drginfo/ia32-64
1
0
Fork 0
forked from NRZCode/ia32-64
Code Pull requests Activity
ia32-64/x86/vfnmadd132ps.vfnmadd213ps.vfnmadd231ps.html

400 lines
19 KiB
HTML
Raw Normal View History

Initial commit
2025-07-08 02:23:29 -03:00
<!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>VFNMADD132PS/VFNMADD213PS/VFNMADD231PS
— Fused Negative Multiply-Add of PackedSingle 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>VFNMADD132PS/VFNMADD213PS/VFNMADD231PS
— Fused Negative Multiply-Add of PackedSingle 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>VEX.128.66.0F38.W0 9C /r VFNMADD132PS xmm1, xmm2, xmm3/m128</td>
<td>A</td>
<td>V/V</td>
<td>FMA</td>
<td>Multiply packed single precision floating-point values from xmm1 and xmm3/mem, negate the multiplication result and add to xmm2 and put result in xmm1.</td></tr>
<tr>
<td>VEX.128.66.0F38.W0 AC /r VFNMADD213PS xmm1, xmm2, xmm3/m128</td>
<td>A</td>
<td>V/V</td>
<td>FMA</td>
<td>Multiply packed single precision floating-point values from xmm1 and xmm2, negate the multiplication result and add to xmm3/mem and put result in xmm1.</td></tr>
<tr>
<td>VEX.128.66.0F38.W0 BC /r VFNMADD231PS xmm1, xmm2, xmm3/m128</td>
<td>A</td>
<td>V/V</td>
<td>FMA</td>
<td>Multiply packed single precision floating-point values from xmm2 and xmm3/mem, negate the multiplication result and add to xmm1 and put result in xmm1.</td></tr>
<tr>
<td>VEX.256.66.0F38.W0 9C /r VFNMADD132PS ymm1, ymm2, ymm3/m256</td>
<td>A</td>
<td>V/V</td>
<td>FMA</td>
<td>Multiply packed single precision floating-point values from ymm1 and ymm3/mem, negate the multiplication result and add to ymm2 and put result in ymm1.</td></tr>
<tr>
<td>VEX.256.66.0F38.W0 AC /r VFNMADD213PS ymm1, ymm2, ymm3/m256</td>
<td>A</td>
<td>V/V</td>
<td>FMA</td>
<td>Multiply packed single precision floating-point values from ymm1 and ymm2, negate the multiplication result and add to ymm3/mem and put result in ymm1.</td></tr>
<tr>
<td>VEX.256.66.0F38.0 BC /r VFNMADD231PS ymm1, ymm2, ymm3/m256</td>
<td>A</td>
<td>V/V</td>
<td>FMA</td>
<td>Multiply packed single precision floating-point values from ymm2 and ymm3/mem, negate the multiplication result and add to ymm1 and put result in ymm1.</td></tr>
<tr>
<td>EVEX.128.66.0F38.W0 9C /r VFNMADD132PS xmm1 {k1}{z}, xmm2, xmm3/m128/m32bcst</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed single precision floating-point values from xmm1 and xmm3/m128/m32bcst, negate the multiplication result and add to xmm2 and put result in xmm1.</td></tr>
<tr>
<td>EVEX.128.66.0F38.W0 AC /r VFNMADD213PS xmm1 {k1}{z}, xmm2, xmm3/m128/m32bcst</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed single precision floating-point values from xmm1 and xmm2, negate the multiplication result and add to xmm3/m128/m32bcst and put result in xmm1.</td></tr>
<tr>
<td>EVEX.128.66.0F38.W0 BC /r VFNMADD231PS xmm1 {k1}{z}, xmm2, xmm3/m128/m32bcst</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed single precision floating-point values from xmm2 and xmm3/m128/m32bcst, negate the multiplication result and add to xmm1 and put result in xmm1.</td></tr>
<tr>
<td>EVEX.256.66.0F38.W0 9C /r VFNMADD132PS ymm1 {k1}{z}, ymm2, ymm3/m256/m32bcst</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed single precision floating-point values from ymm1 and ymm3/m256/m32bcst, negate the multiplication result and add to ymm2 and put result in ymm1.</td></tr>
<tr>
<td>EVEX.256.66.0F38.W0 AC /r VFNMADD213PS ymm1 {k1}{z}, ymm2, ymm3/m256/m32bcst</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed single precision floating-point values from ymm1 and ymm2, negate the multiplication result and add to ymm3/m256/m32bcst and put result in ymm1.</td></tr>
<tr>
<td>EVEX.256.66.0F38.W0 BC /r VFNMADD231PS ymm1 {k1}{z}, ymm2, ymm3/m256/m32bcst</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed single precision floating-point values from ymm2 and ymm3/m256/m32bcst, negate the multiplication result and add to ymm1 and put result in ymm1.</td></tr>
<tr>
<td>EVEX.512.66.0F38.W0 9C /r VFNMADD132PS zmm1 {k1}{z}, zmm2, zmm3/m512/m32bcst{er}</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed single precision floating-point values from zmm1 and zmm3/m512/m32bcst, negate the multiplication result and add to zmm2 and put result in zmm1.</td></tr>
<tr>
<td>EVEX.512.66.0F38.W0 AC /r VFNMADD213PS zmm1 {k1}{z}, zmm2, zmm3/m512/m32bcst{er}</td>
<td>B</td>
<td>V/V</td>
<td>AVX512F</td>
<td>Multiply packed single precision floating-point values from zmm1 and zmm2, negate the multiplication result and add to zmm3/m512/m32bcst and put result in zmm1.</td></tr>
<tr>
<td>EVEX.512.66.0F38.W0 BC /r VFNMADD231PS zmm1 {k1}{z}, zmm2, zmm3/m512/m32bcst{er}</td>
<td>B</td>
<td>V/V</td>
<td>AVX512F</td>
<td>Multiply packed single precision floating-point values from zmm2 and zmm3/m512/m32bcst, negate the multiplication result and add to zmm1 and put result in zmm1.</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>VEX.vvvv (r)</td>
<td>ModRM:r/m (r)</td>
<td>N/A</td></tr>
<tr>
<td>B</td>
<td>Full</td>
<td>ModRM:reg (r, w)</td>
<td>EVEX.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>VFNMADD132PS: Multiplies the four, eight or sixteen packed single precision floating-point values from the first source operand to the four, eight or sixteen packed single precision floating-point values in the third source operand, adds the negated infinite precision intermediate result to the four, eight or sixteen packed single precision floating-point values in the second source operand, performs rounding and stores the resulting four, eight or sixteen packed single precision floating-point values to the destination operand (first source operand).</p>
<p>VFNMADD213PS: Multiplies the four, eight or sixteen packed single precision floating-point values from the second source operand to the four, eight or sixteen packed single precision floating-point values in the first source operand, adds the negated infinite precision intermediate result to the four, eight or sixteen packed single precision floating-point values in the third source operand, performs rounding and stores the resulting the four, eight or sixteen packed single precision floating-point values to the destination operand (first source operand).</p>
<p>VFNMADD231PS: Multiplies the four, eight or sixteen packed single precision floating-point values from the second source operand to the four, eight or sixteen packed single precision floating-point values in the third source operand, adds the negated infinite precision intermediate result to the four, eight or sixteen packed single precision floating-point values in the first source operand, performs rounding and stores the resulting four, eight or sixteen packed single precision floating-point values to the destination operand (first source operand).</p>
<p>EVEX encoded versions: The destination operand (also first source operand) and the second source operand are ZMM/YMM/XMM register. The third source operand is a ZMM/YMM/XMM register, a 512/256/128-bit memory location or a 512/256/128-bit vector broadcasted from a 32-bit memory location. The destination operand is conditionally updated with write mask k1.</p>
<p>VEX.256 encoded version: The destination operand (also first source operand) is a YMM register and encoded in reg_field. The second source operand is a YMM register and encoded in VEX.vvvv. The third source operand is a YMM register or a 256-bit memory location and encoded in rm_field.</p>
<p>VEX.128 encoded version: The destination operand (also first source operand) is a XMM register and encoded in reg_field. The second source operand is a XMM register and encoded in VEX.vvvv. The third source operand is a XMM register or a 128-bit memory location and encoded in rm_field. The upper 128 bits of the YMM destination register are zeroed.</p>
<h3 id="operation">Operation<a class="anchor" href="#operation">
</a></h3>
<pre>In the operations below, “*” and “+” symbols represent multiplication and addition with infinite precision inputs and outputs (no
rounding).
</pre>
<h4 id="vfnmadd132ps-dest--src2--src3--vex-encoded-version-">VFNMADD132PS DEST, SRC2, SRC3 (VEX encoded version)<a class="anchor" href="#vfnmadd132ps-dest--src2--src3--vex-encoded-version-">
</a></h4>
<pre>IF (VEX.128) THEN
MAXNUM := 2
ELSEIF (VEX.256)
MAXNUM := 4
FI
For i = 0 to MAXNUM-1 {
n := 32*i;
DEST[n+31:n] := RoundFPControl_MXCSR(- (DEST[n+31:n]*SRC3[n+31:n]) + SRC2[n+31:n])
}
IF (VEX.128) THEN
DEST[MAXVL-1:128] := 0
ELSEIF (VEX.256)
DEST[MAXVL-1:256] := 0
FI
</pre>
<h4 id="vfnmadd213ps-dest--src2--src3--vex-encoded-version-">VFNMADD213PS DEST, SRC2, SRC3 (VEX encoded version)<a class="anchor" href="#vfnmadd213ps-dest--src2--src3--vex-encoded-version-">
</a></h4>
<pre>IF (VEX.128) THEN
MAXNUM := 2
ELSEIF (VEX.256)
MAXNUM := 4
FI
For i = 0 to MAXNUM-1 {
n := 32*i;
DEST[n+31:n] := RoundFPControl_MXCSR(- (SRC2[n+31:n]*DEST[n+31:n]) + SRC3[n+31:n])
}
IF (VEX.128) THEN
DEST[MAXVL-1:128] := 0
ELSEIF (VEX.256)
DEST[MAXVL-1:256] := 0
FI
</pre>
<h4 id="vfnmadd231ps-dest--src2--src3--vex-encoded-version-">VFNMADD231PS DEST, SRC2, SRC3 (VEX encoded version)<a class="anchor" href="#vfnmadd231ps-dest--src2--src3--vex-encoded-version-">
</a></h4>
<pre>IF (VEX.128) THEN
MAXNUM := 2
ELSEIF (VEX.256)
MAXNUM := 4
FI
For i = 0 to MAXNUM-1 {
n := 32*i;
DEST[n+31:n] := RoundFPControl_MXCSR(- (SRC2[n+31:n]*SRC3[n+31:n]) + DEST[n+31:n])
}
IF (VEX.128) THEN
DEST[MAXVL-1:128] := 0
ELSEIF (VEX.256)
DEST[MAXVL-1:256] := 0
FI
</pre>
<h4 id="vfnmadd132ps-dest--src2--src3--evex-encoded-version--when-src3-operand-is-a-register-">VFNMADD132PS DEST, SRC2, SRC3 (EVEX encoded version, when src3 operand is a register)<a class="anchor" href="#vfnmadd132ps-dest--src2--src3--evex-encoded-version--when-src3-operand-is-a-register-">
</a></h4>
<pre>(KL, VL) = (4, 128), (8, 256), (16, 512)
IF (VL = 512) AND (EVEX.b = 1)
THEN
SET_ROUNDING_MODE_FOR_THIS_INSTRUCTION(EVEX.RC);
ELSE
SET_ROUNDING_MODE_FOR_THIS_INSTRUCTION(MXCSR.RC);
FI;
FOR j := 0 TO KL-1
i := j * 32
IF k1[j] OR *no writemask*
THEN DEST[i+31:i] :=
RoundFPControl(-(DEST[i+31:i]*SRC3[i+31:i]) + SRC2[i+31:i])
ELSE
IF *merging-masking* ; merging-masking
THEN *DEST[i+31:i] remains unchanged*
ELSE ; zeroing-masking
DEST[i+31:i] := 0
FI
FI;
ENDFOR
DEST[MAXVL-1:VL] := 0
</pre>
<h4 id="vfnmadd132ps-dest--src2--src3--evex-encoded-version--when-src3-operand-is-a-memory-source-">VFNMADD132PS DEST, SRC2, SRC3 (EVEX encoded version, when src3 operand is a memory source)<a class="anchor" href="#vfnmadd132ps-dest--src2--src3--evex-encoded-version--when-src3-operand-is-a-memory-source-">
</a></h4>
<pre>(KL, VL) = (4, 128), (8, 256), (16, 512)
FOR j := 0 TO KL-1
i := j * 32
IF k1[j] OR *no writemask*
THEN
IF (EVEX.b = 1)
THEN
DEST[i+31:i] :=
RoundFPControl_MXCSR(-(DEST[i+31:i]*SRC3[31:0]) + SRC2[i+31:i])
ELSE
DEST[i+31:i] :=
RoundFPControl_MXCSR(-(DEST[i+31:i]*SRC3[i+31:i]) + SRC2[i+31:i])
FI;
ELSE
IF *merging-masking* ; merging-masking
THEN *DEST[i+31:i] remains unchanged*
ELSE ; zeroing-masking
DEST[i+31:i] := 0
FI
FI;
ENDFOR
DEST[MAXVL-1:VL] := 0
</pre>
<h4 id="vfnmadd213ps-dest--src2--src3--evex-encoded-version--when-src3-operand-is-a-register-">VFNMADD213PS DEST, SRC2, SRC3 (EVEX encoded version, when src3 operand is a register)<a class="anchor" href="#vfnmadd213ps-dest--src2--src3--evex-encoded-version--when-src3-operand-is-a-register-">
</a></h4>
<pre>(KL, VL) = (4, 128), (8, 256), (16, 512)
IF (VL = 512) AND (EVEX.b = 1)
THEN
SET_ROUNDING_MODE_FOR_THIS_INSTRUCTION(EVEX.RC);
ELSE
SET_ROUNDING_MODE_FOR_THIS_INSTRUCTION(MXCSR.RC);
FI;
FOR j := 0 TO KL-1
i := j * 32
IF k1[j] OR *no writemask*
THEN DEST[i+31:i] :=
RoundFPControl(-(SRC2[i+31:i]*DEST[i+31:i]) + SRC3[i+31:i])
ELSE
IF *merging-masking* ; merging-masking
THEN *DEST[i+31:i] remains unchanged*
ELSE ; zeroing-masking
DEST[i+31:i] := 0
FI
FI;
ENDFOR
DEST[MAXVL-1:VL] := 0
</pre>
<h4 id="vfnmadd213ps-dest--src2--src3--evex-encoded-version--when-src3-operand-is-a-memory-source-">VFNMADD213PS DEST, SRC2, SRC3 (EVEX encoded version, when src3 operand is a memory source)<a class="anchor" href="#vfnmadd213ps-dest--src2--src3--evex-encoded-version--when-src3-operand-is-a-memory-source-">
</a></h4>
<pre>(KL, VL) = (4, 128), (8, 256), (16, 512)
FOR j := 0 TO KL-1
i := j * 32
IF k1[j] OR *no writemask*
THEN
IF (EVEX.b = 1)
THEN
DEST[i+31:i] :=
RoundFPControl_MXCSR(-(SRC2[i+31:i]*DEST[i+31:i]) + SRC3[31:0])
ELSE
DEST[i+31:i] :=
RoundFPControl_MXCSR(-(SRC2[i+31:i]*DEST[i+31:i]) + SRC3[i+31:i])
FI;
ELSE
IF *merging-masking* ; merging-masking
THEN *DEST[i+31:i] remains unchanged*
ELSE ; zeroing-masking
DEST[i+31:i] := 0
FI
FI;
ENDFOR
DEST[MAXVL-1:VL] := 0
</pre>
<h4 id="vfnmadd231ps-dest--src2--src3--evex-encoded-version--when-src3-operand-is-a-register-">VFNMADD231PS DEST, SRC2, SRC3 (EVEX encoded version, when src3 operand is a register)<a class="anchor" href="#vfnmadd231ps-dest--src2--src3--evex-encoded-version--when-src3-operand-is-a-register-">
</a></h4>
<pre>(KL, VL) = (4, 128), (8, 256), (16, 512)
IF (VL = 512) AND (EVEX.b = 1)
THEN
SET_ROUNDING_MODE_FOR_THIS_INSTRUCTION(EVEX.RC);
ELSE
SET_ROUNDING_MODE_FOR_THIS_INSTRUCTION(MXCSR.RC);
FI;
FOR j := 0 TO KL-1
i := j * 32
IF k1[j] OR *no writemask*
THEN DEST[i+31:i] :=
RoundFPControl(-(SRC2[i+31:i]*SRC3[i+31:i]) + DEST[i+31:i])
ELSE
IF *merging-masking* ; merging-masking
THEN *DEST[i+31:i] remains unchanged*
ELSE ; zeroing-masking
DEST[i+31:i] := 0
FI
FI;
ENDFOR
DEST[MAXVL-1:VL] := 0
</pre>
<h4 id="vfnmadd231ps-dest--src2--src3--evex-encoded-version--when-src3-operand-is-a-memory-source-">VFNMADD231PS DEST, SRC2, SRC3 (EVEX encoded version, when src3 operand is a memory source)<a class="anchor" href="#vfnmadd231ps-dest--src2--src3--evex-encoded-version--when-src3-operand-is-a-memory-source-">
</a></h4>
<pre>(KL, VL) = (4, 128), (8, 256), (16, 512)
FOR j := 0 TO KL-1
i := j * 32
IF k1[j] OR *no writemask*
THEN
IF (EVEX.b = 1)
THEN
DEST[i+31:i] :=
RoundFPControl_MXCSR(-(SRC2[i+31:i]*SRC3[31:0]) + DEST[i+31:i])
ELSE
DEST[i+31:i] :=
RoundFPControl_MXCSR(-(SRC2[i+31:i]*SRC3[i+31:i]) + DEST[i+31:i])
FI;
ELSE
IF *merging-masking* ; merging-masking
THEN *DEST[i+31:i] remains unchanged*
ELSE ; zeroing-masking
DEST[i+31:i] := 0
FI
FI;
ENDFOR
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>VFNMADDxxxPS __m512 _mm512_fnmadd_ps(__m512 a, __m512 b, __m512 c);
</pre>
<pre>VFNMADDxxxPS __m512 _mm512_fnmadd_round_ps(__m512 a, __m512 b, __m512 c, int r);
</pre>
<pre>VFNMADDxxxPS __m512 _mm512_mask_fnmadd_ps(__m512 a, __mmask16 k, __m512 b, __m512 c);
</pre>
<pre>VFNMADDxxxPS __m512 _mm512_maskz_fnmadd_ps(__mmask16 k, __m512 a, __m512 b, __m512 c);
</pre>
<pre>VFNMADDxxxPS __m512 _mm512_mask3_fnmadd_ps(__m512 a, __m512 b, __m512 c, __mmask16 k);
</pre>
<pre>VFNMADDxxxPS __m512 _mm512_mask_fnmadd_round_ps(__m512 a, __mmask16 k, __m512 b, __m512 c, int r);
</pre>
<pre>VFNMADDxxxPS __m512 _mm512_maskz_fnmadd_round_ps(__mmask16 k, __m512 a, __m512 b, __m512 c, int r);
</pre>
<pre>VFNMADDxxxPS __m512 _mm512_mask3_fnmadd_round_ps(__m512 a, __m512 b, __m512 c, __mmask16 k, int r);
</pre>
<pre>VFNMADDxxxPS __m256 _mm256_mask_fnmadd_ps(__m256 a, __mmask8 k, __m256 b, __m256 c);
</pre>
<pre>VFNMADDxxxPS __m256 _mm256_maskz_fnmadd_ps(__mmask8 k, __m256 a, __m256 b, __m256 c);
</pre>
<pre>VFNMADDxxxPS __m256 _mm256_mask3_fnmadd_ps(__m256 a, __m256 b, __m256 c, __mmask8 k);
</pre>
<pre>VFNMADDxxxPS __m128 _mm_mask_fnmadd_ps(__m128 a, __mmask8 k, __m128 b, __m128 c);
</pre>
<pre>VFNMADDxxxPS __m128 _mm_maskz_fnmadd_ps(__mmask8 k, __m128 a, __m128 b, __m128 c);
</pre>
<pre>VFNMADDxxxPS __m128 _mm_mask3_fnmadd_ps(__m128 a, __m128 b, __m128 c, __mmask8 k);
</pre>
<pre>VFNMADDxxxPS __m128 _mm_fnmadd_ps (__m128 a, __m128 b, __m128 c);
</pre>
<pre>VFNMADDxxxPS __m256 _mm256_fnmadd_ps (__m256 a, __m256 b, __m256 c);
</pre>
<h3 class="exceptions" id="simd-floating-point-exceptions">SIMD Floating-Point Exceptions<a class="anchor" href="#simd-floating-point-exceptions">
</a></h3>
<p>Overflow, Underflow, Invalid, Precision, Denormal.</p>
<h3 class="exceptions" id="other-exceptions">Other Exceptions<a class="anchor" href="#other-exceptions">
</a></h3>
<p>VEX-encoded instructions, see <span class="not-imported">Table 2-19</span>, “Type 2 Class Exception Conditions.”</p>
<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 Developer’s Manual</a> for anything serious.
</p></footer></body></html>
Copy permalink