Search found 469 matches

by rvida
Mon Apr 15, 2013 10:28 am
Forum: Computer Chess Club: General Topics
Topic: Extracting info from Critter session file
Replies: 5
Views: 1303

Re: Extracting info from Critter session file

I have only a vague idea what such a session file would be used for, so please correct me if I am wrong: this session file contains only (position, score) pairs that can be assigned by the user? Or does it also contain other information about the positions? It is very much like a persistent hash ta...
by rvida
Sun Apr 14, 2013 10:27 am
Forum: Computer Chess Club: General Topics
Topic: Extracting info from Critter session file
Replies: 5
Views: 1303

Re: Extracting info from Critter session file

Is there any way that I can extract text info from the session file No, because positions are not stored in the session file, only hash keys (like in polyglot opening books). or at least have a clue of the learning process that it is going on? There is a console mode command that probes the session...
by rvida
Fri Apr 05, 2013 11:57 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: New 6-piece tablebases
Replies: 194
Views: 104159

Re: New 6-piece tablebases

syzygy wrote: The tables use custom compression. Total compressed size:

Code: Select all

                       WDL          DTZ
up to 5 pieces       378 MB        561 MB
up to 6 pieces       68.3 GB       81.9 GB
Is the custom compression scheme any better than LZMA (which is used in Gaviota) ?
by rvida
Sat Mar 09, 2013 10:28 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: C vs ASM
Replies: 93
Views: 18691

Re: C vs ASM

and x86 throw an interruption every time you have an unaligned memory access, which might explain the significant performance cost. Although x86 can throw exception on unaligned data access, this feature is usually disabled except when using code profiling software. The exception is thrown only if ...
by rvida
Wed Mar 06, 2013 10:44 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: C vs ASM
Replies: 93
Views: 18691

Re: C vs ASM

Well, it seems that results are more random than anything else. The 'modfication' you show should in principle be a degradation on most CPUs. An ALU instruction with memory source does not fall in the 'RISC' class. It is decoded into 2 uOps (exactly the same uOps as the mov mem,reg, cmp reg,reg wou...
by rvida
Wed Mar 06, 2013 10:16 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: C vs ASM
Replies: 93
Views: 18691

Re: C vs ASM

rvida wrote: Slightly off-topic:
Didn't check with latest cpus, but the original K7 arch (iirc P6 too) had a branch-predictor limitation about the number of branch targets per I-cache line.
Ooops. Nothing to do with I-Cache. I meant the instruction fetch buffer.
by rvida
Wed Mar 06, 2013 10:10 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: C vs ASM
Replies: 93
Views: 18691

Re: C vs ASM

I get the feeling it's perhaps an alignment issue? Two thoughts for that: 1. When I activate the debug code the speed gain totally disappears, from 14.7 to 22.2 secs 2. The same happens when I make a small (in principle meaningless) change in the ASM code: old mov EAX,dword ptr key1[EBX*4] // eax=k...
by rvida
Wed Feb 27, 2013 9:39 am
Forum: Computer Chess Club: General Topics
Topic: Stockfish can't evaluate the KP vs K endgame?
Replies: 56
Views: 10486

Re: Stockfish can't evaluate the KP vs K endgame?

Regargind the GCC optimization, I'm not sure what you mean. Here is the code: for (int i = 4, rep = 1; i <= st().rule50; i += 2) { rep += sp[-i].key == st().key; if (rep >= 2 + (sp-i < sp0)) return true; } where sp is the current stack pointer and sp0 the stack...
by rvida
Tue Feb 26, 2013 12:54 pm
Forum: Computer Chess Club: General Topics
Topic: Stockfish can't evaluate the KP vs K endgame?
Replies: 56
Views: 10486

Re: Stockfish can't evaluate the KP vs K endgame?

I commited the bugfix in DiscoCheck https://github.com/lucasart/chess/commit/12f360a52dece35a6cea8d2bf6a26623ac5ac811 What's very important is that there is no (measurable) speed penalty, so it wouldn't hurt elo in anyway . This is a dangerous assumption. This patch makes the search space a bit lar...
by rvida
Fri Feb 15, 2013 12:45 am
Forum: Computer Chess Club: General Topics
Topic: Legal Reverse engineering of software
Replies: 43
Views: 6267

Re: Legal Reverse engineering of software

So Richard Vida was legal to decompile Houdini because he was satisfying his personal curiosity. As long as he is not using the decompiled code directly he is also legally improve Critter with ideas he found in Houdini??? I consulted an attorney skilled in IP matters (cost me a lot of €). I don't s...