drginfo/ia32-64
1
0
Fork 0
forked from NRZCode/ia32-64
Code Pull requests Activity
ia32-64/x86/vfmsub132pd.vfmsub213pd.vfmsub231pd.html
NRZ Code 92622c37d6 Initial commit
2025-07-08 02:23:29 -03:00

401 lines
19 KiB
HTML
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

<!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>VFMSUB132PD/VFMSUB213PD/VFMSUB231PD
— Fused Multiply-Subtract of Packed DoublePrecision 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>VFMSUB132PD/VFMSUB213PD/VFMSUB231PD
— Fused Multiply-Subtract of Packed DoublePrecision 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.W1 9A /r VFMSUB132PD xmm1, xmm2, xmm3/m128</td>
<td>A</td>
<td>V/V</td>
<td>FMA</td>
<td>Multiply packed double precision floating-point values from xmm1 and xmm3/mem, subtract xmm2 and put result in xmm1.</td></tr>
<tr>
<td>VEX.128.66.0F38.W1 AA /r VFMSUB213PD xmm1, xmm2, xmm3/m128</td>
<td>A</td>
<td>V/V</td>
<td>FMA</td>
<td>Multiply packed double precision floating-point values from xmm1 and xmm2, subtract xmm3/mem and put result in xmm1.</td></tr>
<tr>
<td>VEX.128.66.0F38.W1 BA /r VFMSUB231PD xmm1, xmm2, xmm3/m128</td>
<td>A</td>
<td>V/V</td>
<td>FMA</td>
<td>Multiply packed double precision floating-point values from xmm2 and xmm3/mem, subtract xmm1 and put result in xmm1.</td></tr>
<tr>
<td>VEX.256.66.0F38.W1 9A /r VFMSUB132PD ymm1, ymm2, ymm3/m256</td>
<td>A</td>
<td>V/V</td>
<td>FMA</td>
<td>Multiply packed double precision floating-point values from ymm1 and ymm3/mem, subtract ymm2 and put result in ymm1.</td></tr>
<tr>
<td>VEX.256.66.0F38.W1 AA /r VFMSUB213PD ymm1, ymm2, ymm3/m256</td>
<td>A</td>
<td>V/V</td>
<td>FMA</td>
<td>Multiply packed double precision floating-point values from ymm1 and ymm2, subtract ymm3/mem and put result in ymm1.</td></tr>
<tr>
<td>VEX.256.66.0F38.W1 BA /r VFMSUB231PD ymm1, ymm2, ymm3/m256</td>
<td>A</td>
<td>V/V</td>
<td>FMA</td>
<td>Multiply packed double precision floating-point values from ymm2 and ymm3/mem, subtract ymm1 and put result in ymm1.S</td></tr>
<tr>
<td>EVEX.128.66.0F38.W1 9A /r VFMSUB132PD xmm1 {k1}{z}, xmm2, xmm3/m128/m64bcst</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed double precision floating-point values from xmm1 and xmm3/m128/m64bcst, subtract xmm2 and put result in xmm1 subject to writemask k1.</td></tr>
<tr>
<td>EVEX.128.66.0F38.W1 AA /r VFMSUB213PD xmm1 {k1}{z}, xmm2, xmm3/m128/m64bcst</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed double precision floating-point values from xmm1 and xmm2, subtract xmm3/m128/m64bcst and put result in xmm1 subject to writemask k1.</td></tr>
<tr>
<td>EVEX.128.66.0F38.W1 BA /r VFMSUB231PD xmm1 {k1}{z}, xmm2, xmm3/m128/m64bcst</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed double precision floating-point values from xmm2 and xmm3/m128/m64bcst, subtract xmm1 and put result in xmm1 subject to writemask k1.</td></tr>
<tr>
<td>EVEX.256.66.0F38.W1 9A /r VFMSUB132PD ymm1 {k1}{z}, ymm2, ymm3/m256/m64bcst</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed double precision floating-point values from ymm1 and ymm3/m256/m64bcst, subtract ymm2 and put result in ymm1 subject to writemask k1.</td></tr>
<tr>
<td>EVEX.256.66.0F38.W1 AA /r VFMSUB213PD ymm1 {k1}{z}, ymm2, ymm3/m256/m64bcst</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed double precision floating-point values from ymm1 and ymm2, subtract ymm3/m256/m64bcst and put result in ymm1 subject to writemask k1.</td></tr>
<tr>
<td>EVEX.256.66.0F38.W1 BA /r VFMSUB231PD ymm1 {k1}{z}, ymm2, ymm3/m256/m64bcst</td>
<td>B</td>
<td>V/V</td>
<td>AVX512VL AVX512F</td>
<td>Multiply packed double precision floating-point values from ymm2 and ymm3/m256/m64bcst, subtract ymm1 and put result in ymm1 subject to writemask k1.</td></tr>
<tr>
<td>EVEX.512.66.0F38.W1 9A /r VFMSUB132PD zmm1 {k1}{z}, zmm2, zmm3/m512/m64bcst{er}</td>
<td>B</td>
<td>V/V</td>
<td>AVX512F</td>
<td>Multiply packed double precision floating-point values from zmm1 and zmm3/m512/m64bcst, subtract zmm2 and put result in zmm1 subject to writemask k1.</td></tr>
<tr>
<td>EVEX.512.66.0F38.W1 AA /r VFMSUB213PD zmm1 {k1}{z}, zmm2, zmm3/m512/m64bcst{er}</td>
<td>B</td>
<td>V/V</td>
<td>AVX512F</td>
<td>Multiply packed double precision floating-point values from zmm1 and zmm2, subtract zmm3/m512/m64bcst and put result in zmm1 subject to writemask k1.</td></tr>
<tr>
<td>EVEX.512.66.0F38.W1 BA /r VFMSUB231PD zmm1 {k1}{z}, zmm2, zmm3/m512/m64bcst{er}</td>
<td>B</td>
<td>V/V</td>
<td>AVX512F</td>
<td>Multiply packed double precision floating-point values from zmm2 and zmm3/m512/m64bcst, subtract zmm1 and put 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 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>Performs a set of SIMD multiply-subtract computation on packed double precision floating-point values using three source operands and writes the multiply-subtract results in the destination operand. The destination operand is also the first source operand. The second operand must be a SIMD register. The third source operand can be a SIMD register or a memory location.</p>
<p>VFMSUB132PD: Multiplies the two, four or eight packed double precision floating-point values from the first source operand to the two, four or eight packed double precision floating-point values in the third source operand. From the infinite precision intermediate result, subtracts the two, four or eight packed double precision floating-point values in the second source operand, performs rounding and stores the resulting two, four or eight packed double precision floating-point values to the destination operand (first source operand).</p>
<p>VFMSUB213PD: Multiplies the two, four or eight packed double precision floating-point values from the second source operand to the two, four or eight packed double precision floating-point values in the first source operand. From the infinite precision intermediate result, subtracts the two, four or eight packed double precision floating-point values in the third source operand, performs rounding and stores the resulting two, four or eight packed double precision floating-point values to the destination operand (first source operand).</p>
<p>VFMSUB231PD: Multiplies the two, four or eight packed double precision floating-point values from the second source to the two, four or eight packed double precision floating-point values in the third source operand. From the infinite precision intermediate result, subtracts the two, four or eight packed double precision floating-point values in the first source operand, performs rounding and stores the resulting two, four or eight packed double 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 64-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 subtraction with infinite precision inputs and outputs (no
rounding).
</pre>
<h4 id="vfmsub132pd-dest--src2--src3--vex-encoded-versions-">VFMSUB132PD DEST, SRC2, SRC3 (VEX encoded versions)<a class="anchor" href="#vfmsub132pd-dest--src2--src3--vex-encoded-versions-">
</a></h4>
<pre>IF (VEX.128) THEN
MAXNUM := 2
ELSEIF (VEX.256)
MAXNUM := 4
FI
For i = 0 to MAXNUM-1 {
n := 64*i;
DEST[n+63:n] := RoundFPControl_MXCSR(DEST[n+63:n]*SRC3[n+63:n] - SRC2[n+63:n])
}
IF (VEX.128) THEN
DEST[MAXVL-1:128] := 0
ELSEIF (VEX.256)
DEST[MAXVL-1:256] := 0
FI
</pre>
<h4 id="vfmsub213pd-dest--src2--src3--vex-encoded-versions-">VFMSUB213PD DEST, SRC2, SRC3 (VEX encoded versions)<a class="anchor" href="#vfmsub213pd-dest--src2--src3--vex-encoded-versions-">
</a></h4>
<pre>IF (VEX.128) THEN
MAXNUM := 2
ELSEIF (VEX.256)
MAXNUM := 4
FI
For i = 0 to MAXNUM-1 {
n := 64*i;
DEST[n+63:n] := RoundFPControl_MXCSR(SRC2[n+63:n]*DEST[n+63:n] - SRC3[n+63:n])
}
IF (VEX.128) THEN
DEST[MAXVL-1:128] := 0
ELSEIF (VEX.256)
DEST[MAXVL-1:256] := 0
FI
</pre>
<h4 id="vfmsub231pd-dest--src2--src3--vex-encoded-versions-">VFMSUB231PD DEST, SRC2, SRC3 (VEX encoded versions)<a class="anchor" href="#vfmsub231pd-dest--src2--src3--vex-encoded-versions-">
</a></h4>
<pre>IF (VEX.128) THEN
MAXNUM := 2
ELSEIF (VEX.256)
MAXNUM := 4
FI
For i = 0 to MAXNUM-1 {
n := 64*i;
DEST[n+63:n] := RoundFPControl_MXCSR(SRC2[n+63:n]*SRC3[n+63:n] - DEST[n+63:n])
}
IF (VEX.128) THEN
DEST[MAXVL-1:128] := 0
ELSEIF (VEX.256)
DEST[MAXVL-1:256] := 0
FI
</pre>
<h4 id="vfmsub132pd-dest--src2--src3--evex-encoded-versions--when-src3-operand-is-a-register-">VFMSUB132PD DEST, SRC2, SRC3 (EVEX encoded versions, when src3 operand is a register)<a class="anchor" href="#vfmsub132pd-dest--src2--src3--evex-encoded-versions--when-src3-operand-is-a-register-">
</a></h4>
<pre>(KL, VL) = (2, 128), (4, 256), (8, 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 * 64
IF k1[j] OR *no writemask*
THEN DEST[i+63:i] :=
RoundFPControl(DEST[i+63:i]*SRC3[i+63:i] - SRC2[i+63:i])
ELSE
IF *merging-masking* ; merging-masking
THEN *DEST[i+63:i] remains unchanged*
ELSE ; zeroing-masking
DEST[i+63:i] := 0
FI
FI;
ENDFOR
DEST[MAXVL-1:VL] := 0
</pre>
<h4 id="vfmsub132pd-dest--src2--src3--evex-encoded-versions--when-src3-operand-is-a-memory-source-">VFMSUB132PD DEST, SRC2, SRC3 (EVEX encoded versions, when src3 operand is a memory source)<a class="anchor" href="#vfmsub132pd-dest--src2--src3--evex-encoded-versions--when-src3-operand-is-a-memory-source-">
</a></h4>
<pre>(KL, VL) = (2, 128), (4, 256), (8, 512)
FOR j := 0 TO KL-1
i := j * 64
IF k1[j] OR *no writemask*
THEN
IF (EVEX.b = 1)
THEN
DEST[i+63:i] :=
RoundFPControl_MXCSR(DEST[i+63:i]*SRC3[63:0] - SRC2[i+63:i])
ELSE
DEST[i+63:i] :=
RoundFPControl_MXCSR(DEST[i+63:i]*SRC3[i+63:i] - SRC2[i+63:i])
FI;
ELSE
IF *merging-masking* ; merging-masking
THEN *DEST[i+63:i] remains unchanged*
ELSE ; zeroing-masking
DEST[i+63:i] := 0
FI
FI;
ENDFOR
DEST[MAXVL-1:VL] := 0
</pre>
<h4 id="vfmsub213pd-dest--src2--src3--evex-encoded-versions--when-src3-operand-is-a-register-">VFMSUB213PD DEST, SRC2, SRC3 (EVEX encoded versions, when src3 operand is a register)<a class="anchor" href="#vfmsub213pd-dest--src2--src3--evex-encoded-versions--when-src3-operand-is-a-register-">
</a></h4>
<pre>(KL, VL) = (2, 128), (4, 256), (8, 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 * 64
IF k1[j] OR *no writemask*
THEN DEST[i+63:i] :=
RoundFPControl(SRC2[i+63:i]*DEST[i+63:i] - SRC3[i+63:i])
ELSE
IF *merging-masking* ; merging-masking
THEN *DEST[i+63:i] remains unchanged*
ELSE ; zeroing-masking
DEST[i+63:i] := 0
FI
FI;
ENDFOR
DEST[MAXVL-1:VL] := 0
</pre>
<h4 id="vfmsub213pd-dest--src2--src3--evex-encoded-versions--when-src3-operand-is-a-memory-source-">VFMSUB213PD DEST, SRC2, SRC3 (EVEX encoded versions, when src3 operand is a memory source)<a class="anchor" href="#vfmsub213pd-dest--src2--src3--evex-encoded-versions--when-src3-operand-is-a-memory-source-">
</a></h4>
<pre>(KL, VL) = (2, 128), (4, 256), (8, 512)
FOR j := 0 TO KL-1
i := j * 64
IF k1[j] OR *no writemask*
THEN
IF (EVEX.b = 1)
THEN
DEST[i+63:i] :=
RoundFPControl_MXCSR(SRC2[i+63:i]*DEST[i+63:i] - SRC3[63:0])
+31:i])
ELSE
DEST[i+63:i] :=
RoundFPControl_MXCSR(SRC2[i+63:i]*DEST[i+63:i] - SRC3[i+63:i])
FI;
ELSE
IF *merging-masking* ; merging-masking
THEN *DEST[i+63:i] remains unchanged*
ELSE ; zeroing-masking
DEST[i+63:i] := 0
FI
FI;
ENDFOR
DEST[MAXVL-1:VL] := 0
</pre>
<h4 id="vfmsub231pd-dest--src2--src3--evex-encoded-versions--when-src3-operand-is-a-register-">VFMSUB231PD DEST, SRC2, SRC3 (EVEX encoded versions, when src3 operand is a register)<a class="anchor" href="#vfmsub231pd-dest--src2--src3--evex-encoded-versions--when-src3-operand-is-a-register-">
</a></h4>
<pre>(KL, VL) = (2, 128), (4, 256), (8, 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 * 64
IF k1[j] OR *no writemask*
THEN DEST[i+63:i] :=
RoundFPControl(SRC2[i+63:i]*SRC3[i+63:i] - DEST[i+63:i])
ELSE
IF *merging-masking* ; merging-masking
THEN *DEST[i+63:i] remains unchanged*
ELSE ; zeroing-masking
DEST[i+63:i] := 0
FI
FI;
ENDFOR
DEST[MAXVL-1:VL] := 0
</pre>
<h4 id="vfmsub231pd-dest--src2--src3--evex-encoded-versions--when-src3-operand-is-a-memory-source-">VFMSUB231PD DEST, SRC2, SRC3 (EVEX encoded versions, when src3 operand is a memory source)<a class="anchor" href="#vfmsub231pd-dest--src2--src3--evex-encoded-versions--when-src3-operand-is-a-memory-source-">
</a></h4>
<pre>(KL, VL) = (2, 128), (4, 256), (8, 512)
FOR j := 0 TO KL-1
i := j * 64
IF k1[j] OR *no writemask*
THEN
IF (EVEX.b = 1)
THEN
DEST[i+63:i] :=
RoundFPControl_MXCSR(SRC2[i+63:i]*SRC3[63:0] - DEST[i+63:i])
ELSE
DEST[i+63:i] :=
RoundFPControl_MXCSR(SRC2[i+63:i]*SRC3[i+63:i] - DEST[i+63:i])
FI;
ELSE
IF *merging-masking* ; merging-masking
THEN *DEST[i+63:i] remains unchanged*
ELSE ; zeroing-masking
DEST[i+63: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>VFMSUBxxxPD __m512d _mm512_fmsub_pd(__m512d a, __m512d b, __m512d c);
</pre>
<pre>VFMSUBxxxPD __m512d _mm512_fmsub_round_pd(__m512d a, __m512d b, __m512d c, int r);
</pre>
<pre>VFMSUBxxxPD __m512d _mm512_mask_fmsub_pd(__m512d a, __mmask8 k, __m512d b, __m512d c);
</pre>
<pre>VFMSUBxxxPD __m512d _mm512_maskz_fmsub_pd(__mmask8 k, __m512d a, __m512d b, __m512d c);
</pre>
<pre>VFMSUBxxxPD __m512d _mm512_mask3_fmsub_pd(__m512d a, __m512d b, __m512d c, __mmask8 k);
</pre>
<pre>VFMSUBxxxPD __m512d _mm512_mask_fmsub_round_pd(__m512d a, __mmask8 k, __m512d b, __m512d c, int r);
</pre>
<pre>VFMSUBxxxPD __m512d _mm512_maskz_fmsub_round_pd(__mmask8 k, __m512d a, __m512d b, __m512d c, int r);
</pre>
<pre>VFMSUBxxxPD __m512d _mm512_mask3_fmsub_round_pd(__m512d a, __m512d b, __m512d c, __mmask8 k, int r);
</pre>
<pre>VFMSUBxxxPD __m256d _mm256_mask_fmsub_pd(__m256d a, __mmask8 k, __m256d b, __m256d c);
</pre>
<pre>VFMSUBxxxPD __m256d _mm256_maskz_fmsub_pd(__mmask8 k, __m256d a, __m256d b, __m256d c);
</pre>
<pre>VFMSUBxxxPD __m256d _mm256_mask3_fmsub_pd(__m256d a, __m256d b, __m256d c, __mmask8 k);
</pre>
<pre>VFMSUBxxxPD __m128d _mm_mask_fmsub_pd(__m128d a, __mmask8 k, __m128d b, __m128d c);
</pre>
<pre>VFMSUBxxxPD __m128d _mm_maskz_fmsub_pd(__mmask8 k, __m128d a, __m128d b, __m128d c);
</pre>
<pre>VFMSUBxxxPD __m128d _mm_mask3_fmsub_pd(__m128d a, __m128d b, __m128d c, __mmask8 k);
</pre>
<pre>VFMSUBxxxPD __m128d _mm_fmsub_pd (__m128d a, __m128d b, __m128d c);
</pre>
<pre>VFMSUBxxxPD __m256d _mm256_fmsub_pd (__m256d a, __m256d b, __m256d 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>
View git blame Copy permalink