Changes between Version 32 and Version 33 of GcnInstrsVop3


Ignore:
Timestamp:
11/25/17 22:00:28 (6 years ago)
Author:
trac
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • GcnInstrsVop3

    v32 v33  
    7474</tbody>
    7575</table>
    76 <p>List of fields for VOP3A/VOP3B encoding (GCN 1.2):</p>
     76<p>List of fields for VOP3A/VOP3B encoding (GCN 1.2/1.4):</p>
    7777<table>
    7878<thead>
     
    9393<td>ABS</td>
    9494<td>Absolute modifiers for source operands (VOP3A)</td>
     95</tr>
     96<tr>
     97<td>11-14</td>
     98<td>OPSEL</td>
     99<td>Operand selection (VOP3A) (GCN 1.4)</td>
    95100</tr>
    96101<tr>
     
    149154<li>-SRC - negate floating point value from source operand. Applied after ABS modifier.</li>
    150155<li>ABS(SRC), |SRC| - apply absolute value to source operand</li>
     156<li>OPSEL:VALUE|[B0,...] - operand half selection (0 - lower 16-bits, 1 - bits)</li>
    151157</ul>
    152158<p>NOTE: OMOD modifier doesn't work if output denormals are allowed
     
    157163<p>Negation and absolute value can be combined: <code>-ABS(V0)</code>. Modifiers CLAMP and
    158164OMOD (MUL:2, MUL:4 and DIV:2) can be given in random order.</p>
     165<p>Operand half selection (OPSEL) take value with bits number depends of number operands.
     166Last bit control destination operand. Zero in bit choose lower 16-bits in dword,
     167one choose higher 16-bits. Example: op_sel:[0,1,1] - higher 16-bits in second source and
     168in destination.</p>
    159169<p>Limitations for operands:</p>
    160170<ul>