Thanks Martin. Missed that.Martin Thoresen wrote:He has already done this:Graham Banks wrote: It would be very much appreciated if Vadim made a version available that didn't automatically allocate large pages usage.
http://www.spaadyshare.com/VYF5Y7GDBYF4 ... _no_LP.zip
Best,
Martin
Something new: Gull 1.2
Moderators: hgm, Rebel, chrisw
-
- Posts: 41428
- Joined: Sun Feb 26, 2006 10:52 am
- Location: Auckland, NZ
Re: Something new: Gull 1.2
gbanksnz at gmail.com
-
- Posts: 2041
- Joined: Wed Mar 08, 2006 8:30 pm
Re: Something new: Gull 1.2
Both exe(s) Gull 1_2 w32.exe and Gull 1_2 w32 tunable.exe have the (usual) options:
option name Hash type spin min 1 max 1024 default 64
option name Ponder type check default false
option name MultiPV type spin min 1 max 64 default 1
option name Clear Hash type button
Gull 1_2 w32 tunable.exe has the following complementary options:
option name Aspiration Search type check default true
option name 1/2 Ply Search type check default false
option name LMR type check default true
option name PV Hashing type check default false
option name Extended Info type check default true
option name Space Evaluation type check default false
option name Favor type combo default None var White var None var Black
option name +- Evaluation type button
option name = Evaluation type button
option name -+ Evaluation type button
Author ThinkingALot calls these last options "tunable compile" (see readme.txt)
Any idea what they mean (or link to that)?
option name Hash type spin min 1 max 1024 default 64
option name Ponder type check default false
option name MultiPV type spin min 1 max 64 default 1
option name Clear Hash type button
Gull 1_2 w32 tunable.exe has the following complementary options:
option name Aspiration Search type check default true
option name 1/2 Ply Search type check default false
option name LMR type check default true
option name PV Hashing type check default false
option name Extended Info type check default true
option name Space Evaluation type check default false
option name Favor type combo default None var White var None var Black
option name +- Evaluation type button
option name = Evaluation type button
option name -+ Evaluation type button
Author ThinkingALot calls these last options "tunable compile" (see readme.txt)
Any idea what they mean (or link to that)?
-
- Posts: 20943
- Joined: Mon Feb 27, 2006 7:30 pm
- Location: Birmingham, AL
Re: Something new: Gull 1.2
Why would you want it to run in a less than optimal (slower) way? large pages reduces TLB thrashing and can make a significant speed difference for many programs...Graham Banks wrote:Thanks Martin. Missed that.Martin Thoresen wrote:He has already done this:Graham Banks wrote: It would be very much appreciated if Vadim made a version available that didn't automatically allocate large pages usage.
http://www.spaadyshare.com/VYF5Y7GDBYF4 ... _no_LP.zip
Best,
Martin
-
- Posts: 2041
- Joined: Wed Mar 08, 2006 8:30 pm
Re: Something new: Gull 1.2
Well, for those who like games between 2 engines, not sure what happens if both engines grab for their own share of LP...bob wrote:Why would you want it to run in a less than optimal (slower) way?
-
- Posts: 1833
- Joined: Thu Jun 22, 2006 12:07 am
Re: Something new: Gull 1.2
Exactly, for analyzing LP is definitely a good idea.ernest wrote: Well, for those who like games between 2 engines, not sure what happens if both engines grab for their own share of LP...
But there are only problems when 2 engines meet in a game and both wants LP.
Best,
Martin
-
- Posts: 41428
- Joined: Sun Feb 26, 2006 10:52 am
- Location: Auckland, NZ
Re: Something new: Gull 1.2
My understanding was that your computer required a significant amount of memory in order to run large pages?bob wrote:Why would you want it to run in a less than optimal (slower) way? large pages reduces TLB thrashing and can make a significant speed difference for many programs...Graham Banks wrote:Thanks Martin. Missed that.Martin Thoresen wrote:He has already done this:Graham Banks wrote: It would be very much appreciated if Vadim made a version available that didn't automatically allocate large pages usage.
http://www.spaadyshare.com/VYF5Y7GDBYF4 ... _no_LP.zip
Best,
Martin
gbanksnz at gmail.com
-
- Posts: 2871
- Joined: Wed Mar 08, 2006 10:09 pm
- Location: Germany
- Full name: Werner Schüle
Re: Something new: Gull 1.2
Yes - I started 2 matches with Gull on my quad and here always one match stopped: The gui could´nt load Gull. Only one match ran ok.Martin Thoresen wrote:But there are only problems when 2 engines meet in a game and both wants LP.
Best,
Martin
So a am please to see a better solution now.
regards
Werenr
Werner
-
- Posts: 2010
- Joined: Sun May 25, 2008 11:12 pm
- Location: Whitchurch. Shropshire, UK.
- Full name: Harvey Williamson
Re: Something new: Gull 1.2
the one demanding large pages can also take resources from the one that does not even with ponder off.Werner wrote:Yes - I started 2 matches with Gull on my quad and here always one match stopped: The gui could´nt load Gull. Only one match ran ok.Martin Thoresen wrote:But there are only problems when 2 engines meet in a game and both wants LP.
Best,
Martin
So a am please to see a better solution now.
regards
Werenr
-
- Posts: 3245
- Joined: Thu Mar 09, 2006 9:10 am
Re: Something new: Gull 1.2
Even if only one engine uses LP, things can still get messy in game-play mode.Martin Thoresen wrote:Exactly, for analyzing LP is definitely a good idea.ernest wrote: Well, for those who like games between 2 engines, not sure what happens if both engines grab for their own share of LP...
But there are only problems when 2 engines meet in a game and both wants LP.
Best,
Martin
LP should always be under control of user.
Matthias.
My engine was quite strong till I added knowledge to it.
http://www.chess.hylogic.de
http://www.chess.hylogic.de