Chess4u
Would you like to react to this message? Create an account in a few clicks or log in to continue.

Chess4uLog in

Welcome to the chess for you Forum.chess for you is a place to discuss and share anything related a Computer Chess.Join A Community That Cares About You.

Free Chess Opening Book for Every 15 Days

Stockfish Latest Versions

power_settings_newLogin to reply
+2
Chess4U
Waldemar
6 posters

descriptionStockfish Latest Versions EmptyStockfish Latest Versions

more_horiz
Stockfish Latest Versions

[You must be registered and logged in to see this link.]

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
TRY THIS

!! latest version !!


Windows x64 for Haswell CPUs
Windows x64 for modern computers
Windows x64
Windows 32
Linux x64 for Haswell CPUs
Linux x64 for modern computers
Linux x64
Author: mbootsector 
Date: Mon Nov 2 10:05:43 2015 +0000 
Timestamp: 1446458743 

Pick bestmove from the deepest thread. 

STC: 
LLR: 2.96 (-2.94,2.94) [0.00,5.00] 
Total: 26930 W: 4441 L: 4214 D: 18275 

LTC: 
LLR: 2.96 (-2.94,2.94) [0.00,5.00] 
Total: 7783 W: 1017 L: 876 D: 5890 

No functional change in single thread mode 

Resolves #485 

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
!! latest version !!


Windows x64 for Haswell CPUs
Windows x64 for modern computers
Windows x64
Windows 32
Linux x64 for Haswell CPUs
Linux x64 for modern computers
Linux x64
Author: Marco Costalba 
Date: Tue Nov 3 11:27:00 2015 +0100 
Timestamp: 1446546420 

Get rid of timer thread 

Unfortunately std::condition_variable::wait_for() 
is not accurate in general case and the timer thread 
can wake up also after tens or even hundreds of 
millisecs after time has elapsded. CPU load, process 
priorities, number of concurrent threads, even from 
other processes, will have effect upon it. 

Even official documentation says: "This function may 
block for longer than timeout_duration due to scheduling 
or resource contention delays." 

So retire timer and use a polling scheme based on a 
local thread counter that counts search() calls and 
a small trick to keep polling frequency constant, 
independently from the number of threads. 

Tested for no regression at very fast TC 2+0.05 th 7: 
LLR: 2.96 (-2.94,2.94) [-3.00,1.00] 
Total: 32969 W: 6720 L: 6620 D: 19629 

TC 2+0.05 th 1: 
LLR: 2.95 (-2.94,2.94) [-3.00,1.00] 
Total: 7765 W: 1917 L: 1765 D: 4083 

And at STC TC, both single thread 
LLR: 2.96 (-2.94,2.94) [-3.00,1.00] 
Total: 15587 W: 3036 L: 2905 D: 9646 

And with 7 threads 
LLR: 2.95 (-2.94,2.94) [-3.00,1.00] 
Total: 8149 W: 1367 L: 1227 D: 5555 

bench: 8639247 

descriptionStockfish Latest Versions EmptyOrka 021115

more_horiz
Orka 021115
[You must be registered and logged in to see this link.] :

descriptionStockfish Latest Versions EmptySF 06/11/15 MZ x64

more_horiz
SF 06/11/15 MZ x64

-SF061115MZ 64 BMI2
-SF061115MZ 64 POPCNT
-SF061115MZ 64

[You must be registered and logged in to see this link.]

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
THX.

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
thank you

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
Stockfish 7 coming soon [You must be registered and logged in to see this image.]

a word from SF maintainer

zamar 

Nov 7

"
Dear Stockfish contributors and enthusiasts,

I've been very positively surprised by the progress that has been made since the release of Stockfish 6: 
* The latest regression test showed ELO gain of 60 points in Single Threaded mode
* On top of this we have significant SMP improvements.

This is much more than I thought was possible. So well done, everybody! Together, with the right tools, we can make seemingly impossible things possible.

TCEC Superfinal has just started up. It's a good check point to see where we currently stand compared to Komodo. Again nobody can predict the result, which makes the final exciting to follow.
Game 1 (where SF unfortunately was on the losing side) demonstrated how well modern engines are able to understand positional compensation. Despite white being two pawns down, both SF and Komodo were able to see that due to positional factors (and maybe due to very very deep tactics) white was actually better. This wasn't the case 5 years ago.

Is it time for SF7? Soon, it will be. Again we will want to make a high quality release. We have just introduced new SMP implementation, and changed timer implementation, so we must not rush too much, to make sure that we give people enough time to test the new implementation and report any problems, and of course to see how we do in TCEC finals. So my current preference would be to release SF7 in early January 2016. We should have the same kind of 1-2 weeks freeze before the release as last time (We caught many important bugs last time during the freeze!).

Meanwhile keep the patches coming! :-)"

[You must be registered and logged in to see this link.]

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
Good news!

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
Windows x64 for Haswell CPUs
Windows x64 for modern computers
Windows x64
Windows 32
Linux x64 for Haswell CPUs
Linux x64 for modern computers
Linux x64
Author: lucasart 
Date: Tue Nov 10 21:41:42 2015 +0000 
Timestamp: 1447191702 

Ensure that rootDepth < DEPTH_MAX 

Indeed, if we use a depth >= DEPTH_MAX, we start having negative depth in the 
TT (due to int8_t cast). 

No functional change in single thread mode 

Resolves #490 

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
!! latest version !!


Windows x64 for Haswell CPUs
Windows x64 for modern computers
Windows x64
Windows 32
Linux x64 for Haswell CPUs
Linux x64 for modern computers
Linux x64
Author: lucasart 
Date: Tue Nov 10 21:46:02 2015 +0000 
Timestamp: 1447191962 

Avoid friend 

operator<<(os, pos) does not need to access any private members of pos. 

No functional change. 

Resolves #492 

descriptionStockfish Latest Versions EmptyStockfish 12/11/15 IP

more_horiz
Stockfish 12/11/15 IP

Stockfish 121115IP x32 gen
Stockfish 121115IP x32 old
Stockfish 121115IP x32 sse
Stockfish 121115IP x64 bmi2
Stockfish 121115IP x64 gen
Stockfish 121115IP x64 pop
Stockfish 121115IP x64 sse

[You must be registered and logged in to see this link.]

Thanks to Ipman sunny

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
Windows x64 for Haswell CPUs
Windows x64 for modern computers
Windows x64
Windows 32
Linux x64 for Haswell CPUs
Linux x64 for modern computers
Linux x64
Author: Marco Costalba 
Date: Fri Nov 13 08:13:59 2015 +0100 
Timestamp: 1447398839 

Fix broken UCI 'wait for stop' 

When we reach the maximum depth, we can finish the 
search without a raise of Signals.stop. However, if 
we are pondering or in an infinite search, the UCI 
protocol states that we shouldn't print the best move 
before the GUI sends a "stop" or "ponderhit" command. 

It was broken by lazy smp. Fix it by moving the stopping 
of the threads after waiting for GUI. 

No functional change. 

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
!! latest version !!


Windows x64 for Haswell CPUs
Windows x64 for modern computers
Windows x64
Windows 32
Linux x64 for Haswell CPUs
Linux x64 for modern computers
Linux x64
Author: Marco Costalba 
Date: Fri Nov 13 08:22:44 2015 +0100 
Timestamp: 1447399364 

Retire ThreadBase 

Now that we don't have anymore TimerThread, there is 
no need of this long class hierarchy. 

Also assorted reformatting while there. 

To verify no regression, passed at STC with 7 threads: 
LLR: 2.97 (-2.94,2.94) [-5.00,0.00] 
Total: 30990 W: 4945 L: 4942 D: 21103 

No functional change. 

descriptionStockfish Latest Versions EmptyStockfish Vovakor MinGW 14/11/15

more_horiz
Stockfish Vovakor MinGW 14/11/15

Code:

http://www.mediafire.com/download/ozuz8xdlfafcbih/stockfish+141115+POPCNT_64.exe

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
King Asad compile very strong

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
thanks you all......anaconda advise many to become vip here in this forum.......they listen to me........coss they know anaconda very well as a player....thanks to all in this forum..........bless

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
!! latest version !!


Windows x64 for Haswell CPUs
Windows x64 for modern computers
Windows x64
Windows 32
Linux x64 for Haswell CPUs
Linux x64 for modern computers
Linux x64
Author: VoyagerOne 
Date: Sun Nov 15 20:40:09 2015 -0800 
Timestamp: 1447648809 

History Pruning: Don't prune the main killer move. 

Also increased pruned depth to 4. 

STC: 
LLR: 2.96 (-2.94,2.94) [0.00,5.00] 
Total: 23380 W: 4581 L: 4350 D: 14449 

LTC: 
LLR: 2.96 (-2.94,2.94) [0.00,5.00] 
Total: 28934 W: 4329 L: 4105 D: 20500 

Bench: 8369743 

Resolves #498  

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
!! latest version !!


Windows x64 for Haswell CPUs
Windows x64 for modern computers
Windows x64
Windows 32
Linux x64 for Haswell CPUs
Linux x64 for modern computers
Linux x64
Author: mstembera 
Date: Fri Nov 20 23:30:07 2015 -0800 
Timestamp: 1448091007 

Clean up RootMove less operator 

This is used by std::stable_sort() to sort moves from highest score to lowest score. 

1) The comment is incorrect since highest to lowest means descending. 
2) It's more natural to implement a less operator using another less operator rather than a greater operator. 

No functional change. 

Resolves #504 

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
!! latest version !!


Windows x64 for Haswell CPUs
Windows x64 for modern computers
Windows x64
Windows 32
Linux x64 for Haswell CPUs
Linux x64 for modern computers
Linux x64
Author: Joona Kiiski 
Date: Tue Nov 24 17:24:45 2015 -0800 
Timestamp: 1448414685 

Revert "Allow cross compilation of Windows binaries on a Linux system" 

This reverts commit 388630ae285b3f9f0c8ee4f30e754bde6688c57c. 

Confuses fishtest build system 

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
Windows x64 for Haswell CPUs
Windows x64 for modern computers
Windows x64
Windows 32
Linux x64 for Haswell CPUs
Linux x64 for modern computers
Linux x64
Author: Joona Kiiski 
Date: Tue Nov 24 17:24:45 2015 -0800 
Timestamp: 1448414685 

Revert "Allow cross compilation of Windows binaries on a Linux system" 

This reverts commit 388630ae285b3f9f0c8ee4f30e754bde6688c57c. 

Confuses fishtest build system

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
Windows x64 for Haswell CPUs
Windows x64 for modern computers
Windows x64
Windows 32
Linux x64 for Haswell CPUs
Linux x64 for modern computers
Linux x64
Author: Marco Costalba 
Date: Sat Dec 5 11:16:19 2015 +0100 
Timestamp: 1449310579 

Simplify outpost code 

Also inline defintions of SpaceMask and CenterBindMask. 

Verified from assembly that compiler computes the values 
at compile time, so it is also theoretical faster. 

While there factor out scale factor evaluation. 

No functional change. 

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
!! latest version !!


Windows x64 for Haswell CPUs
Windows x64 for modern computers
Windows x64
Windows 32
Linux x64 for Haswell CPUs
Linux x64 for modern computers
Linux x64
Author: absimaldata 
Date: Sat Dec 5 10:31:00 2015 +0000 
Timestamp: 1449311460 

New Tuned Weights 

More accurate evaluation weights 

Performed better at STC 

LLR: 1.32 (-2.94,2.94) [0.00,4.00] 
Total: 190043 W: 37433 L: 36675 D: 115935 

Passed LTC 

LLR: 2.95 (-2.94,2.94) [0.00,4.00] 
Total: 30157 W: 4540 L: 4303 D: 21314 

Bench: 9264977 

Resolves #515 

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
DeepFishMZ v5.2

Last src+LP Revert "Fix compiling of 32 bit binary on 64-bit Windows"

[size=13][You must be registered and logged in to see this link.]
[/size]




File Size: 582.37KB

[size=13][You must be registered and logged in to see this link.]
[/size]

[url=http://www.mediafire.com/download/7fosrhj7frlo2hj/DeepFishMZ+v5.2.7z][size=13][You must be registered and logged in to see this link.]
[/url][/size]

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
Stockfish_151225b

[You must be registered and logged in to see this link.]

descriptionStockfish Latest Versions EmptyRe: Stockfish Latest Versions

more_horiz
privacy_tip Permissions in this forum:
You cannot reply to topics in this forum
power_settings_newLogin to reply