16 months agobinarytrees-threadpool: use compact class and mutex master binarytrees.vala-9.vala
Marco Trevisan (Treviño) [Mon, 17 Jan 2011 17:56:24 +0000] 
binarytrees-threadpool: use compact class and mutex

Mutex seems faster than atomic operations... Ohhh! :o

16 months agobinarytrees-thread: use compact class and mutex binarytrees.vala-8.vala
Marco Trevisan (Treviño) [Mon, 17 Jan 2011 17:41:13 +0000] 
binarytrees-thread: use compact class and mutex

Mutex seems faster than atomic operations... Ohhh! :o

16 months agobinarytrees: use compact class binarytrees.vala-7.vala
Marco Trevisan (Treviño) [Mon, 17 Jan 2011 17:35:22 +0000] 
binarytrees: use compact class

16 months agospectralnorm: first version based on pure C# implementation spectralnorm.vala
Marco Trevisan (Treviño) [Fri, 14 Jan 2011 18:36:46 +0000] 
spectralnorm: first version based on pure C# implementation

Un-threaded version.

16 months agoobject-allocator: new benchmark that allocates many objects and refs them. objectallocator.vala
Marco Trevisan (Treviño) [Fri, 14 Jan 2011 18:32:53 +0000] 
object-allocator: new benchmark that allocates many objects and refs them.

Compilable both with gobject and dova profiles. C and C# versions available at http://go.3v1n0.net/ieGMS8

16 months agobinarytrees-dova-thread: added new binarytree version for dova profile
Marco Trevisan (Treviño) [Fri, 14 Jan 2011 02:06:30 +0000] 
binarytrees-dova-thread: added new binarytree version for dova profile

It has great performances, finally! Added some little changes to the other binarytree versions too, but they are a looot slower (2,6 secs vs 26!)

16 months agobinarytrees-dova: removed the atoi dependency.
Marco Trevisan (Treviño) [Thu, 13 Jan 2011 19:23:07 +0000] 
binarytrees-dova: removed the atoi dependency.

Added static method for calculating the string

16 months agobinarytrees-dova: added support for Vala Dova profile
Marco Trevisan (Treviño) [Thu, 13 Jan 2011 18:29:55 +0000] 
binarytrees-dova: added support for Vala Dova profile

Version of binarytrees benchmark using Dova (http://gitorious.org/dova/dova-core) as vala profile, it shows a great performance improvement (it can be compared to C# in this case)

16 months agospectralnorm-thread: added the threaded version of this benchmark
Marco Trevisan (Treviño) [Mon, 10 Jan 2011 11:26:40 +0000] 
spectralnorm-thread: added the threaded version of this benchmark

The code works, but the output is not the same when there's more than 1 processor working; need to check.

16 months agofasta: added Fasta benchmark fasta.vala
Marco Trevisan (Treviño) [Mon, 10 Jan 2011 06:20:12 +0000] 
fasta: added Fasta benchmark

16 months agofannkuchredux-thread: added missing return value in thread run fannkuchredux.vala-6.vala
Marco Trevisan (Treviño) [Thu, 6 Jan 2011 04:23:37 +0000] 
fannkuchredux-thread: added missing return value in thread run

16 months agobuild-system: added meteor benchmark
Marco Trevisan (Treviño) [Wed, 5 Jan 2011 04:05:04 +0000] 
build-system: added meteor benchmark

16 months agometeor: fixed segfault, the benchmark now works! meteor.vala
Marco Trevisan (Treviño) [Wed, 5 Jan 2011 03:59:53 +0000] 
meteor: fixed segfault, the benchmark now works!

Yuppy! :P
But vala really needs fixing... :(

16 months agometeor: use fixed arrays when possible
Marco Trevisan (Treviño) [Wed, 5 Jan 2011 02:36:59 +0000] 
meteor: use fixed arrays when possible

16 months agometeor: implemented the Soln to_string() method
Marco Trevisan (Treviño) [Wed, 5 Jan 2011 02:33:58 +0000] 
meteor: implemented the Soln to_string() method

16 months agometeor: added first version translittered by the java implementation
Marco Trevisan (Treviño) [Wed, 5 Jan 2011 02:24:51 +0000] 
meteor: added first version translittered by the java implementation

This code still segfaults by the way :P

16 months agobinarytrees-threadpool: use AtomicInt instead of mutex binarytrees.vala-6.vala
Marco Trevisan (Treviño) [Tue, 4 Jan 2011 17:07:11 +0000] 
binarytrees-threadpool: use AtomicInt instead of mutex

16 months agofannkuchredux-thread: use AtomicInt instead of locks
Marco Trevisan (Treviño) [Tue, 4 Jan 2011 16:49:40 +0000] 
fannkuchredux-thread: use AtomicInt instead of locks

This improves speed!

16 months agofannkuchredux-thread: using fixed arrays also for return values... fannkuchredux.vala-5.vala
Marco Trevisan (Treviño) [Tue, 4 Jan 2011 14:30:06 +0000] 
fannkuchredux-thread: using fixed arrays also for return values...

16 months agofannkuchredux-thread: use fixed-size array for saving per-thread results
Marco Trevisan (Treviño) [Tue, 4 Jan 2011 14:12:09 +0000] 
fannkuchredux-thread: use fixed-size array for saving per-thread results

16 months agofannkuchredux-thread: always free the array... fannkuchredux.vala-4.vala
Marco Trevisan (Treviño) [Tue, 4 Jan 2011 04:17:37 +0000] 
fannkuchredux-thread: always free the array...

This reduces the memory usage, but increase the execution time :(

16 months agofannkuchredux-thread: use an GLib.Array instead of int* arrays in return
Marco Trevisan (Treviño) [Tue, 4 Jan 2011 04:08:08 +0000] 
fannkuchredux-thread: use an GLib.Array instead of int* arrays in return

This allow to make vala free the returned values

16 months agofannkuchredux-thread: use the GLib.Thread return functionality fannkuchredux.vala-3.vala
Marco Trevisan (Treviño) [Tue, 4 Jan 2011 03:51:01 +0000] 
fannkuchredux-thread: use the GLib.Thread return functionality

This improves the performances, but actually it seems to cause a memory
leak, cause the pointer is not free'd.

16 months agofannkuchredux-thread: added a first threaded version fannkuchredux.vala-2.vala
Marco Trevisan (Treviño) [Tue, 4 Jan 2011 03:12:15 +0000] 
fannkuchredux-thread: added a first threaded version

16 months agobuild system: valac enable multithreading support by default
Marco Trevisan (Treviño) [Tue, 4 Jan 2011 01:47:10 +0000] 
build system: valac enable multithreading support by default

16 months agobinarytrees-thread: use the thread return value for printing binarytrees.vala-5.vala
Marco Trevisan (Treviño) [Tue, 4 Jan 2011 01:39:37 +0000] 
binarytrees-thread: use the thread return value for printing

16 months agofannkuchredux: added first version based on un-threaded C# one fannkuchredux.vala
Marco Trevisan (Treviño) [Tue, 4 Jan 2011 01:04:06 +0000] 
fannkuchredux: added first version based on un-threaded C# one

16 months agobuild system: added an automake based build system
Marco Trevisan (Treviño) [Mon, 3 Jan 2011 23:11:44 +0000] 
build system: added an automake based build system

16 months agobinarytrees-threadpool: some optimizations binarytrees.vala-4.vala
Marco Trevisan (Treviño) [Mon, 3 Jan 2011 23:10:19 +0000] 
binarytrees-threadpool: some optimizations

16 months agobinarytrees-threadpool: added new threaded version using ThreadPool
Marco Trevisan (Treviño) [Mon, 3 Jan 2011 23:04:25 +0000] 
binarytrees-threadpool: added new threaded version using ThreadPool

It has better performances than the standard thread version

16 months agobinarytrees-thread: use the Sched.CpuSet get_cpu_count() func... binarytrees.vala-3.vala
Marco Trevisan (Treviño) [Mon, 3 Jan 2011 22:26:37 +0000] 
binarytrees-thread: use the Sched.CpuSet get_cpu_count() func...

16 months agosched.vapi: updated to sync the real library
Marco Trevisan (Treviño) [Mon, 3 Jan 2011 22:25:30 +0000] 
sched.vapi: updated to sync the real library

Plus added the convenience func get_cpu_count()

16 months agobinarytrees-thread: using sched.h for getting the CPU number
Marco Trevisan (Treviño) [Mon, 3 Jan 2011 18:55:03 +0000] 
binarytrees-thread: using sched.h for getting the CPU number

Added a vapi implementation

16 months agobinarytrees-thread: better TreeNode implementation.
Marco Trevisan (Treviño) [Mon, 3 Jan 2011 14:13:25 +0000] 
binarytrees-thread: better TreeNode implementation.

16 months agobinarytrees-thread: added max thread number.
Marco Trevisan (Treviño) [Mon, 3 Jan 2011 14:12:34 +0000] 
binarytrees-thread: added max thread number.

Actually just fixed to 4, to be adapted to CPU number

16 months agobinarytree: reduce the memory usage and complexity removing the Next class binarytrees.vala-2.vala
Marco Trevisan (Treviño) [Fri, 31 Dec 2010 08:31:03 +0000] 
binarytree: reduce the memory usage and complexity removing the Next class

16 months agobinarytrees: added the threaded version
Marco Trevisan (Treviño) [Fri, 31 Dec 2010 01:13:53 +0000] 
binarytrees: added the threaded version

16 months agonbody: avoid sun to be scanned when calculating px, py and pz nbody.vala
Marco Trevisan (Treviño) [Thu, 30 Dec 2010 23:01:31 +0000] 
nbody: avoid sun to be scanned when calculating px, py and pz

Add a slice to bodies, to avoid un-need computation

16 months agonbody: fixed error, use classes instead of structs
Marco Trevisan (Treviño) [Thu, 30 Dec 2010 22:41:29 +0000] 
nbody: fixed error, use classes instead of structs

This allows to modify the real variables

16 months agonbody: code cleanup, better read-ability...
Marco Trevisan (Treviño) [Thu, 30 Dec 2010 22:39:28 +0000] 
nbody: code cleanup, better read-ability...

16 months agoAdded nbody benchmark, but still not working as expected
Marco Trevisan (Treviño) [Thu, 30 Dec 2010 15:06:16 +0000] 
Added nbody benchmark, but still not working as expected

Based on the C# implementation by Isaac Gouy and Robert F. Tobler

16 months agoFirst commit, added binarytrees benchmark binarytrees.vala
Marco Trevisan (Treviño) [Thu, 30 Dec 2010 15:01:49 +0000] 
First commit, added binarytrees benchmark

Based on vala version by Marek Safar