Vs. Alioth binary-trees

[The Anh Tran]

dsimcha wrote:
> == Quote from The Anh Tran (trtheanh at gmail.com)'s article
>> Hi,
>> Could you measure spinlock with this bench?
>> http://shootout.alioth.debian.org/u32q/benchmark.php?test=binarytrees&lang=all
>> Thanks.
> 
> I tried to reply to your post last night with my modified gcx.d, but apparently
> posting attachments of that size (~80kb) silently fails.
> 
> There is no multithreaded implementation of the binary trees benchmark for D that
> I was able to find.  As far as the single-threaded version, spinlocks would make
> absolutely no difference because the druntime GC uses thread_needLock() to avoid
> any kind of lock on single-threaded code.
> 
> If you or anyone else wants to play around w/ my modified gcx.d code and try it
> under different use cases, I've posted it to http://cis.jhu.edu/~dsimcha/gcx.d.

I've failed to recompile druntime with your gcx.d. I'm still a D newbie 
:|. Ie: need someone to hold my hand and guide steps by steps ;)

This is my multithread alloc implementation.
If it uses std.gc, and runs with 2 threads; it is 5 times slower than 1 
thread !???
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: test.d
URL: <http://lists.puremagic.com/pipermail/digitalmars-d/attachments/20081209/3ea0eaf2/attachment.ksh>