CCRL Discussion Board

I was thinking more of an extension of your computation order: b) pawnless endings -> all with 1 pawn -> all with 2 pawns -> ... -> all with 14 pawns I realize I made some mistakes, but the idea should still hold. (To compute positions without few piece types, then add piece types as you can afford ...

Instead of limiting the number of pieces, you could limit the number of piece types. Doing up to 4 piece types should be doable now but the full set is about 2^16 times bigger. (actually less as this does not consider that pawns can't be on final rank). Compression seems very unlikely to buy you ver...

Yes, there are certainly other moves that are irreversible (other than P-push, capture, e.p.-capture rejection and castling-rights-changing moves). Your example is indeed one such - nice - however, I think it's not possible to identify all of these (or maybe any) with a simple, computer-implementab...

1. Sharing information across cores Don't! It will only complicate things without improving performance much. 2. Auto-sensory of prioritization, keeping parallelism in mind Just use one core. That helps keep complexity of writing the generator down. 3. Bidirectional Asymmetric Processing That is an...

Ok, 4 years then. That is reasonably quick. I recommend testing the program on commodity hardware first too see how much the cpu gets used when computing a 7-man. The computation does not even need to finish to do this. If the system you think of building has 12 processors, I'd recommend using mothe...

Thinking about hardware can be fun. Have you considered waiting 4 years before starting? Then you should be able to buy twice as fast hardware at half price and still finish 3.5 months earlier than if you buy now

Ah, so speed is about 8 min/ending. That means 7-man should take about 1 month/ending.
Managed to download after figuring out there was a captcha. No source code for generator as far as I can see.

I could not get download to work, but from the posted documentation it seems to be a program for windows that is limited mainly by ram size and disk speed. If anyone has a link to the source I would like to have a look at it. I don't understand you about the speed. Do you mean a five man ending take...

I am talking about improving/writing programs for computing and using 6 annd 7-man endgame tablebases. I think I have seen mention of some improvements in last 6 months and as for what I have been doing is changing my planned algorithm to make it easier to implement. (caring less about theoretical p...

Yes, work is still going on, but only very little.

Got it

$ time wget --recursive http://olympuschess.com/egtb/gaviota/
...
FÄRDIG --2010-06-15 11:59:22--
Hämtade: 482 filer, 14G i 38m 44s (6,13 MB/s)

real 40m34.440s
user 0m9.205s
sys 1m37.062s

The idea of hgm is to examine part of the attacking sides moves at the same time as all the defending sides moves. That leads to different partitioning possibilities and also avoids the need to store partial results for both sides. If you only compute one side at a time, the defending side is able t...

Rather than doing something practical I have done some more theorethical optimizing. When computing black to move (BTM) positions from white to move (WTM) positions you need the following info: White wins with distance to zeroing (DTZ) = N-1, white wins with DTZ = N-2, white wins with DTZ < N. Given...

... if i may ask a (semi-) naive question: how much time cpu-wise were you planning on devoting to the codec? i ask because currently most of the computation is (or should be) done in the 512k L1 cache of the cpu, while loading to and from RAM and disc as necessary. given that L1 RAM is considerabl...

A 64-bit computer running a 32-bit operating system is a software problem that has already been solved. Just replace with a 64-bit OS. I hear there is some free software for that.

I wrote an overview at: viewtopic.php?f=6&t=3652&p=47585#p47585
Then this thread has most explained: viewtopic.php?f=6&t=3458&p=34282#p34282

The problem with SSD drives is that the bytes on it are more expensive. The advantage of mine and h.g.mullers algorithm is that they work well with normal hard disks because they don't need all that many seeks. With mine, reading and writing disk in blocks of 16 MB is reasonable on a machine with 6-...

Ok, I have removed the tablebase.sesse.net folder to avoid confusion.

I put 3-4-5 from http://www.olympuschess.com/tb/ at http://www.teo.se/345/ as I download it.
I think my connection has more available bandwidth.

GPU cards might later on be useful for compressing the database. I imagine that might be limited by CPU rather than by memory bandwidth.
Generating endgame tables is still much more a software problem than a hardware problem.

It's now 2008 and even I can afford a 1-2tb every couple months. So disk space doesnt seem to be the issue anymore. So where are the 7-men tables and has anyone have any updates? Here is one interesting post from Rybka forum: After reading this thread I’ve made a conclusion that most people can’t i...

I have written some C++ code for computing illegality/invalidity status of positions. Using one core it does this in approximately 15min for a 7-man endgame. (simulated by 640*KRRRvK) There are many opportunities for optimization left so it should be possible to do it quickly enough that it can be u...

Avoiding storing illegal positions looks rather promising, the more pieces, the more useful is my guess. I'm Experimenting with Erlang (still learning) and for KRRvK I get: [8325184,16777216,0.4962196350097656] legal for [white_rook,white_rook] and for KRRvKR: [508975152,1073741824,0.474020048975944...

Adding: #include <malloc.h> #include <stdlib.h> and casting pointers to int for printing "%x" makes the warnings go away when compiling with gcc -O2 -m32. Still crashes though, and when I compile without -O2 it crashes quickly: time ./a.out e7dbf008 e7dc0000 KBB_KN Segmentation fault real ...

Yes, changing SIZE the way I did was a bad idea. I did it cause the subtraction looked strange to me and the program crashed before the change but not after. The alignment code change is more sensible. What I would like is having the code work correctly on my 64-bit system. Maybe I'll go look for so...