<font size=+2 face=Verdana><b><a name="Trends">Changes over Time</font></b>
</td></tr></table></td></tr></table></center><p>
-GNUplot <b>charts</b> showing a couple of things:
-<ol>
-<li>Compile Time
-<li>Testing Time
-<li>Time to run some programs, say all of Olden
-<li>Time to compile something, say Olden.
-<li>Size of bytecode files for Olden
-</ol>
+
+Here are some charts showing how the LLVM optimizer and code generators are
+changing over time. For now we use the Olden benchmark suite to measure this,
+but eventually we will switch to using SPEC CPU2000. All programs are run with
+"LARGE_PROBLEM_SIZE" enabled. Click on any of the charts to get a larger
+version.<p>
+
+<h2>Compilation Measurements:</h2>
+
+<table border="0" align=center>
+<tr>
+<td width=50% align=center>
+<a href="running_Olden_bytecode_large.png"><img width=480 height=360 border=0 src="running_Olden_bytecode.png"></a><br>
+Size of LLVM bytecode files
+</td>
+<td width=50% align=center>
+<a href="running_Olden_machcode_large.png"><img width=480 height=360 border=0 src="running_Olden_machcode.png"></a><br>
+Size of native machine code for each program (generated by the JIT)
+</td></tr>
+<tr>
+<td align=center>
+<a href="running_Olden_opt_time_large.png"><img width=480 height=360 border=0 src="running_Olden_opt_time.png"></a><br>
+Time to run the LLVM optimizer on each program
+</td>
+<td></td></tr>
+</table>
+
+<h2>Program Execution Measurements:</h2>
+
+<table border="0" align=center>
+<tr>
+<td width=50% align=center>
+<a href="running_Olden_cbe_time_large.png"><img width=480 height=360 border=0 src="running_Olden_cbe_time.png"></a><br>
+Execution time for CBE generated executable
+</td>
+<td width=50% align=center>
+<a href="running_Olden_llc_time_large.png"><img width=480 height=360 border=0 src="running_Olden_llc_time.png"></a><br>
+Execution time for the LLC generated executable
+</td></tr>
+
+<tr>
+<td align=center>
+<a href="running_Olden_jit_time_large.png"><img width=480 height=360 border=0 src="running_Olden_jit_time.png"></a><br>
+Execution time for program in the JIT
+</td>
+<td></td></tr>
+</table>
+
+
+
<br><br><center>
<table border="0" cellspacing="0" cellpadding="2"><tr><td bgcolor="#000000">
GCC output: greater than 1 is a speedup, less than 1 is a slowdown.</li>
<li><a name="GCC/CBE">GCC/CBE</a> - The speed-up of the CBE output vs the native
GCC output: greater than 1 is a speedup, less than 1 is a slowdown.</li>
+<li><a name="LLC-LS">LLC-LS</a> - How long does the program generated by the static
+ backend LLC take to execute the program, when compiled with the linear scan
+ register allocator. This is temporary, for tuning.</li>
</ol><p>
A complete log of testing