forked from NRZCode/ia32-64
114 lines
6.9 KiB
HTML
114 lines
6.9 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>RCPPS
|
||
— Compute Reciprocals of Packed Single Precision Floating-Point Values</title></head><body><header><nav><ul><li><a href='index.html'>Index</a></li><li>December 2023</li></ul></nav></header><h1>RCPPS
|
||
— Compute Reciprocals of Packed Single Precision Floating-Point Values</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>NP 0F 53 /r RCPPS xmm1, xmm2/m128</td>
|
||
<td>RM</td>
|
||
<td>V/V</td>
|
||
<td>SSE</td>
|
||
<td>Computes the approximate reciprocals of the packed single precision floating-point values in xmm2/m128 and stores the results in xmm1.</td></tr>
|
||
<tr>
|
||
<td>VEX.128.0F.WIG 53 /r VRCPPS xmm1, xmm2/m128</td>
|
||
<td>RM</td>
|
||
<td>V/V</td>
|
||
<td>AVX</td>
|
||
<td>Computes the approximate reciprocals of packed single precision values in xmm2/mem and stores the results in xmm1.</td></tr>
|
||
<tr>
|
||
<td>VEX.256.0F.WIG 53 /r VRCPPS ymm1, ymm2/m256</td>
|
||
<td>RM</td>
|
||
<td>V/V</td>
|
||
<td>AVX</td>
|
||
<td>Computes the approximate reciprocals of packed single precision values in ymm2/mem and stores the results in ymm1.</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>
|
||
<th>Operand 4</th></tr>
|
||
<tr>
|
||
<td>RM</td>
|
||
<td>ModRM:reg (w)</td>
|
||
<td>ModRM:r/m (r)</td>
|
||
<td>N/A</td>
|
||
<td>N/A</td></tr></table>
|
||
<h2 id="description">Description<a class="anchor" href="#description">
|
||
¶
|
||
</a></h2>
|
||
<p>Performs a SIMD computation of the approximate reciprocals of the four packed single precision floating-point values in the source operand (second operand) stores the packed single precision floating-point results in the destination operand. The source operand can be an XMM register or a 128-bit memory location. The destination operand is an XMM register. See <span class="not-imported">Figure 10-5</span> in the Intel<sup>®</sup> 64 and IA-32 Architectures Software Developer’s Manual, Volume 1, for an illustration of a SIMD single precision floating-point operation.</p>
|
||
<p>The relative error for this approximation is:</p>
|
||
<p>|Relative Error| ≤ 1.5 ∗ 2<sup>−12</sup></p>
|
||
<p>The RCPPS instruction is not affected by the rounding control bits in the MXCSR register. When a source value is a 0.0, an ∞ of the sign of the source value is returned. A denormal source value is treated as a 0.0 (of the same sign). Tiny results (see Section 4.9.1.5, “Numeric Underflow Exception (#U)” in Intel<sup>®</sup> 64 and IA-32 Architectures Software Developer’s Manual, Volume 1) are always flushed to 0.0, with the sign of the operand. (Input values greater than or equal to |1.11111111110100000000000B∗2<sup>125</sup>| are guaranteed to not produce tiny results; input values less than or equal to |1.00000000000110000000001B*2<sup>126</sup>| are guaranteed to produce tiny results, which are in turn flushed to 0.0; and input values in between this range may or may not produce tiny results, depending on the implementation.) When a source value is an SNaN or QNaN, the SNaN is converted to a QNaN or the source QNaN is returned.</p>
|
||
<p>In 64-bit mode, using a REX prefix in the form of REX.R permits this instruction to access additional registers (XMM8-XMM15).</p>
|
||
<p>128-bit Legacy SSE version: The second source can be an XMM register or an 128-bit memory location. The destination is not distinct from the first source XMM register and the upper bits (MAXVL-1:128) of the corresponding YMM register destination are unmodified.</p>
|
||
<p>VEX.128 encoded version: the first source operand is an XMM register or 128-bit memory location. The destination operand is an XMM register. The upper bits (MAXVL-1:128) of the corresponding YMM register destination are zeroed.</p>
|
||
<p>VEX.256 encoded version: The first source operand is a YMM register. The second source operand can be a YMM register or a 256-bit memory location. The destination operand is a YMM register.</p>
|
||
<p>Note: In VEX-encoded versions, VEX.vvvv is reserved and must be 1111b, otherwise instructions will #UD.</p>
|
||
<h2 id="operation">Operation<a class="anchor" href="#operation">
|
||
¶
|
||
</a></h2>
|
||
<h3 id="rcpps--128-bit-legacy-sse-version-">RCPPS (128-bit Legacy SSE Version)<a class="anchor" href="#rcpps--128-bit-legacy-sse-version-">
|
||
¶
|
||
</a></h3>
|
||
<pre>DEST[31:0] := APPROXIMATE(1/SRC[31:0])
|
||
DEST[63:32] := APPROXIMATE(1/SRC[63:32])
|
||
DEST[95:64] := APPROXIMATE(1/SRC[95:64])
|
||
DEST[127:96] := APPROXIMATE(1/SRC[127:96])
|
||
DEST[MAXVL-1:128] (Unmodified)
|
||
</pre>
|
||
<h3 id="vrcpps--vex-128-encoded-version-">VRCPPS (VEX.128 Encoded Version)<a class="anchor" href="#vrcpps--vex-128-encoded-version-">
|
||
¶
|
||
</a></h3>
|
||
<pre>DEST[31:0] := APPROXIMATE(1/SRC[31:0])
|
||
DEST[63:32] := APPROXIMATE(1/SRC[63:32])
|
||
DEST[95:64] := APPROXIMATE(1/SRC[95:64])
|
||
DEST[127:96] := APPROXIMATE(1/SRC[127:96])
|
||
DEST[MAXVL-1:128] := 0
|
||
</pre>
|
||
<h3 id="vrcpps--vex-256-encoded-version-">VRCPPS (VEX.256 Encoded Version)<a class="anchor" href="#vrcpps--vex-256-encoded-version-">
|
||
¶
|
||
</a></h3>
|
||
<pre>DEST[31:0] := APPROXIMATE(1/SRC[31:0])
|
||
DEST[63:32] := APPROXIMATE(1/SRC[63:32])
|
||
DEST[95:64] := APPROXIMATE(1/SRC[95:64])
|
||
DEST[127:96] := APPROXIMATE(1/SRC[127:96])
|
||
DEST[159:128] := APPROXIMATE(1/SRC[159:128])
|
||
DEST[191:160] := APPROXIMATE(1/SRC[191:160])
|
||
DEST[223:192] := APPROXIMATE(1/SRC[223:192])
|
||
DEST[255:224] := APPROXIMATE(1/SRC[255:224])
|
||
</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>RCCPS __m128 _mm_rcp_ps(__m128 a)
|
||
</pre>
|
||
<pre>RCPPS __m256 _mm256_rcp_ps (__m256 a);
|
||
</pre>
|
||
<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 2-21</span>, “Type 4 Class Exception Conditions,” additionally:</p>
|
||
<table>
|
||
<tr>
|
||
<td>#UD</td>
|
||
<td>If VEX.vvvv ≠ 1111B.</td></tr></table><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>
|