Changes between Version 167 and Version 168 of benchmarks

Timestamp:

Jan 29, 2013, 1:55:13 PM (13 years ago)

Author:

goddard

Comment:

--

benchmarks

@@ -197 +197 @@
 <td> 1913 <td> 1856  <td> 207 <td> 5334 ? <td> 420
 <td>1509 <td>12.1 <td>10.9 <td> 36.0 <td> 21.0 <td> 60.2
-<td><a href="../ticket/8459">8459</a>, solid must be wrong.
+<td><a href="../ticket/8459">8459</a>, solid must be wrong, note <a href="#note7">7</a>.
 <td>5/10
 <td>1.4.1_b30365
@@ -215 +215 @@
 <td> 1889 <td> 1750+  <td> 226 <td> ? <td> 512
 <td>1322 <td>603 <td>501 <td> 1895 <td> 170 <td> 142
-<td><a href="../ticket/11148">11727</a>
+<td><a href="../ticket/11148">11727</a>, note <a href="#note7">7</a>.
 <td>1/13
 <td>1.7_b38101 32-bit
@@ -3280 +3280 @@
 </p>
 
-
 <a name="note6">
 <p>
@@ -3294 +3293 @@
 </p>
 
+<a name="note7">
+<p>
+<b>(7)</b>
+The solid benchmark sometimes reports incorrect high scores, for example 10000, I believe because the texture allocation failed
+and Chimera is merely drawing a yellow outline box instead of drawing the data.  Something else must be going on because a
+size 10000 test would use a 10000 cubed volume which would not fit in memory.  Needs further investigation.
+</p>
 }}}