forked from NRZCode/ia32-64
216 lines
10 KiB
HTML
216 lines
10 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>VFCMADDCPH/VFMADDCPH
|
||
— Complex Multiply and Accumulate FP16 Values</title></head><body><header><nav><ul><li><a href='index.html'>Index</a></li><li>December 2023</li></ul></nav></header><h1>VFCMADDCPH/VFMADDCPH
|
||
— Complex Multiply and Accumulate 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.F2.MAP6.W0 56 /r VFCMADDCPH xmm1{k1}{z}, xmm2, xmm3/m128/m32bcst</td>
|
||
<td>A</td>
|
||
<td>V/V</td>
|
||
<td>AVX512-FP16 AVX512VL</td>
|
||
<td>Complex multiply a pair of FP16 values from xmm2 and complex conjugate of xmm3/m128/m32bcst, add to xmm1 and store the result in xmm1 subject to writemask k1.</td></tr>
|
||
<tr>
|
||
<td>EVEX.256.F2.MAP6.W0 56 /r VFCMADDCPH ymm1{k1}{z}, ymm2, ymm3/m256/m32bcst</td>
|
||
<td>A</td>
|
||
<td>V/V</td>
|
||
<td>AVX512-FP16 AVX512VL</td>
|
||
<td>Complex multiply a pair of FP16 values from ymm2 and complex conjugate of ymm3/m256/m32bcst, add to ymm1 and store the result in ymm1 subject to writemask k1.</td></tr>
|
||
<tr>
|
||
<td>EVEX.512.F2.MAP6.W0 56 /r VFCMADDCPH zmm1{k1}{z}, zmm2, zmm3/m512/m32bcst {er}</td>
|
||
<td>A</td>
|
||
<td>V/V</td>
|
||
<td>AVX512-FP16</td>
|
||
<td>Complex multiply a pair of FP16 values from zmm2 and complex conjugate of zmm3/m512/m32bcst, add to zmm1 and store the result in zmm1 subject to writemask k1.</td></tr>
|
||
<tr>
|
||
<td>EVEX.128.F3.MAP6.W0 56 /r VFMADDCPH xmm1{k1}{z}, xmm2, xmm3/m128/m32bcst</td>
|
||
<td>A</td>
|
||
<td>V/V</td>
|
||
<td>AVX512-FP16 AVX512VL</td>
|
||
<td>Complex multiply a pair of FP16 values from xmm2 and xmm3/m128/m32bcst, add to xmm1 and store the result in xmm1 subject to writemask k1.</td></tr>
|
||
<tr>
|
||
<td>EVEX.256.F3.MAP6.W0 56 /r VFMADDCPH ymm1{k1}{z}, ymm2, ymm3/m256/m32bcst</td>
|
||
<td>A</td>
|
||
<td>V/V</td>
|
||
<td>AVX512-FP16 AVX512VL</td>
|
||
<td>Complex multiply a pair of FP16 values from ymm2 and ymm3/m256/m32bcst, add to ymm1 and store the result in ymm1 subject to writemask k1.</td></tr>
|
||
<tr>
|
||
<td>EVEX.512.F3.MAP6.W0 56 /r VFMADDCPH zmm1{k1}{z}, zmm2, zmm3/m512/m32bcst {er}</td>
|
||
<td>A</td>
|
||
<td>V/V</td>
|
||
<td>AVX512-FP16</td>
|
||
<td>Complex multiply a pair of FP16 values from zmm2 and zmm3/m512/m32bcst, add to zmm1 and store the result in zmm1 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 (r, 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 performs a complex multiply and accumulate operation. There are normal and complex conjugate forms of the operation.</p>
|
||
<p>The broadcasting and masking for this operation is done on 32-bit quantities representing a pair of FP16 values.</p>
|
||
<p>Rounding is performed at every FMA (fused multiply and add) boundary. Execution occurs as if all MXCSR exceptions are masked. MXCSR status bits are updated to reflect exceptional conditions.</p>
|
||
<h3 id="operation">Operation<a class="anchor" href="#operation">
|
||
¶
|
||
</a></h3>
|
||
<h4 id="vfcmaddcph-dest-k1---src1--src2--avx512-">VFCMADDCPH dest{k1}, src1, src2 (AVX512)<a class="anchor" href="#vfcmaddcph-dest-k1---src1--src2--avx512-">
|
||
¶
|
||
</a></h4>
|
||
<pre>VL = 128, 256, 512
|
||
KL := VL / 32
|
||
FOR i := 0 to KL-1:
|
||
IF k1[i] or *no writemask*:
|
||
IF broadcasting and src2 is memory:
|
||
tsrc2.fp16[2*i+0] := src2.fp16[0]
|
||
tsrc2.fp16[2*i+1] := src2.fp16[1]
|
||
ELSE:
|
||
tsrc2.fp16[2*i+0] := src2.fp16[2*i+0]
|
||
tsrc2.fp16[2*i+1] := src2.fp16[2*i+1]
|
||
FOR i := 0 to KL-1:
|
||
IF k1[i] or *no writemask*:
|
||
tmp[2*i+0] := dest.fp16[2*i+0] + src1.fp16[2*i+0] * tsrc2.fp16[2*i+0]
|
||
tmp[2*i+1] := dest.fp16[2*i+1] + src1.fp16[2*i+1] * tsrc2.fp16[2*i+0]
|
||
FOR i := 0 to KL-1:
|
||
IF k1[i] or *no writemask*:
|
||
// conjugate version subtracts odd final term
|
||
dest.fp16[2*i+0] := tmp[2*i+0] + src1.fp16[2*i+1] * tsrc2.fp16[2*i+1]
|
||
dest.fp16[2*i+1] := tmp[2*i+1] - src1.fp16[2*i+0] * tsrc2.fp16[2*i+1]
|
||
ELSE IF *zeroing*:
|
||
dest.fp16[2*i+0] := 0
|
||
dest.fp16[2*i+1] := 0
|
||
DEST[MAXVL-1:VL] := 0
|
||
</pre>
|
||
<h4 id="vfmaddcph-dest-k1---src1--src2--avx512-">VFMADDCPH dest{k1}, src1, src2 (AVX512)<a class="anchor" href="#vfmaddcph-dest-k1---src1--src2--avx512-">
|
||
¶
|
||
</a></h4>
|
||
<pre>VL = 128, 256, 512
|
||
KL := VL / 32
|
||
FOR i := 0 to KL-1:
|
||
IF k1[i] or *no writemask*:
|
||
IF broadcasting and src2 is memory:
|
||
tsrc2.fp16[2*i+0] := src2.fp16[0]
|
||
tsrc2.fp16[2*i+1] := src2.fp16[1]
|
||
ELSE:
|
||
tsrc2.fp16[2*i+0] := src2.fp16[2*i+0]
|
||
tsrc2.fp16[2*i+1] := src2.fp16[2*i+1]
|
||
FOR i := 0 to KL-1:
|
||
IF k1[i] or *no writemask*:
|
||
tmp[2*i+0] := dest.fp16[2*i+0] + src1.fp16[2*i+0] * tsrc2.fp16[2*i+0]
|
||
tmp[2*i+1] := dest.fp16[2*i+1] + src1.fp16[2*i+1] * tsrc2.fp16[2*i+0]
|
||
FOR i := 0 to KL-1:
|
||
IF k1[i] or *no writemask*:
|
||
// non-conjugate version subtracts even term
|
||
dest.fp16[2*i+0] := tmp[2*i+0] - src1.fp16[2*i+1] * tsrc2.fp16[2*i+1]
|
||
dest.fp16[2*i+1] := tmp[2*i+1] + src1.fp16[2*i+0] * tsrc2.fp16[2*i+1]
|
||
ELSE IF *zeroing*:
|
||
dest.fp16[2*i+0] := 0
|
||
dest.fp16[2*i+1] := 0
|
||
DEST[MAXVL-1:VL] := 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>VFCMADDCPH __m128h _mm_fcmadd_pch (__m128h a, __m128h b, __m128h c);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m128h _mm_mask_fcmadd_pch (__m128h a, __mmask8 k, __m128h b, __m128h c);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m128h _mm_mask3_fcmadd_pch (__m128h a, __m128h b, __m128h c, __mmask8 k);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m128h _mm_maskz_fcmadd_pch (__mmask8 k, __m128h a, __m128h b, __m128h c);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m256h _mm256_fcmadd_pch (__m256h a, __m256h b, __m256h c);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m256h _mm256_mask_fcmadd_pch (__m256h a, __mmask8 k, __m256h b, __m256h c);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m256h _mm256_mask3_fcmadd_pch (__m256h a, __m256h b, __m256h c, __mmask8 k);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m256h _mm256_maskz_fcmadd_pch (__mmask8 k, __m256h a, __m256h b, __m256h c);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m512h _mm512_fcmadd_pch (__m512h a, __m512h b, __m512h c);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m512h _mm512_mask_fcmadd_pch (__m512h a, __mmask16 k, __m512h b, __m512h c);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m512h _mm512_mask3_fcmadd_pch (__m512h a, __m512h b, __m512h c, __mmask16 k);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m512h _mm512_maskz_fcmadd_pch (__mmask16 k, __m512h a, __m512h b, __m512h c);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m512h _mm512_fcmadd_round_pch (__m512h a, __m512h b, __m512h c, const int rounding);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m512h _mm512_mask_fcmadd_round_pch (__m512h a, __mmask16 k, __m512h b, __m512h c, const int rounding);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m512h _mm512_mask3_fcmadd_round_pch (__m512h a, __m512h b, __m512h c, __mmask16 k, const int rounding);
|
||
</pre>
|
||
<pre>VFCMADDCPH __m512h _mm512_maskz_fcmadd_round_pch (__mmask16 k, __m512h a, __m512h b, __m512h c, const int rounding);
|
||
</pre>
|
||
<pre>VFMADDCPH __m128h _mm_fmadd_pch (__m128h a, __m128h b, __m128h c);
|
||
</pre>
|
||
<pre>VFMADDCPH __m128h _mm_mask_fmadd_pch (__m128h a, __mmask8 k, __m128h b, __m128h c);
|
||
</pre>
|
||
<pre>VFMADDCPH __m128h _mm_mask3_fmadd_pch (__m128h a, __m128h b, __m128h c, __mmask8 k);
|
||
</pre>
|
||
<pre>VFMADDCPH __m128h _mm_maskz_fmadd_pch (__mmask8 k, __m128h a, __m128h b, __m128h c);
|
||
</pre>
|
||
<pre>VFMADDCPH __m256h _mm256_fmadd_pch (__m256h a, __m256h b, __m256h c);
|
||
</pre>
|
||
<pre>VFMADDCPH __m256h _mm256_mask_fmadd_pch (__m256h a, __mmask8 k, __m256h b, __m256h c);
|
||
</pre>
|
||
<pre>VFMADDCPH __m256h _mm256_mask3_fmadd_pch (__m256h a, __m256h b, __m256h c, __mmask8 k);
|
||
</pre>
|
||
<pre>VFMADDCPH __m256h _mm256_maskz_fmadd_pch (__mmask8 k, __m256h a, __m256h b, __m256h c);
|
||
</pre>
|
||
<pre>VFMADDCPH __m512h _mm512_fmadd_pch (__m512h a, __m512h b, __m512h c);
|
||
</pre>
|
||
<pre>VFMADDCPH __m512h _mm512_mask_fmadd_pch (__m512h a, __mmask16 k, __m512h b, __m512h c);
|
||
</pre>
|
||
<pre>VFMADDCPH __m512h _mm512_mask3_fmadd_pch (__m512h a, __m512h b, __m512h c, __mmask16 k);
|
||
</pre>
|
||
<pre>VFMADDCPH __m512h _mm512_maskz_fmadd_pch (__mmask16 k, __m512h a, __m512h b, __m512h c);
|
||
</pre>
|
||
<pre>VFMADDCPH __m512h _mm512_fmadd_round_pch (__m512h a, __m512h b, __m512h c, const int rounding);
|
||
</pre>
|
||
<pre>VFMADDCPH __m512h _mm512_mask_fmadd_round_pch (__m512h a, __mmask16 k, __m512h b, __m512h c, const int rounding);
|
||
</pre>
|
||
<pre>VFMADDCPH __m512h _mm512_mask3_fmadd_round_pch (__m512h a, __m512h b, __m512h c, __mmask16 k, const int rounding);
|
||
</pre>
|
||
<pre>VFMADDCPH __m512h _mm512_maskz_fmadd_round_pch (__mmask16 k, __m512h a, __m512h b, __m512h c, const int rounding);
|
||
</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 2-49</span>, “Type E4 Class Exception Conditions.”</p>
|
||
<p>Additionally:</p>
|
||
<table>
|
||
<tr>
|
||
<td>#UD</td>
|
||
<td>If (dest_reg == src1_reg) or (dest_reg == src2_reg).</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>
|