Hi All,<div><br></div><div>�� �thanks to everyone finding bug cases. �Please keep it up. �I&#39;m anxious to find a reproducible case for a crash in (what I thought was fixed) bytecode to machine code pc mapping with the r2333 VMs which has a stack trace like the following:</div>
<div><br></div><div><div>Thread 0 Crashed: �Dispatch queue: com.apple.main-thread</div><div>0 � com.teleplace.Teleplace � � � � 0x000b8ca6 mapForbcpcperformUntilarg + 166 (cogit.c:10929)</div><div>1 � com.teleplace.Teleplace � � � � 0x000b8f3c mcPCForstartBcpcin + 44 (cogit.c:11480)</div>
<div><br></div><div>I get it only intermittently (ironically when pushing out a new version of the VM sources) and would love to find a reproducible case. �So if anyone sees this I&#39;d be indebted if you could spend some time trying to make it easily reproducible.</div>
<div><br></div><div>TIA</div><div>Eliot</div><br><div class="gmail_quote">On Thu, Dec 30, 2010 at 6:02 PM, Eliot Miranda <span dir="ltr">&lt;<a href="mailto:eliot.miranda@gmail.com">eliot.miranda@gmail.com</a>&gt;</span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">Hi All,<div><br></div><div>�� � I&#39;ve released a new version of Cog that has a substantially improved code generator along the lines of Peter Deutsch&#39;s HPS (VisualWorks) and various of Ian Piumarta&#39;s VMs. �These all use a simple tecnique to identify constant references in bytecode and to support a register-based calling convention. �While this does produce faster code it tends to accelerate low-level code much more than high-level code as you can see by the following benchmarks:</div>

<div><br></div><div><div>SimpleStackBasedCogit: [1 to: 100000000 do: [:i|]] timeToRun 691</div><div>StackToRegisterMappingCogit: [1 to: 100000000 do: [:i|]] timeToRun 192</div><div>192 - 691 / 6.91 -72%</div><div><br></div>

<div>SimpleStackBasedCogit: 0 tinyBenchmarks &#39;753495217 bytecodes/sec; 64769127 sends/sec&#39;</div><div>StackToRegisterMappingCogit: 0 tinyBenchmarks &#39;931756141 bytecodes/sec; 128157989 sends/sec&#39;</div><div>
931756141 - 753495217 / 7534952.17 -24%</div>
<div>128157989 - 64769127 / 647691.27 -98%</div><div><br></div><div>SimpleStackBasedCogit: [Compiler recompileAll] timeToRun 47013 (no transcript</div><div>StackToRegisterMappingCogit: [Compiler recompileAll] timeToRun 43406 (no transcript)</div>

<div>43406 - 47013 / 470.13 -7.67234594686576</div></div><div><br></div><div>The status of this code is essentially beta. �The test suite runs the same on the new code generator as on the old, but I think there are still bugs because I get the occasional transient error. �I am therefore very interested in any reproducible errors you can find.</div>

<div><br></div><div>The VMs (<a href="http://www.mirandabanda.org/files/Cog/VM/VM.r2334/" target="_blank">http://www.mirandabanda.org/files/Cog/VM/VM.r2334/</a>) contain a few other important changes:�</div><div><br></div>
<div>- a bug fix to bytecode&lt;-&gt;native pc mappng that produced incorrect results for methods containing blocks with ^-returns in them. �One symptom is incorrect highlighting of the pc in the debugger, althoguh symptoms could be much serious.</div>

<div><br></div><div>- jitting interpreted methods on backward branches. �Currently any interpreted method that performs more than 20 backward branches will be considered for JIT compilation and if it is suitable (default, &lt;= 60 literals) will be compiled to native code.</div>

<div><br></div><div>- new callback support. �I need to commit some changes to the Alien package to provide access to this but essentially the VM&#39;s callback support is now able to be ported to architectures with register-based calling conventions (ARM, PowerPC, SPARC etc). �I&#39;ll try and get the Alien code released soon, and to back-port the changes to the standard VM before the end of the holiday.</div>

<div><br></div><div>One thing that is still /not/ fixed is the lack of a SoundPlugin on win32. �Apologies. �I&#39;ll try and get a fix for this before the end of the holidays too, but time might be too tight. �There are other priorities such as harmonising the standard and Cog VMs for the 4.2 release.</div>

<div><br></div><div>best</div><div>Eliot</div><font color="#888888"><div><br></div>
</font></blockquote></div><br></div>