
||
<td>Return the maximum scalar double precision floating-point value between xmm3/m64 and xmm2.</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>N/A</td>
|
||
<td>ModRM:reg (r, w)</td>
|
||
<td>ModRM:r/m (r)</td>
|
||
<td>N/A</td>
|
||
<td>N/A</td></tr>
|
||
<tr>
|
||
<td>B</td>
|
||
<td>N/A</td>
|
||
<td>ModRM:reg (w)</td>
|
||
<td>VEX.vvvv (r)</td>
|
||
<td>ModRM:r/m (r)</td>
|
||
<td>N/A</td></tr>
|
||
<tr>
|
||
<td>C</td>
|
||
<td>Tuple1 Scalar</td>
|
||
<td>ModRM:reg (w)</td>
|
||
<td>EVEX.vvvv (r)</td>
|
||
<td>ModRM:r/m (r)</td>
|
||
<td>N/A</td></tr></table>
|
||
<h2 id="description">Description<a class="anchor" href="#description">
|
||
¶
|
||
</a></h2>
|
||
<p>Compares the low double precision floating-point values in the first source operand and the second source operand, and returns the maximum value to the low quadword of the destination operand. The second source operand can be an XMM register or a 64-bit memory location. The first source and destination operands are XMM registers. When the second source operand is a memory operand, only 64 bits are accessed.</p>
|
||
<p>If the values being compared are both 0.0s (of either sign), the value in the second source operand is returned. If a value in the second source operand is an SNaN, that SNaN is returned unchanged to the destination (that is, a QNaN version of the SNaN is not returned).</p>
|
||
<p>If only one value is a NaN (SNaN or QNaN) for this instruction, the second source operand, either a NaN or a valid floating-point value, is written to the result. If instead of this behavior, it is required that the NaN of either source operand be returned, the action of MAXSD can be emulated using a sequence of instructions, such as, a comparison followed by AND, ANDN, and OR.</p>
|
||
<p>128-bit Legacy SSE version: The destination and first source operand are the same. Bits (MAXVL-1:64) of the corresponding destination register remain unchanged.</p>
|
||
<p>VEX.128 and EVEX encoded version: Bits (127:64) of the XMM register destination are copied from corresponding bits in the first source operand. Bits (MAXVL-1:128) of the destination register are zeroed.</p>
|
||
<p>EVEX encoded version: The low quadword element of the destination operand is updated according to the writemask.</p>
|
||
<p>Software should ensure VMAXSD is encoded with VEX.L=0. Encoding VMAXSD with VEX.L=1 may encounter unpredictable behavior across different processor generations.</p>
|
||
<h2 id="operation">Operation<a class="anchor" href="#operation">
|
||
¶
|
||
</a></h2>
|
||
<pre>MAX(SRC1, SRC2)
|
||
{
|
||
IF ((SRC1 = 0.0) and (SRC2 = 0.0)) THEN DEST := SRC2;
|
||
ELSE IF (SRC1 = NaN) THEN DEST := SRC2; FI;
|
||
ELSE IF (SRC2 = NaN) THEN DEST := SRC2; FI;
|
||
ELSE IF (SRC1 > SRC2) THEN DEST := SRC1;
|
||
ELSE DEST := SRC2;
|
||
FI;
|
||
}
|
||
</pre>
|
||
<h3 id="vmaxsd--evex-encoded-version-">VMAXSD (EVEX Encoded Version)<a class="anchor" href="#vmaxsd--evex-encoded-version-">
|
||
¶
|
||
</a></h3>
|
||
<pre>IF k1[0] or *no writemask*
|
||
THEN DEST[63:0] := MAX(SRC1[63:0], SRC2[63:0])
|
||
ELSE
|
||
IF *merging-masking* ; merging-masking
|
||
THEN *DEST[63:0] remains unchanged*
|
||
ELSE ; zeroing-masking
|
||
DEST[63:0] := 0
|
||
FI;
|
||
FI;
|
||
DEST[127:64] := SRC1[127:64]
|
||
DEST[MAXVL-1:128] := 0
|
||
</pre>
|
||
<h3 id="vmaxsd--vex-128-encoded-version-">VMAXSD (VEX.128 Encoded Version)<a class="anchor" href="#vmaxsd--vex-128-encoded-version-">
|
||
¶
|
||
</a></h3>
|
||
<pre>DEST[63:0] := MAX(SRC1[63:0], SRC2[63:0])
|
||
DEST[127:64] := SRC1[127:64]
|
||
DEST[MAXVL-1:128] := 0
|
||
</pre>
|
||
<h3 id="maxsd--128-bit-legacy-sse-version-">MAXSD (128-bit Legacy SSE Version)<a class="anchor" href="#maxsd--128-bit-legacy-sse-version-">
|
||
¶
|
||
</a></h3>
|
||
<pre>DEST[63:0] := MAX(DEST[63:0], SRC[63:0])
|
||
DEST[MAXVL-1:64] (Unmodified)
|
||
</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>VMAXSD __m128d _mm_max_round_sd( __m128d a, __m128d b, int);
|
||
</pre>
|
||
<pre>VMAXSD __m128d _mm_mask_max_round_sd(__m128d s, __mmask8 k, __m128d a, __m128d b, int);
|
||
</pre>
|
||
<pre>VMAXSD __m128d _mm_maskz_max_round_sd( __mmask8 k, __m128d a, __m128d b, int);
|
||
</pre>
|
||
<pre>MAXSD __m128d _mm_max_sd(__m128d a, __m128d b)
|
||
</pre>
|
||
<h2 class="exceptions" id="simd-floating-point-exceptions">SIMD Floating-Point Exceptions<a class="anchor" href="#simd-floating-point-exceptions">
|
||
¶
|
||
</a></h2>
|
||
<p>Invalid (Including QNaN Source Operand), Denormal.</p>
|
||
<h2 class="exceptions" id="other-exceptions">Other Exceptions<a class="anchor" href="#other-exceptions">
|
||
¶
|
||
</a></h2>
|
||
<p>Non-EVEX-encoded instruction, see <span class="not-imported">Table 2-20</span>, “Type 3 Class Exception Conditions.”</p>
|
||
<p>EVEX-encoded instruction, 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>
|