Scid vs. PC 4.14

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
CositasBuenas
Posts: 105
Joined: Tue Aug 03, 2010 5:36 pm

Re: Scid vs. PC 4.14

Post by CositasBuenas » Sun May 24, 2015 7:03 pm

Very stange this. You could also try ChessX or Xboard.

User avatar
reflectionofpower
Posts: 1456
Joined: Fri Mar 01, 2013 4:28 pm
Location: USA

Re: Scid vs. PC 4.14

Post by reflectionofpower » Sun May 24, 2015 7:13 pm

CositasBuenas wrote:Very stange this. You could also try ChessX or Xboard.
ALL Linux programs that are supposed to be a cinch are not with these UCI engines. I looked at Xboard, u have to compile the damn thing and I looked at it and it wadsn't just a simple ./comfigure
make deal either so that is out for sure

ChessX (tgz) thats another one thats out, "Compiling the source is simple. Execute:
qmake
and then
make"

NOT

Windows 8.1 here I come
"hodie mihi, cras tibi"

Lonnie

User avatar
Ponti
Posts: 454
Joined: Wed Mar 15, 2006 5:13 am
Location: Curitiba - PR - BRAZIL
Contact:

Re: Scid vs. PC 4.14

Post by Ponti » Mon May 25, 2015 5:21 am

CositasBuenas wrote:Sounds like you need an OS like Android where you simply tap the screen to get what you want. Everything else seems to lead to mental overload.
Nope. He should get iOS. :lol:
A. Ponti
AMD Ryzen 1800x, Windows 10.
FIDE current ratings: standard 1913, rapid 1931

User avatar
Ponti
Posts: 454
Joined: Wed Mar 15, 2006 5:13 am
Location: Curitiba - PR - BRAZIL
Contact:

Re: Scid vs. PC 4.14 - A Question about upgrade:

Post by Ponti » Wed May 27, 2015 4:09 am

I compiled it and installed it from source a time ago, currently running SCID vs PC 4.12 with Manjaro Ascella 0.8.12..

I want to upgrade to 4.14... do I have to compile it again to another new directory ?

I'm running Stockfish 6 linux without any problems... :-)
A. Ponti
AMD Ryzen 1800x, Windows 10.
FIDE current ratings: standard 1913, rapid 1931

User avatar
Ponti
Posts: 454
Joined: Wed Mar 15, 2006 5:13 am
Location: Curitiba - PR - BRAZIL
Contact:

Re: Scid vs. PC 4.14

Post by Ponti » Wed May 27, 2015 4:48 am

There's obviously a difference when you're compiling from source vs when you are installing software in Linux with a package (and that resembles better the way Windows users install softwares).

It could be easier if the author of SCID vs PC could build some packages for the main Linux distros.

Many newbie Linux users do not understand that to compile sofware from source there are some prerequisites, and it is a pain to read all the manuals, ask in the foruns, search for the answers... I know, because I've also done that for many years (I installed Red Hat from diskettes, for example, just doing what every weird guy told me- RTFM !!).

The best advantage I see in Linux is: I can forget my PC running analysing or doing anything I want and it won't shut down unless something very serious happens. I tried to do that with Windows, but it can' t. Did not try with Mac, maybe it can, because it is actually FreeBSD...

Unfortunately, chess software for Linux and Mac are not at the same level at their counterpart for Windows. Hopefully it will be a matter of time... IMHO
A. Ponti
AMD Ryzen 1800x, Windows 10.
FIDE current ratings: standard 1913, rapid 1931

Ferdy
Posts: 4079
Joined: Sun Aug 10, 2008 1:15 pm
Location: Philippines

Re: Another apparent bug

Post by Ferdy » Wed Jun 10, 2015 10:21 am

I make a tool to analyze a game in a pgn file automatically using an engine. Make it output its first 2 best pv. This is still under development, just try to make an attempt to create a smart analyst.

Here are some notes that perhaps may be of use to your development.
1. In the analysis I output the pv once the eval of first pv of the engine is better by 15cp or more than the eval of the game move and of course the move in the first pv should not be the same as that of the game move.

2. At move 8 there is only 1 alternative pv shown though I set the engine to do multipv 2. This is because when the move of the 2nd pv has a move similar to the game move then do not include it in the output.

3. At move 9, 2 alternative pv were given, both are better than the game move eval (my tool is side pov). Looks fine so far, but I'd like to add a comment before the pv starts, something like this
(Better is 9.Nf4 ... +0.28/14)
(Also playable is 9.Bd2 ... +0.19/14)
The word better is, and playable, is dependent on the score difference of the move evals. So if it may happen that the 2nd pv is worst, say -0.15/14, perhpas the output would be.
(A mistake is 9.b3 ... -0.15/14)
This would warn the user even without looking at the eval score. The idea is that the user is informed that there is also dangerous move in this position, by playing thru that line the user will get an idea of what will make his position vulnerable. If indeed the 2nd pv will bring the score to negative, and the first pv is positive then I plan to increase the default analysis time of the analyzing engine, say 3x especially when the best pv is not an easy move like recapture. I thought of recording the complexity value of every position so that whenever the engine changes its best pv move more than once, it would be a candidate for increasing the analysis time.

4. The game analyzed here is my game and playing white, at move 13. h3 this is the move where I spent a lot of time (TC is 5min+15sec inc)
[d]r1bq1rk1/pp3ppp/2pb1n2/3p4/3P4/P1NBP3/1PQ2PPP/R1B2RK1 w - - 2 13
Black has some sort of a threat, targeting my h2 sq, Ng4 followed by Qh4, or Qc7, or Bc7 followed by Qd6, if I play g3 then black may play h5 then h4, or even Bh3. So the plan I adopt is to hold the Ng4 by h3, and if my dark diagonal squares are attacked, I will just play f4, my e3 is weak in this case but it could not be exploited easily since I have the dak bishop. My other plan of playing f3 followed by e4 to gain central control is now abandoned.
In other words this is the position where I like the engine to spend more time analyzing the position. But look at the smart analyst :) it does not give me any analysis at all (bad or good), I will accept anything the engine would throw here even Lyudmil :).
The problem here is in no. 1, I set the engine to give me the pvs if the game move is worst by 15cp or more. So how to make an engine and tool smart on this. One solution is to mark 13.h3 with something say 13.h3!? the point is wherever the move has this symbol then make the engine spent more time on this position and throw some pvs here, say default multipv 2 * 3 = multipv 6. After the analysis remove the move mark !? in move 13.h3!?.

5. Stockfish will show some eval parameters values like mobility, and others when eval command is sent, utilize this feature and include a comment showing the eval parameters that have high values for a given side. Say if my king safety is low compared to black by a given margin, add a comment, {Your king is in danger}, I don't know if there is an equivalent NAGS on this.
Image

Here is the pgn. This also looks fine in Fritz 13, as I use some of its pgn output format into my tool.

[pgn]
[Event "Training"]
[Site "Home"]
[Date "2015.06.07"]
[Round "1"]
[White "Player"]
[Black "CENTURY 3.0"]
[Result "0-1"]

1. d4 Nf6 2. c4 e6 3. Nc3 Bb4 4. e3 O-O 5. Ne2 d5 6. a3
Be7 7. cxd5 Nxd5
8. Nxd5 $0 {+0.05/14} (8. e4 Nxc3 9. Nxc3 Nc6 10. Be3 f5 11. exf5 $14 {+0.24/14})exd5
9. Qc2 $0 {-0.08/13} (9. Nf4 g5 10. Nh5 Bf5 11. Be2 b6 12. Bd2 $14 {+0.28/14})
(9. Bd2 c6 10. Nf4 Bf5 11. Be2 Nd7 12. O-O $14 {+0.19/14})
Nd7 $0 {-0.13/15} (9... g6 10. Nf4 Bg5 11. Bd3 Bxf4 12. exf4 Nc6 $10 {+0.08/13})
(9... Re8 10. Ng3 b6 11. Bd3 g6 12. b4 Bb7 $10 {-0.02/13}) 10. Nc3 Nf6 11. Bd3 c6 12. O-O
Bd6 13. h3 Bc7 14. f4 Re8 15. Bd2 Bd7 16. Rae1 a6 17. Rf2
g6 $0 {-0.00/13} (17... h6 18. Rc1 b5 19. Ne2 a5 20. Ng3 Bb6 $15 {+0.22/13})
(17... a5 18. Rc1 h6 19. Rff1 Bd6 20. e4 dxe4 $15 {+0.16/13}) 18. Ref1
Qe7 19. g4
b6 $0 {-0.00/12} (19... h6 20. Rc1 Rad8 21. Qb3 Rb8 22. Qc2 Kh8 $15 {+0.20/13})
(19... Qd6 20. Rg2 Kh8 21. f5 a5 22. Qb3 Rab8 $15 {+0.13/13})
20. f5 $0 {-0.37/13} (20. Rg2 Qd6 21. Na4 Qe7 22. Nc3 $10 {+0.00/12})
(20. Rc1 Qd6 21. b3 Qe7 22. a4 a5 23. g5 $10 {-0.08/12}) g5 0-1
[/pgn]

User avatar
Ozymandias
Posts: 1103
Joined: Sun Oct 25, 2009 12:30 am

Re: Scid vs. PC 4.14

Post by Ozymandias » Mon Sep 14, 2015 7:49 am

So, I've been itching for some time, to switch to Yosemite, but I'd like to get rid of BootCamp at the same time. What does this have to do with Scid? Well, I initially installed Windows, because of my lack of knowledge about OSX apps. Over the years I've been finding replacements for all the programs I use; all but one. Even though ScidvsMac looks much better, and has the same functionality that ScidvsPC, I find myself using the latter, every time I'm going to do some tree searches, simply because it's 10 times faster. I find it odd, this being a rMBP and all. Is this a known problem or is it just me?

stevenaaus
Posts: 602
Joined: Wed Oct 13, 2010 7:44 am
Location: Australia
Contact:

Re: Scid vs. PC 4.14

Post by stevenaaus » Mon Sep 14, 2015 11:43 am

Wow - you're right. The problem is progressbar updates are 10x slower on OS X due to buggy TCL . Searches are probably the same. To test, change the 5000 const in sc_tree_search to 1000,000 and recompile tkscid.

I'll have to do something... just not sure what, laugh. Maybe remove the tree progressbar for OS X. ... but ???

I'll look at it on the weekend.
---------
Ferdinand - Analysis annotation is a very complicated feature. Very hard to make small changes. Big changes like you suggest are not possible.

User avatar
Ozymandias
Posts: 1103
Joined: Sun Oct 25, 2009 12:30 am

Re: Scid vs. PC 4.14

Post by Ozymandias » Mon Sep 14, 2015 1:21 pm

stevenaaus wrote:Maybe remove the tree progress bar for OS X.
That's an option, but there should be some indication about the search ending. When you enable "fast and slow mode", the stats are updated when "fast" search concludes, but if they don't differ from the final ones (in a noticeable way), there's no other way to know, that the job's done. You could use an italic font, by default, for the provisional stats the fast search produces, and reinstate the active font when the slow search finishes.

stevenaaus
Posts: 602
Joined: Wed Oct 13, 2010 7:44 am
Location: Australia
Contact:

Re: Scid vs. PC 4.14

Post by stevenaaus » Tue Sep 15, 2015 9:21 am

That's an option, but there should be some indication about the search ending. When you enable "fast and slow mode", the stats are updated when "fast" search concludes, but if they don't differ from the final ones (in a noticeable way), there's no other way to know, that the job's done. You could use an italic font, by default, for the provisional stats the fast search produces, and reinstate the active font when the slow search finishes.
Ok. It's very likely the solution will be an option to hide the tree progress bar.
It still gives some feedback (stop button), speeds up OS X, and gives more screen room. And it's a very simple fix.

You can do it yourself by editing ScidvsMac.app/Contents/MacOS/scid
and removing this line (around line number 87809)

Code: Select all

--- tcl/windows/tree.tcl        (revision 2342)
+++ tcl/windows/tree.tcl        (working copy)
@@ -208,7 +208,6 @@
   # pack $w.status -side bottom -fill x
 
   pack [frame $w.buttons -relief sunken] -side bottom -fill x -pady 5
-  pack $w.progress -side bottom
   pack $w.f -side top -expand 1 -fill both
 
   button $w.buttons.best -image b_list -command "::tree::toggleBest $baseNumber"
How the progress bar affects search speed is less obvious.
My Linux and OS X hardware is different.
Using the same hardware, could you please do a couple of material and board searches in windows/OS X, and tell me how they compare.
I'm not sure the impact is so bad.

BTW - i really don't have a great handle on the fast/slow/fast+slow search code. It could probably do with some overhaul since the tree search is now interuptible. Any comments ?

Post Reply