Paolo, >The files are in cvs, will add them to the tarball for next release.
I need to finish my assignment in the next couple of weeks. So could you please send me this document on mini. As I finish my invest on JIT I will start looking into garbage collection. I already did some measurement on the MS GC. So any documentation on the specific use of the Boehm collector in Mono is also welcome. I am starting to look into this the next couple of days. Thanks for helping me out, Tom PS: by the way I saw that in Mono verion 0.24 it is no longer possible to generate assembly code!? I used this previously to examine the code generated for a number of small benchmarks I have written. -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Paolo Molaro Sent: Wednesday, May 28, 2003 9:18 AM To: [EMAIL PROTECTED] Subject: Re: [Mono-list] Need some short answers on the new Mono JIT / Mini On 05/27/03 Tom Fransen wrote: > - My first question is: were can I find the document on mini that is > mentioned in the release notes??? > I searched the Mono TAR files but I cannot find it. I don't know if it is > stored in CVS. > I have no experience with CVS , who can help me out here? The files are in cvs, will add them to the tarball for next release. > - If I start a program I can use the option -O=all to enable all > optimizations. > Which optimizations are used when I do not use the -O option, what are the > default optimizations ??? The default optimizations may change over time as we add more optimizations and as we consider them stable enough. Optimizations that are costly to perform may not enter into the default set, though: maybe some student will be interested in doing some profile-directed recompilation of methods... :-) The default optimizations also depend on the processor in the system where you run the JIT. If you want to set some defined optimizations, you can use: -O=-all,your_optimizations The '-all' part will disable all the default optimizations. > - I assume that the mono JIT compiles the code on a method basis (like MS > JIT). Yes. > However if I use the -O=all option the difference in duration between the > first > and the second invocation is almost zero!! > So in other words there is no initial cost for the JIT. > Can anybody explain why this is??????????? My guess is that the method you used to measure the time is incorrect or not precise enough to give reliable results:-) lupus -- ----------------------------------------------------------------- [EMAIL PROTECTED] debian/rules [EMAIL PROTECTED] Monkeys do it better _______________________________________________ Mono-list maillist - [EMAIL PROTECTED] http://lists.ximian.com/mailman/listinfo/mono-list _______________________________________________ Mono-list maillist - [EMAIL PROTECTED] http://lists.ximian.com/mailman/listinfo/mono-list
