| <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd"> |
| |
| <html> |
| |
| <head> |
| <title>cmp<kind></title> |
| <link rel=stylesheet href="opcode.css"> |
| </head> |
| |
| <body> |
| |
| <h1>cmp<kind></h1> |
| |
| <h2>Purpose</h2> |
| |
| <p> |
| Perform the indicated floating point or long comparison, storing 0 if the two |
| arguments are equal, 1 if the second argument is larger, or -1 if the first |
| argument is larger. The "bias" listed for the floating point operations |
| indicates how NaN comparisons are treated: "Gt bias" instructions return 1 for |
| NaN comparisons, and "lt bias" instructions return -1. |
| </p> |
| <p> |
| For example, to check to see if floating point a < b, then it is advisable to |
| use cmpg-float; a result of -1 indicates that the test was true, and the other |
| values indicate it was false either due to a valid comparison or because one |
| or the other values was NaN. |
| </p> |
| |
| <h2>Details</h2> |
| |
| <table class="instruc"> |
| <thead> |
| <tr> |
| <th>Op & Format</th> |
| <th>Mnemonic / Syntax</th> |
| <th>Arguments</th> |
| </tr> |
| </thead> |
| <tbody> |
| <tr> |
| <td>2d..31 23x</td> |
| <td>cmp<i>kind</i> vAA, vBB, vCC<br/> |
| 2d: cmpl-float <i>(lt bias)</i><br/> |
| 2e: cmpg-float <i>(gt bias)</i><br/> |
| 2f: cmpl-double <i>(lt bias)</i><br/> |
| 30: cmpg-double <i>(gt bias)</i><br/> |
| 31: cmp-long |
| </td> |
| <td><code>A:</code> destination register (8 bits)<br/> |
| <code>B:</code> first source register or pair<br/> |
| <code>C:</code> second source register or pair</td> |
| </tr> |
| </tbody> |
| </table> |
| |
| <h2>Constraints</h2> |
| |
| <ul> |
| <li> |
| A, B and C must be valid register indices in the current stack frame. |
| </li> |
| <li> |
| For the two -float variants, both vB and vC must be of type float. |
| </li> |
| <li> |
| For the two -double variants, both vB and vC must be the lower part of a |
| register pair holding a double value. |
| </li> |
| <li> |
| For the -long variant, both both vB and vC must be the lower part of a |
| register pair holding a long value. |
| </li> |
| </ul> |
| |
| <h2>Behavior</h2> |
| |
| <ul> |
| <li> |
| The values of registers vB and vC are compared. The result, which is stored |
| in vA, is one of the following three: |
| <ul> |
| <li> |
| If vB < vC, then vA'=-1. |
| </li> |
| <li> |
| If vB == vC, then vA'=0. |
| </li> |
| <li> |
| If vC > vC, then vA'=1. |
| </li> |
| </ul> |
| </li> |
| <li> |
| For the -float and -double variants, an addition "bias" specifies what |
| happens if one or both of the arguments are NaN: |
| <ul> |
| <li> |
| A "lt bias" results in vA'=-1. |
| </li> |
| <li> |
| A "gt bias" results in vA'=1. |
| </li> |
| </ul> |
| </li> |
| <li> |
| If v(A-1) is the lower half of a register pair, v(A-1)' becomes undefined. |
| </li> |
| <li> |
| If v(A+1) is the upper half of a register pair, v(A+1)' becomes undefined. |
| </li> |
| </ul> |
| |
| <h2>Exceptions</h2> |
| |
| <p> |
| None. |
| </p> |
| |
| </body> |
| </html> |
