Latest Updates:
Very Hot Topic (More than 25 Replies) The best analysis program? (Read 211580 times)
an ordinary chessplayer
God Member
*****
Offline


I used to be not bad.

Posts: 1637
Location: Columbus, OH (USA)
Joined: 01/02/15
Re: The best analysis program?
Reply #189 - 08/18/23 at 03:44:00
Post Tools
Quote:
Glenn Flear wrote:
Every opening where White has a space advantage involving a pawn on d5 (whether passed or not) are over-hyped by the engines...

That sounds right. I said something similar on chess.com (yetanotheraoc is me):

Quote:
Engines love a space advantage like the one after d4-d5, but that's just prejudice. An engine isn't a positional genius like Karpov. The point is, nowadays people analyze with an engine all the time, so you can expect to see d4-d5 played in more positions, even when it's not necessarily the best procedure.
https://www.chess.com/forum/view/chess-openings/tension-in-the-centre-in-the-clo...

There is a corollary here: you can adopt a black repertoire based on one of these space-disadvantage structures and do well with them when your opponents over-estimate their chances. Preparing such an opening just requires some experience and judgment to know that position X at +1.05 is fine for black while position Y at +0.90 position is unacceptable.

I wouldn't be able to hold any of these positions against an engine, but that doesn't mean the evaluation is correct!
  
Back to top
 
IP Logged
 
FreeRepublic
God Member
*****
Offline


I Love ChessPublishing!

Posts: 662
Location: Georgia
Joined: 06/08/17
Gender: Male
Re: The best analysis program?
Reply #188 - 08/17/23 at 22:24:48
Post Tools
Glenn Flear in his August 2023 Daring Defense update:

"Time for a pronouncement: I have a theory! Every opening where White has a space advantage involving a pawn on d5 (whether passed or not) are over-hyped by the engines i.e. the silicon monster gives White as having a healthy plus when humans consider Black to be perfectly fine. Has anybody else noticed this?

Check out some main line King's Indians, Chigorin Ruy Lopezs, Benkos, Benonis and various Grünfelds involving a pawn on d5, and note what Stockfish 16 thinks. See what I mean!?"

It's hard to ignore computer evaluations. They evaluate so many positions, so quickly. Yet if there is a flaw in their evaluations, then you might be misled. One might be unduly pessimistic about playing the lines mentioned above with the Black pieces.
  
Back to top
 
IP Logged
 
Pawnpusher
Senior Member
****
Offline


I Love ChessPublishing!

Posts: 273
Joined: 01/04/18
Gender: Male
Re: The best analysis program?
Reply #187 - 06/13/19 at 10:54:55
Post Tools
Has anyone tried to use chess position trainer and LcO?
  
Back to top
 
IP Logged
 
tony37
God Member
*****
Offline


I Love ChessPublishing!

Posts: 619
Joined: 10/16/10
Re: The best analysis program?
Reply #186 - 06/12/19 at 08:52:17
Post Tools
after installing those drivers I couldn't boot into my system and had to revert to the older drivers (I probably should upgrade to Ubuntu 19.04 (update: this works fine with standard drivers; what also worked was installing the 5.1 kernel on 18.04, even without newer drivers))

there seems to be a bug in this thread where the latest post isn't always displayed
« Last Edit: 06/12/19 at 22:54:29 by tony37 »  
Back to top
 
IP Logged
 
tony37
God Member
*****
Offline


I Love ChessPublishing!

Posts: 619
Joined: 10/16/10
Re: The best analysis program?
Reply #185 - 06/11/19 at 19:25:24
Post Tools
Has anyone tried LeelaFish yet? I'd see it as Leela with Stockfish tactical blunder check enabled. Supposedly it's 30 elo higher than Leela alone. See https://github.com/killerducky/lc0/wiki/LeelaFish for more info, don't forget to  link 'AuxEngineFile' to Stockfish, otherwise it's just lc0 running.

Probably not important for most, but after installing the latest Linux drivers for my AMD GPU (from ppa:oibaf/graphics-drivers for those interested), my mouse isn't hanging anymore when I put lc0 on.
  
Back to top
 
IP Logged
 
Pawnpusher
Senior Member
****
Offline


I Love ChessPublishing!

Posts: 273
Joined: 01/04/18
Gender: Male
Re: The best analysis program?
Reply #184 - 06/06/19 at 22:09:18
Post Tools
Ahh but the quest to optimize!!
  
Back to top
 
IP Logged
 
JFugre
Junior Member
**
Offline


I Love ChessPublishing!

Posts: 89
Joined: 01/22/19
Re: The best analysis program?
Reply #183 - 06/06/19 at 07:55:38
Post Tools
Pawnpusher wrote on 06/05/19 at 15:52:11:
Thanks guys, I get the eval better now. Does anyone know how to find the best settings for a GPU (nvidia 1060)?


Defaults are fine.
  
Back to top
 
IP Logged
 
Pawnpusher
Senior Member
****
Offline


I Love ChessPublishing!

Posts: 273
Joined: 01/04/18
Gender: Male
Re: The best analysis program?
Reply #182 - 06/05/19 at 15:52:11
Post Tools
Thanks guys, I get the eval better now. Does anyone know how to find the best settings for a GPU (nvidia 1060)?
  
Back to top
 
IP Logged
 
JFugre
Junior Member
**
Offline


I Love ChessPublishing!

Posts: 89
Joined: 01/22/19
Re: The best analysis program?
Reply #181 - 06/05/19 at 08:53:48
Post Tools
Pawnpusher wrote on 06/04/19 at 23:40:57:
I do use Cuda, the evaluation puzzles me. I am used to seeing scores far lower than LcO gives for positions. Have you noticed this?


The latest lc0 version has a re-scaled formula to convert winning percentage to centipawns. The scores are now much less extreme, and were calibrated to the average of Stockfish/Komodo scores in TCEC15, or something. So it's more comparable to a classic engine now.

I believe the one playing in TCEC had the old formula. You can change it in the settings.
  
Back to top
 
IP Logged
 
brabo
God Member
*****
Offline


Welcome chessfriend

Posts: 1064
Joined: 02/02/07
Re: The best analysis program?
Reply #180 - 06/05/19 at 00:18:33
Post Tools
Pawnpusher wrote on 06/04/19 at 23:40:57:
I do use Cuda, the evaluation puzzles me. I am used to seeing scores far lower than LcO gives for positions. Have you noticed this?

Yes it is explained at https://github.com/LeelaChessZero/lc0/wiki/Technical-Explanation-of-Leela-Chess-...
+3,36 would mean 90% probabily of a win for white.

For e.g. Houdini only +1,5 already means 90% probability of a win for white see http://www.cruxis.com/chess/houdini.htm

So you should make a conversion from 1 scale to another.
  
Back to top
WWW  
IP Logged
 
brabo
God Member
*****
Offline


Welcome chessfriend

Posts: 1064
Joined: 02/02/07
Re: The best analysis program?
Reply #179 - 06/05/19 at 00:18:15
Post Tools
tony37 wrote on 06/04/19 at 23:26:56:
brabo wrote on 06/04/19 at 11:15:32:
tony37 wrote on 06/04/19 at 09:30:49:
ordo elo (or self-play elo) can be very misleading, as I wrote a few posts before
can I ask what AMD card you're using? probably way better than mine


My AMD FX(tm)-6300 6 core processor 3,5Ghz is configured with a Nvidia Geforce GTX 960. That is average nowadays.

but since you have an Nvidia card, you probably use the (fast) Cuda version of Lc0 after all
I think JFugre was talking about using the OpenCL version on an AMD graphics card like Radeon

No I am using the OpenCL version. When I try the Cuda version then the lc0 exe just closes without having done any computation. I thought it probably has to do with that I first have to install https://developer.nvidia.com/cuda-zone but then I get the message that I should also install the video studio to let it work properly. I also did that but it kept warning that the video studio is not ok so then I gave up before making some changes which I can't restore anymore afterwards.
  
Back to top
WWW  
IP Logged
 
Pawnpusher
Senior Member
****
Offline


I Love ChessPublishing!

Posts: 273
Joined: 01/04/18
Gender: Male
Re: The best analysis program?
Reply #178 - 06/04/19 at 23:40:57
Post Tools
I do use Cuda, the evaluation puzzles me. I am used to seeing scores far lower than LcO gives for positions. Have you noticed this?
  
Back to top
 
IP Logged
 
tony37
God Member
*****
Offline


I Love ChessPublishing!

Posts: 619
Joined: 10/16/10
Re: The best analysis program?
Reply #177 - 06/04/19 at 23:26:56
Post Tools
brabo wrote on 06/04/19 at 11:15:32:
tony37 wrote on 06/04/19 at 09:30:49:
ordo elo (or self-play elo) can be very misleading, as I wrote a few posts before
can I ask what AMD card you're using? probably way better than mine


My AMD FX(tm)-6300 6 core processor 3,5Ghz is configured with a Nvidia Geforce GTX 960. That is average nowadays.

but since you have an Nvidia card, you probably use the (fast) Cuda version of Lc0 after all
I think JFugre was talking about using the OpenCL version on an AMD graphics card like Radeon
  
Back to top
 
IP Logged
 
brabo
God Member
*****
Offline


Welcome chessfriend

Posts: 1064
Joined: 02/02/07
Re: The best analysis program?
Reply #176 - 06/04/19 at 11:15:32
Post Tools
tony37 wrote on 06/04/19 at 09:30:49:
ordo elo (or self-play elo) can be very misleading, as I wrote a few posts before
can I ask what AMD card you're using? probably way better than mine

Yes I also don't trust the ordo elo much. It is an indication but I consider the match much more valuable.

My AMD FX(tm)-6300 6 core processor 3,5Ghz is configured with a Nvidia Geforce GTX 960. That is average nowadays.
  
Back to top
WWW  
IP Logged
 
Pawnpusher
Senior Member
****
Offline


I Love ChessPublishing!

Posts: 273
Joined: 01/04/18
Gender: Male
Re: The best analysis program?
Reply #175 - 06/04/19 at 11:09:02
Post Tools
Yes I think Tony37 is on point according to what I have read. If I understand it is not only the ELO of a network but the number of games (in other words the number of positions) in the network. I have an Nvidia Geforce 1060 card, and am looking to find the best settings as I read that LcO is sort of difficult to get optimized. Having said that I am impressed at the play of LcO.
  
Back to top
 
IP Logged
 
tony37
God Member
*****
Offline


I Love ChessPublishing!

Posts: 619
Joined: 10/16/10
Re: The best analysis program?
Reply #174 - 06/04/19 at 09:30:49
Post Tools
brabo wrote on 06/04/19 at 09:05:37:
JFugre wrote on 06/04/19 at 06:18:45:
OpenCL still gets reasonable performance on AMD cards.


Correct. I ran last week a 100 game match 15 minutes + 10 seconds between the commercial Komodo 11 and Lc0 on my AMD (2,5 years old) and Komodo only won with 56 - 44. That is about 50 elo difference. I expect with a longer timecontrol that Lc0 would maybe even win such match. I was using the opencl and a network = 3214 ordo elo. Today there is already a network = 3253 ordo elo.

ordo elo (or self-play elo) can be very misleading, as I wrote a few posts before
can I ask what AMD card you're using? probably way better than mine
  
Back to top
 
IP Logged
 
brabo
God Member
*****
Offline


Welcome chessfriend

Posts: 1064
Joined: 02/02/07
Re: The best analysis program?
Reply #173 - 06/04/19 at 09:05:37
Post Tools
JFugre wrote on 06/04/19 at 06:18:45:
OpenCL still gets reasonable performance on AMD cards.


Correct. I ran last week a 100 game match 15 minutes + 10 seconds between the commercial Komodo 11 and Lc0 on my AMD (2,5 years old) and Komodo only won with 56 - 44. That is about 50 elo difference. I expect with a longer timecontrol that Lc0 would maybe even win such match. I was using the opencl and a network = 3214 ordo elo. Today there is already a network = 3253 ordo elo.

I will most likely write an article on my blog about it as I guess it is interesting to see what the performance of Lc0 today is on more or less commodity hardware.
  
Back to top
WWW  
IP Logged
 
JFugre
Junior Member
**
Offline


I Love ChessPublishing!

Posts: 89
Joined: 01/22/19
Re: The best analysis program?
Reply #172 - 06/04/19 at 06:18:45
Post Tools
TopNotch wrote on 06/04/19 at 01:58:39:
I've tried this and essentially LC0 is utterly useless running on windows unless you have a Nvidia Graphics Card. I honestly don't understand why they even offer blas, opencl, its a joke.


OpenCL still gets reasonable performance on AMD cards. OpenCL on a CPU or an Intel iGPU is too slow. They just lack the horsepower.

It's mainly when comparing with the RTX cards that the gap widens up to ludicrous levels (i.e. an RTX 2060 is 15 times faster than a high end AMD card, whereas with a GTX 1060 it's only ~2.8).
  
Back to top
 
IP Logged
 
an ordinary chessplayer
God Member
*****
Offline


I used to be not bad.

Posts: 1637
Location: Columbus, OH (USA)
Joined: 01/02/15
Re: The best analysis program?
Reply #171 - 06/04/19 at 03:22:18
Post Tools
TopNotch wrote on 06/04/19 at 01:58:39:
I honestly don't understand why they even offer blas, opencl, its a joke.
Not a joke. It's a software experiment, they are trying not to limit the target hardware. Alphazero didn't run on stock x86 hardware, and lc0 need not, either. Some systems don't have graphics at all, but sufficiently powerful CPUs (or TPUs in Alphazero's case) can still make it interesting. Who knows what kind of crazy homegrown system someone might want to try it on?
  
Back to top
 
IP Logged
 
TopNotch
God Member
*****
Offline


I only look 1 move ahead,
but its always the best

Posts: 2211
Joined: 01/04/03
Gender: Male
Re: The best analysis program?
Reply #170 - 06/04/19 at 01:58:39
Post Tools
I've tried this and essentially LC0 is utterly useless running on windows unless you have a Nvidia Graphics Card. I honestly don't understand why they even offer blas, opencl, its a joke.
  

The man who tries to do something and fails is infinitely better than he who tries to do nothing and succeeds - Lloyd Jones Smiley
Back to top
 
IP Logged
 
Pawnpusher
Senior Member
****
Offline


I Love ChessPublishing!

Posts: 273
Joined: 01/04/18
Gender: Male
Re: The best analysis program?
Reply #169 - 06/03/19 at 14:46:16
Post Tools
Yes, I think for cuda to work you have to have some sort of Nvidia GPU, other GPUs use open cl, and if no graphics card the blas. I think that is how it works.
  
Back to top
 
IP Logged
 
tony37
God Member
*****
Offline


I Love ChessPublishing!

Posts: 619
Joined: 10/16/10
Re: The best analysis program?
Reply #168 - 06/03/19 at 08:09:44
Post Tools
Leon_Trotsky wrote on 06/02/19 at 22:37:00:
I followed the instructions, but still analysis window is totally blank  Shocked

for Windows there are three versions (blas, opencl and cuda), did you try all three?
the blas and opencl versions are likely to be very slow, so don't expect too much
  
Back to top
 
IP Logged
 
Pawnpusher
Senior Member
****
Offline


I Love ChessPublishing!

Posts: 273
Joined: 01/04/18
Gender: Male
Re: The best analysis program?
Reply #167 - 06/03/19 at 00:41:47
Post Tools
I just used the directions from the Github LcO site and it did take a couple of tries, I also looked at the Youtube site quoted above, and got it going. I looked on a computer chess forum. It was originally called Talkchess because I am trying to find the best settings for my GPU. It was mostly interesting and somewhat useful. Hope this helps.
  
Back to top
 
IP Logged
 
Leon_Trotsky
Senior Member
****
Offline


Кто был никем — тот станет
всем!

Posts: 499
Location: Barcelona, CAT
Joined: 08/11/17
Gender: Male
Re: The best analysis program?
Reply #166 - 06/02/19 at 22:37:00
Post Tools
I followed the instructions, but still analysis window is totally blank  Shocked

  
Back to top
 
IP Logged
 
proustiskeen
YaBB Moderator
*****
Offline


Hello from Omaha!

Posts: 678
Joined: 08/11/08
Re: The best analysis program?
Reply #165 - 06/02/19 at 18:24:53
Post Tools
@Leon_Trotsky - this may be of assistance.

https://www.youtube.com/watch?v=T2SB4Z-o5ZQ
  
Back to top
WWW  
IP Logged
 
tony37
God Member
*****
Offline


I Love ChessPublishing!

Posts: 619
Joined: 10/16/10
Re: The best analysis program?
Reply #164 - 06/02/19 at 16:04:57
Post Tools
update: I'm really not sure this weights file with 4305 elo is actually better than more recent ones about 3300
the first one gives some really weird evaluation in the following position, it initially thinks white is better there (the more recent file doesn't)

* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
*

update2: Now I finally got it: self-play elo doesn't seem to be important, you have to look at https://docs.google.com/spreadsheets/d/1XSJiCcQpCLv0fNwrUn7jXjdkZFU63YFEWpdXv6dS... (I seem to remember this site was down when I first looked there)
So network number 42000 is strongest available for now
  
Back to top
 
IP Logged
 
tony37
God Member
*****
Offline


I Love ChessPublishing!

Posts: 619
Joined: 10/16/10
Re: The best analysis program?
Reply #163 - 06/02/19 at 10:29:37
Post Tools
Pawnpusher wrote on 06/01/19 at 18:31:26:
I am using Lc0, I'm sort of a gamer so I have an NVidia gpu. I understand that the better the machine the better Leela works and mine is by no means state of the art. But it is the way to go currently, it does awesome analysis. I think the weights file (?) I downloaded is about 3300 elo? It is a very positional engine, and supposedly misses some tactics, but I haven't noticed.

I finally found out some older weights files have much highter elo than recent ones, the highest elo I found is 4305 (date: 2018-08-22 03:52:53 at http://lczero.org/networks). These elos probably shouldn't be compared to other rankings. No idea why recent files have much lower elo...

Leon_Trotsky wrote on 06/02/19 at 05:38:25:
I am having a problem setting Leela on ChessBase 15. How did you make it work there to analyse  ż

does it work in another program like Arena?
  
Back to top
 
IP Logged
 
Leon_Trotsky
Senior Member
****
Offline


Кто был никем — тот станет
всем!

Posts: 499
Location: Barcelona, CAT
Joined: 08/11/17
Gender: Male
Re: The best analysis program?
Reply #162 - 06/02/19 at 05:38:25
Post Tools
I am having a problem setting Leela on ChessBase 15. How did you make it work there to analyse  ż
  
Back to top
 
IP Logged
 
Pawnpusher
Senior Member
****
Offline


I Love ChessPublishing!

Posts: 273
Joined: 01/04/18
Gender: Male
Re: The best analysis program?
Reply #161 - 06/01/19 at 18:31:26
Post Tools
I am using Lc0, I'm sort of a gamer so I have an NVidia gpu. I understand that the better the machine the better Leela works and mine is by no means state of the art. But it is the way to go currently, it does awesome analysis. I think the weights file (?) I downloaded is about 3300 elo? It is a very positional engine, and supposedly misses some tactics, but I haven't noticed.
  
Back to top
 
IP Logged
 
tony37
God Member
*****
Offline


I Love ChessPublishing!

Posts: 619
Joined: 10/16/10
Re: The best analysis program?
Reply #160 - 05/19/19 at 11:01:01
Post Tools
Leon_Trotsky wrote on 05/18/19 at 09:07:42:
I tried download Leela in https://github.com/LeelaChessZero/lc0/releases but engine analysis not showing in ChessBase 15 when loaded.

The small window where engine evaluation should show is totally blank on my computer.

did you follow the instructions at https://github.com/LeelaChessZero/lc0/wiki/Getting-Started ?
first you need a 'weights'-file from http://lczero.org/networks, presumably one with high elo, and then you need to open the lc0.exe file to initialise the configuration (this takes a while)
also important is to have the right drivers for the GPU (for Cuda or for Opencl), this took me a whole day to get this right at my new Ubuntu laptop, on an old Windows 7 laptop the lc0 program just crashed after a while
now that it's finally working I have to say it's pretty slow and I barely can move my mouse while it's analysing, I suppose you need a GPU with Cuda support (I have to use Opencl) for it to make any sense using Leela for analysis

edit: when I use 'blas' as backend instead of 'opencl' (so using CPU iso GPU), I can actually use my computer while analysing but the engine is of course even slower then
for info: I have AMD A6-9220 radeon r4 as processor with AMD Stoney as graphics card, so new but not very strong
« Last Edit: 05/19/19 at 12:40:56 by tony37 »  
Back to top
 
IP Logged
 
BeeCaves
Junior Member
**
Offline


I Love ChessPublishing!

Posts: 66
Joined: 03/09/18
Re: The best analysis program?
Reply #159 - 05/18/19 at 17:38:07
Post Tools
Has anyone who installed Leela have any feedback on the evaluation function? 

Despite computers being so much stronger than us, it seems like it can be tough some time to use them to assess if a decision was "practical" in a game you made or if you should go for a line where you are defending ... it might be a draw but if one side needs to be much more precise from a human perspective, this can be a problem.

I saw some comment that AlphaZero plays at IM strength with "search turned off" and so I wonder if it is "more human" and easier to learn from.

I don't know how similar Leela plays to Alpha?

  
Back to top
 
IP Logged
 
an ordinary chessplayer
God Member
*****
Offline


I used to be not bad.

Posts: 1637
Location: Columbus, OH (USA)
Joined: 01/02/15
Re: The best analysis program?
Reply #158 - 05/18/19 at 14:28:10
Post Tools
http://blog.lczero.org/2018/09/guide-setting-up-leela-on-chess-gui.html

Quote:
This means that in order to run Leela in full strength you will need a GPU preferably one that supports CUDA.
Running it on GPU with non CUDA Lc0 binaries means you will get many times worse performance.
Even worse running it on a CPU like normal traditional engines run, means you will get many many times worse performance.
Neural net weights in order to run fast enough they need a GPU. On CPUs they are slow as turtles.

My laptop GPU is non CUDA, so I would have to buy new hardware, otherwise Stockfish is stronger. I'm too frugal for that, but also, I honestly don't see why I should care. In most cases I just use whatever engine is available in the GUI of the moment. That might be Shredder, Stockfish (10 or 9 or even 8), Crafty, Fritz.

My old car already goes faster than the speed limit, I don't care that a Tesla can go from 0 to 100 km/h in 2.3 seconds. If you point out that I would care in a race, my answer is that in the race we won't be using the car...
  
Back to top
 
IP Logged
 
Leon_Trotsky
Senior Member
****
Offline


Кто был никем — тот станет
всем!

Posts: 499
Location: Barcelona, CAT
Joined: 08/11/17
Gender: Male
Re: The best analysis program?
Reply #157 - 05/18/19 at 09:07:42
Post Tools
I tried download Leela in https://github.com/LeelaChessZero/lc0/releases but engine analysis not showing in ChessBase 15 when loaded.

The small window where engine evaluation should show is totally blank on my computer.
  
Back to top
 
IP Logged
 
brabo
God Member
*****
Offline


Welcome chessfriend

Posts: 1064
Joined: 02/02/07
Re: The best analysis program?
Reply #156 - 05/18/19 at 08:50:45
Post Tools
The superfinal of TCEC season 15 is ongoing and it looks like Leela has surpassed Stockfish. At this moment the score is 23-18: https://tcec.chessdom.com/

It is time to start using Leela instead of Stockfish as main-engine. I hear some people having troubles to let Leela work properly in Chessbase. I didn't try it out yet so does anybody has experience with it?
  
Back to top
WWW  
IP Logged
 
IsaVulpes
Senior Member
****
Offline


No.

Posts: 345
Joined: 12/09/07
Re: The best analysis program?
Reply #155 - 05/17/19 at 10:17:37
Post Tools
IsaVulpes wrote on 11/30/18 at 11:50:55:
GMTonyKosten wrote on 11/30/18 at 08:12:32:
Now that Stockfish 10 has been released, maybe someone can update this thread - I assume SF 10 is clearly strongest?

It should be, but running Lc0 at the same time is probably still going to be worth it in some positions

SF10 most likely isn't the strongest engine anymore now Wink

Time to get an RTX 2060.. What a time to be alive, buying strong GPUs to play.. Chess. Cheesy
  
Back to top
 
IP Logged
 
IsaVulpes
Senior Member
****
Offline


No.

Posts: 345
Joined: 12/09/07
Re: The best analysis program?
Reply #154 - 11/30/18 at 11:50:55
Post Tools
GMTonyKosten wrote on 11/30/18 at 08:12:32:
Now that Stockfish 10 has been released, maybe someone can update this thread - I assume SF 10 is clearly strongest?

It should be, but running Lc0 at the same time is probably still going to be worth it in some positions
  
Back to top
 
IP Logged
 
GMTonyKosten
YaBB Administrator
*****
Offline


Mr Dynamic?

Posts: 3149
Location: Clermont-Ferrand
Joined: 12/19/02
Gender: Male
Re: The best analysis program?
Reply #153 - 11/30/18 at 08:12:32
Post Tools
Now that Stockfish 10 has been released, maybe someone can update this thread - I assume SF 10 is clearly strongest?
  
Back to top
IP Logged
 
tdv
YaBB Newbies
*
Offline


Let's play chess!

Posts: 18
Joined: 03/27/18
Re: The best analysis program?
Reply #152 - 03/30/18 at 22:05:01
Post Tools
Quite agree that one do much better with only 1 good engine than a few years ago... even if I still think that "difference" always brings something in making an opinion (like in medicine Smiley)  If I had 2 computers available, I'd run 2 different engines at the same time anyway.
  
Back to top
IP Logged
 
brabo
God Member
*****
Offline


Welcome chessfriend

Posts: 1064
Joined: 02/02/07
Re: The best analysis program?
Reply #151 - 02/28/18 at 17:11:35
Post Tools
How much sense does it still have to use more than 1 engine today? I gave a short status of this in my just published article: https://chess-brabo.blogspot.be/2018/02/to-analyze-using-computer-part-3.html
  
Back to top
WWW  
IP Logged
 
TonyRo
God Member
*****
Offline


I'm gonna crack your skull!

Posts: 1826
Location: Cleveland, OH
Joined: 11/26/07
Gender: Male
Re: The best analysis program?
Reply #150 - 10/30/17 at 17:04:54
Post Tools
Me too - I make no claims about the accuracy!  Cool
  
Back to top
 
IP Logged
 
TD
God Member
*****
Offline


Feyenoord forever!

Posts: 624
Location: Rotterdam, NLD
Joined: 02/12/11
Re: The best analysis program?
Reply #149 - 10/30/17 at 14:41:43
Post Tools
TonyRo wrote on 10/30/17 at 14:17:41:
Hell, nowadays with lichess.org "studies", you can essentially create multi-game PGNs, save them in the cloud, use an opening explorer/DB, Stockfish using your local CPUs, draw arrows and highlight squares, analyze with friends, etc. All for free, all in one place, saved automatically, with no installations needed. It's kind of insane. For most hobbyists that's an all in one deal. As an example, you can check out a study I made a while ago as a companion to a YouTube video: https://lichess.org/study/MgtQ3N9a

I have to check your study someday, because I like the Catalan for White!
  
Back to top
 
IP Logged
 
TD
God Member
*****
Offline


Feyenoord forever!

Posts: 624
Location: Rotterdam, NLD
Joined: 02/12/11
Re: The best analysis program?
Reply #148 - 10/30/17 at 14:39:48
Post Tools
TonyRo wrote on 10/30/17 at 14:17:41:
Stockfish will always be open source and totally free, and most likely world class. Unless you have some really specific reason for paying for Houdini or Komodo, SF will always be sufficient for most hobbyists.

Thanks TonyRo, a friend of mine said just about the same.
« Last Edit: 10/30/17 at 18:02:50 by TD »  
Back to top
 
IP Logged
 
TonyRo
God Member
*****
Offline


I'm gonna crack your skull!

Posts: 1826
Location: Cleveland, OH
Joined: 11/26/07
Gender: Male
Re: The best analysis program?
Reply #147 - 10/30/17 at 14:17:41
Post Tools
Stockfish will always be open source and totally free, and most likely world class. Unless you have some really specific reason for paying for Houdini or Komodo, SF will always be sufficient for most hobbyists.

Hell, nowadays with lichess.org "studies", you can essentially create multi-game PGNs, save them in the cloud, use an opening explorer/DB, Stockfish using your local CPUs, draw arrows and highlight squares, analyze with friends, etc. All for free, all in one place, saved automatically, with no installations needed. It's kind of insane. For most hobbyists that's an all in one deal. As an example, you can check out a study I made a while ago as a companion to a YouTube video: https://lichess.org/study/MgtQ3N9a
  
Back to top
 
IP Logged
 
TD
God Member
*****
Offline


Feyenoord forever!

Posts: 624
Location: Rotterdam, NLD
Joined: 02/12/11
Re: The best analysis program?
Reply #146 - 10/29/17 at 14:30:23
Post Tools
At the moment I am using Stockfish 6 for my game analysis, but I am thinking of bying a new program. That would be Houdini 6 or Komodo 11. Any comments, preferences, pluses & minuses?

Update: I suddenly got the briljant idea of updating Stockfish 6 to Stockfish 8...
  
Back to top
 
IP Logged
 
ChessIsWar
Junior Member
**
Offline



Posts: 63
Location: New York State
Joined: 12/10/14
Gender: Male
Re: Looking for Safe Chess Engine Download or Purchase
Reply #145 - 02/29/16 at 04:47:57
Post Tools
Quote:
http://computerchess.org.uk/ccrl/4040/cgi/engine_details.cgi?print=Details&each_... 7 64-bit 4CPU#Stockfish_7_64-bit_4CPU

Stockfish 7 is beating Komodo 9.3 28-24 in a 40 40 time control.
Stockfish 7 also beats Komodo 9.2 76.5 - 59.5


You know, I had seen these results last night after I posted in this thread already. These do not surprise me. In blitz, Stockfish 290116 64 POPCNT is an absolute monster. It's beating Komodo 9.1 on my machine, using chessbase's live book, at a 4-1 clip. It's elo is so much higher that 9.3 is obviously rated lower than Stockfish 7/Dev at this point. They even have 9.3 performing "worse" than 9.2. Kind of makes it impossible for me to talk myself into buying 9.3 just to see. I know I wouldn't be happy with the results.

I've looked endlessly online and can't find the start date for TCEC 9. Should be extremely interesting.
  
Back to top
 
IP Logged
 
NegiRefutes
Ex Member


Re: Looking for Safe Chess Engine Download or Purchase
Reply #144 - 02/28/16 at 19:23:43
Post Tools
http://computerchess.org.uk/ccrl/4040/cgi/engine_details.cgi?print=Details&each_...

Stockfish 7 is beating Komodo 9.3 28-24 in a 40 40 time control.
Stockfish 7 also beats Komodo 9.2 76.5 - 59.5
  
Back to top
 
IP Logged
 
ChessIsWar
Junior Member
**
Offline



Posts: 63
Location: New York State
Joined: 12/10/14
Gender: Male
Re: Looking for Safe Chess Engine Download or Purchase
Reply #143 - 02/28/16 at 00:00:37
Post Tools
Stockfish 7 is outperforming it's own post release development versions and handling Komodo 9.1 quite nicely on my machine. Currently wondering whether dropping the $60 (for 9.3) or $100 (for the year) on Komodo 9.3. I am by no means a high rated player (I'm about a 1500 USCF skill level but have reached that rating in a short period after beginning to play chess as an adult). Software like Chessbase, Fritz 15, and engines are totally up my ally so I've been spending some time using them lately. I'm considering playing cpu vs cpu chess on PlayChess a little bit, too.

Anyone running K9.3 vs SF7? How's it usually play out? Is it worth it to upgrade from 9.1 to 9.3? Thanks.

  
Back to top
 
IP Logged
 
GMTonyKosten
YaBB Administrator
*****
Offline


Mr Dynamic?

Posts: 3149
Location: Clermont-Ferrand
Joined: 12/19/02
Gender: Male
Re: The best analysis program?
Reply #142 - 01/27/16 at 12:19:42
Post Tools
BenkoFan wrote on 01/12/16 at 10:46:12:
Also seems to be up there with Houdini and Komodo on the computer rating lists.


Stockfish is very strong, I use it all the time. Smiley
  
Back to top
IP Logged
 
BenkoFan
Junior Member
**
Offline


I Love ChessPublishing!

Posts: 58
Joined: 09/16/15
Re: The best analysis program?
Reply #141 - 01/12/16 at 10:46:12
Post Tools
StockFish is open source software. Also seems to be up there with Houdini and Komodo on the computer rating lists.
  
Back to top
 
IP Logged
 
TonyRo
God Member
*****
Offline


I'm gonna crack your skull!

Posts: 1826
Location: Cleveland, OH
Joined: 11/26/07
Gender: Male
Re: The best analysis program?
Reply #140 - 01/11/16 at 01:15:17
Post Tools
Yes.
  
Back to top
 
IP Logged
 
Smyslov_Fan
YaBB Moderator
Correspondence fan
*****
Offline


Progress depends on the
unreasonable man. ~GBS

Posts: 6902
Joined: 06/15/05
Re: The best analysis program?
Reply #139 - 01/11/16 at 00:29:08
Post Tools
Stockfish 7? Is that free?
  
Back to top
 
IP Logged
 
Keano
God Member
*****
Offline


Money doesn't talk, it
swears.

Posts: 2915
Location: Toulouse
Joined: 05/25/05
Gender: Male
Re: The best analysis program?
Reply #138 - 01/08/16 at 18:11:26
Post Tools
After hearing so much of Komodo I got it and installed it. So far I am hugely dissapointed, I still trust far more the evaluations of Houdini 1.5
  
Back to top
 
IP Logged
 
GMTonyKosten
YaBB Administrator
*****
Offline


Mr Dynamic?

Posts: 3149
Location: Clermont-Ferrand
Joined: 12/19/02
Gender: Male
Re: The best analysis program?
Reply #137 - 01/08/16 at 11:29:21
Post Tools
I've started using Stockfish 7, and it seems pretty good.
  
Back to top
IP Logged
 
mattchess
Junior Member
**
Offline


I Love ChessPublishing!

Posts: 63
Location: Seattle
Joined: 08/30/13
Re: The best analysis program?
Reply #136 - 12/07/15 at 18:40:47
Post Tools
I use Komodo 9.3 for analysis but I must say I love the new Fritz 15 engine/GUI combo for training.  Play a game in friend mode and after it presents training positions to review where you missed tactics or threats.
  
Back to top
WWW  
IP Logged
 
proustiskeen
YaBB Moderator
*****
Offline


Hello from Omaha!

Posts: 678
Joined: 08/11/08
Re: The best analysis program?
Reply #135 - 12/07/15 at 16:37:32
Post Tools
It's Rybka 4.1 + 40 elo points or so. Way stronger than Deep Fritz 14, but not in the same league as Komodo, Stockfish or even Houdini.

The GUI does have some nice new features. Most importantly, the database functions have been improved a bit, so while it's no replacement for ChessBase 13, it could work as a first database program for lower rated players.
  
Back to top
WWW  
IP Logged
 
GMTonyKosten
YaBB Administrator
*****
Offline


Mr Dynamic?

Posts: 3149
Location: Clermont-Ferrand
Joined: 12/19/02
Gender: Male
Re: The best analysis program?
Reply #134 - 12/07/15 at 13:15:44
Post Tools
Has anyone tried Fritz 15 yet?
  
Back to top
IP Logged
 
Keano
God Member
*****
Offline


Money doesn't talk, it
swears.

Posts: 2915
Location: Toulouse
Joined: 05/25/05
Gender: Male
Re: The best analysis program?
Reply #133 - 05/01/15 at 18:52:07
Post Tools
will give this Komodo 6 a go, I am still on Houdini 1.5, not sure if that is bad  Embarrassed
  
Back to top
 
IP Logged
 
brabo
God Member
*****
Offline


Welcome chessfriend

Posts: 1064
Joined: 02/02/07
Re: The best analysis program?
Reply #132 - 05/01/15 at 07:05:19
Post Tools
DenVerdsligeRejsende wrote on 04/30/15 at 23:43:07:
I have Komodo 6 and Stockfisk 6 (64-bit). I am now analysing the Caro-Kann Advance for Black with 3...Lf5 as home analyses. I tried both left on overnight on a critical position past move 20 and it made my laptop slower, although my laptop is new and fast (HP brand). Is Komodo should be the main engine nowadays then? When I quit one engine from the two it makes my computer faster again.

Even more confusing to me is that when I have both on I get things like this:

Komodo +0,32 Stockfisk +0,07

Sometimes also it goes one negative one positive:

Komodo +0,23 Stockfisk -0,02

So is one seeing something than the other?


I always use 1 engine per computer. Today I am using 2 engines for analyzing positions so I have 2 computers. This avoids the loss of time for swapping between both engines. An extra benefit is that I have also an extra screen on which I can permanently monitor both engines in parallel. That is still one less than e.g. Shipov when doing his online commentary see http://www.chessintranslation.com/2012/05/two-cats-and-three-computers-sergey-sh...

If you use the last versions of Komodo and Stockfish then I don't think there is any substantial difference in strength overall. That was also proven in the previous TCEC final. However each engine will have specific positions which are played a bit better than the other. So it makes sense to do the analysis with both.

Each engine has its own evaluationmethod. A small difference of evaluation between engines doesn't necessarily mean that some engine sees anything extra just that it evaluates the position a bit different. It is up to the operator to detect if the difference in evaluation is a matter of pure evaluation or more important moves. Sometimes it is sufficient to look at the mainlines but often more analysis is needed to understand the difference in evaluation.
  
Back to top
WWW  
IP Logged
 
DenVerdsligeRejsende
Senior Member
****
Offline


I Love ChessPublishing!

Posts: 269
Location: Křbenhavn
Joined: 05/16/14
Re: The best analysis program?
Reply #131 - 04/30/15 at 23:43:07
Post Tools
I have Komodo 6 and Stockfisk 6 (64-bit). I am now analysing the Caro-Kann Advance for Black with 3...Lf5 as home analyses. I tried both left on overnight on a critical position past move 20 and it made my laptop slower, although my laptop is new and fast (HP brand). Is Komodo should be the main engine nowadays then? When I quit one engine from the two it makes my computer faster again.

Even more confusing to me is that when I have both on I get things like this:

Komodo +0,32 Stockfisk +0,07

Sometimes also it goes one negative one positive:

Komodo +0,23 Stockfisk -0,02

So is one seeing something than the other?
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #130 - 04/30/15 at 16:14:17
Post Tools
Keano wrote on 04/30/15 at 13:45:42:
Vass wrote on 04/29/15 at 14:42:45:
A new candidate for best analysis engine: Komodo 9
http://komodochess.com/
Roll Eyes


Very interesting - is there a free version of this engine available to test?


Unfortunately no, there is not!
But, Komodo 6 is free already: http://komodochess.com/pub/Komodo-6.zip
Wink
  
Back to top
 
IP Logged
 
Keano
God Member
*****
Offline


Money doesn't talk, it
swears.

Posts: 2915
Location: Toulouse
Joined: 05/25/05
Gender: Male
Re: The best analysis program?
Reply #129 - 04/30/15 at 13:45:42
Post Tools
Vass wrote on 04/29/15 at 14:42:45:
A new candidate for best analysis engine: Komodo 9
http://komodochess.com/
Roll Eyes


Very interesting - is there a free version of this engine available to test?
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #128 - 04/29/15 at 14:42:45
Post Tools
A new candidate for best analysis engine: Komodo 9
http://komodochess.com/
Roll Eyes
  
Back to top
 
IP Logged
 
amza iulian
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 10
Joined: 02/28/15
Re: The best analysis program?
Reply #127 - 03/03/15 at 03:18:46
Post Tools
Vass wrote on 03/02/15 at 09:08:26:

I have one game in corr game with +1 but all engine is not able to see the victory. maybe becouse is not possible. ( the position is closed, is only one column free. ) Cry
  
Back to top
 
IP Logged
 
amza iulian
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 10
Joined: 02/28/15
Re: The best analysis program?
Reply #126 - 03/02/15 at 10:14:46
Post Tools
I suppose Stockfish 6 remain for while the best choice. For accuracy, you can use 2 engine to compare. And your method, to advance one more or two moves ahead look good but take times and patience.
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #125 - 03/02/15 at 09:08:26
Post Tools
  
Back to top
 
IP Logged
 
amza iulian
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 10
Joined: 02/28/15
Re: The best analysis program?
Reply #124 - 03/02/15 at 04:06:41
Post Tools
Vass wrote on 03/01/15 at 21:34:39:
amza iulian wrote on 03/01/15 at 04:18:08:
Btw, after 32 or max 40 is very rare to see same change in first choice of program. For open positions is clear, Stockfish is good analyzer, but for closed position Komodo is the best? Maybe is better to use 3 core...one for K, one for S, one for H...What your opinion,Mr. Vass?


TalJechin said it already:

TalJechin wrote on 02/03/15 at 16:00:30:
One should use at least one engine from every one of the major 'families' as they all have different kinds of positions where one or two of them 'understand' the positions better than the rest.


And no, I never noticed Komodo to be especially good at analysis of closed position. It has other qualities though.

The question remain without answer. What engine is better in closed position?
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #123 - 03/01/15 at 21:34:39
Post Tools
amza iulian wrote on 03/01/15 at 04:18:08:
Btw, after 32 or max 40 is very rare to see same change in first choice of program. For open positions is clear, Stockfish is good analyzer, but for closed position Komodo is the best? Maybe is better to use 3 core...one for K, one for S, one for H...What your opinion,Mr. Vass?


TalJechin said it already:

TalJechin wrote on 02/03/15 at 16:00:30:
One should use at least one engine from every one of the major 'families' as they all have different kinds of positions where one or two of them 'understand' the positions better than the rest.


And no, I never noticed Komodo to be especially good at analysis of closed position. It has other qualities though.
  
Back to top
 
IP Logged
 
amza iulian
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 10
Joined: 02/28/15
Re: The best analysis program?
Reply #122 - 03/01/15 at 04:18:08
Post Tools
Btw, after 32 or max 40 is very rare to see same change in first choice of program. For open positions is clear, Stockfish is good analyzer, but for closed position Komodo is the best? Maybe is better to use 3 core...one for K, one for S, one for H...What your opinion,Mr. Vass?
  
Back to top
 
IP Logged
 
amza iulian
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 10
Joined: 02/28/15
Re: The best analysis program?
Reply #121 - 03/01/15 at 04:12:45
Post Tools
I know your point. Still, if you buy a new processor 5960X I think, for your correspondence chess i can loose against me, if I use for exemple only i5 quad core, no matter is I will stay more time than you to analyze. Is not only speed, is also RAM, FSB..etc...
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #120 - 02/28/15 at 17:33:46
Post Tools
amza iulian wrote on 02/28/15 at 10:38:10:
I am not sure about this. If I use octocore FX 8350 1 h sure is better than single core 8h.


Reading is always a good thing to do!
http://www.cruxis.com/chess/manual/index.html
And especially:
2.7 Some Frequently Asked Questions
3.4 Cores and Threads Management
Wink
  
Back to top
 
IP Logged
 
amza iulian
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 10
Joined: 02/28/15
Re: The best analysis program?
Reply #119 - 02/28/15 at 10:40:02
Post Tools
TalJechin wrote on 02/03/15 at 16:00:30:
One should use at least one engine from every one of the major 'families' as they all have different kinds of positions where one or two of them 'understand' the positions better than the rest.

TalJechin wrote on 02/03/15 at 16:00:30:
One should use at least one engine from every one of the major 'families' as they all have different kinds of positions where one or two of them 'understand' the positions better than the rest.

TalJechin wrote on 02/03/15 at 16:00:30:
One should use at least one engine from every one of the major 'families' as they all have different kinds of positions where one or two of them 'understand' the positions better than the rest.

TalJechin wrote on 02/03/15 at 16:00:30:
One should use at least one engine from every one of the major 'families' as they all have different kinds of positions where one or two of them 'understand' the positions better than the rest.

TalJechin wrote on 02/03/15 at 16:00:30:
One should use at least one engine from every one of the major 'families' as they all have different kinds of positions where one or two of them 'understand' the positions better than the rest.

I think for closed position Komodo can be a solution. For sharp could be Stockfish.
  
Back to top
 
IP Logged
 
amza iulian
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 10
Joined: 02/28/15
Re: The best analysis program?
Reply #118 - 02/28/15 at 10:38:10
Post Tools
I am not sure about this. If I use octocore FX 8350 1 h sure is better than single core 8h.
Vass wrote on 06/21/14 at 19:49:49:
Configuring the number of cores, hash size, endgame tablebases support and...pondering on/off and setting the proper use of opening books (if you do match tests between engines) is enough for a start.
For infinite analysis you have to configure maximum hash (irrespective of the engine you use). It means that if your PC has 2048 MB RAM, then your maximum hash size is 1024 MB. For matches between engines it depends on what time control they play. If it's the classical 5' blitz control - 128MB for each engine is enough (reading a big-sized hash file while playing blitz is counter-productive). 256MB hash for each engine is good for matches on 60'+15" time control for example...and so on..
"Large pages" can improve the speed of analysis with 10 to 15 per cents, but not for long (and not every engine can handle them right, too). And besides, you have to be experienced in order to set properly the "large pages" parameters on your PC (if it supports them, of course).
Usually, engines are not good with openings or opening-related positions. That's why their developpers recommend using human-made small opening books (up to 8-15 moves only). Do not rely on proper evaluation of the openings by engines up to the 15th move!
If you let them play in matches on their own, they'll probably start with 1.d4 d5 2. Bf4 Nf6 3.Nc3 Nc6, or similar, all the time.  Grin
...And yes, analysing on one core for two hours is better than analysing on two cores for one hour only.

  
Back to top
 
IP Logged
 
TalJechin
God Member
*****
Offline


There is no secret ingredient.

Posts: 2892
Location: Malmö
Joined: 08/12/04
Gender: Male
Re: The best analysis program?
Reply #117 - 02/03/15 at 16:00:30
Post Tools
One should use at least one engine from every one of the major 'families' as they all have different kinds of positions where one or two of them 'understand' the positions better than the rest.
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #116 - 02/03/15 at 11:29:51
Post Tools
My humble opinion: very good for infinite analysis and not so good for the Aquarium's IDeA way of analysis.
Still, my preferred chess engine is Houdini 4.  Wink
  
Back to top
 
IP Logged
 
GMTonyKosten
YaBB Administrator
*****
Offline


Mr Dynamic?

Posts: 3149
Location: Clermont-Ferrand
Joined: 12/19/02
Gender: Male
Re: The best analysis program?
Reply #115 - 02/03/15 at 00:42:59
Post Tools
Vass wrote on 02/02/15 at 17:45:26:
The new Stockfish 6 is available for download


Yes, and it's very good! Wink
  
Back to top
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #114 - 02/02/15 at 17:45:26
Post Tools
The new Stockfish 6 is available for download at https://stockfishchess.org/ and at https://stockfishchess.org/download/

It may be a good try for analyzing with it!  Wink
  
Back to top
 
IP Logged
 
HgMan
God Member
*****
Offline


Demand me nothing: What
you know, you know

Posts: 2330
Location: Up on Cripple Creek
Joined: 11/09/04
Gender: Male
Re: The best analysis program?
Reply #113 - 12/05/14 at 22:34:03
Post Tools
A few years ago, I found I got a lot of value out of Junior 10. Typically, it provided inferior analysis to Rybka, Deep Shredder, and Deep Hiarcs, but it also offered the most radically different options. Pairing that more aggressive style with careful analysis of the ensuing positions yielded me a number of points in correspondence chess.
  

"Luck favours the prepared mind."  --Louis Pasteur
Back to top
 
IP Logged
 
GeneM
Senior Member
****
Offline


Tournament winner gets
two fun filled knights!

Posts: 303
Location: near Seattle WA USA
Joined: 01/12/08
Re: The best analysis program?
Reply #112 - 11/08/14 at 04:40:21
Post Tools
.
Vass wrote on 06/21/14 at 19:49:49:
Do not rely on proper evaluation of the openings by engines up to the 15th move!
If you let [chess engines] play in matches on their own, they'll probably start with 1.d4 d5 2. Bf4 Nf6 3.Nc3 Nc6, or similar, all the time.  Grin

Vass, Your 2.Bf4 reminds me of a new book that is coming out in a few months.
For Black, after 1.d4 d5 2.c4 (or 2.Nf3) it recommends 2...Bf5.

Book: The Liberated Bishop Defence, by Alexey Bezgodov, due 2015-January

http://www.amazon.com/Liberated-Bishop-Defence-Surprising-Repertoire/dp/90569154...
.
  

GeneM , CastleLong.com , FRC-chess960
Back to top
WWW  
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #111 - 09/08/14 at 11:54:31
Post Tools
bearheart wrote on 09/08/14 at 11:32:41:
Please, pardon  my total ignorance, but how does one run it?  I extracted it after downloading & unzipping, however, when I when to run as administrator I get a DOS window that appears to show only the coders name.  It has been too many decades since I have played with command lines prior to the creation of GUIs. 
Can anyone give me the directions?   
Too old to remember... Embarrassed

Well, bearheart, you need a GUI to run this engine in a proper mode.
If you don't have Chessbase, Fritz, ChessOK Aquarium or alike, you can download a free GUI (for example Arena at http://www.playwitharena.com/ , Scid Vs PC at http://scidvspc.sourceforge.net/ or even my favourite free GUI called ChessX at http://chessx.sourceforge.net/) and install the engine under one of these.  Wink

EDIT: Of course, Rybka Aquarium will do.
  
Back to top
 
IP Logged
 
bearheart
YaBB Newbies
*
Offline


I love ChessPublishing.com!

Posts: 23
Location: Mound
Joined: 02/19/05
Re: The best analysis program?
Reply #110 - 09/08/14 at 11:50:34
Post Tools
Oh well, never mind.  I figured it out and can use it with my Rybka Aquarium program.   Roll Eyes
  
Back to top
 
IP Logged
 
bearheart
YaBB Newbies
*
Offline


I love ChessPublishing.com!

Posts: 23
Location: Mound
Joined: 02/19/05
Re: The best analysis program?
Reply #109 - 09/08/14 at 11:32:41
Post Tools
Please, pardon  my total ignorance, but how does one run it?  I extracted it after downloading & unzipping, however, when I when to run as administrator I get a DOS window that appears to show only the coders name.  It has been too many decades since I have played with command lines prior to the creation of GUIs. 
Can anyone give me the directions?   
Too old to remember... Embarrassed
  
Back to top
 
IP Logged
 
Vladimir
Junior Member
**
Offline



Posts: 60
Joined: 05/22/09
Re: The best analysis program?
Reply #108 - 09/08/14 at 01:41:07
Post Tools
Vass wrote on 09/04/14 at 16:37:02:
A new candidate for best analysis program: Komodo 8 engine at http://komodochess.com/ is out now.
The developpers promised +50 ELO points for this new engine version. First tests started...
For now, some engine-testers confirm +17 ELO points, though more games are obviously necessary.  Wink


Concurrent with this, the previous versions Komodo 5.1 and Komodo CCT are also now released for free to download at the Komodo Chess website.

It's great having Stockfish available for free, and now a second opinion that's also free and comparable in strength.
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #107 - 09/04/14 at 16:37:02
Post Tools
A new candidate for best analysis program: Komodo 8 engine at http://komodochess.com/ is out now.
The developpers promised +50 ELO points for this new engine version. First tests started...
For now, some engine-testers confirm +17 ELO points, though more games are obviously necessary.  Wink
  
Back to top
 
IP Logged
 
Michael Ayton
God Member
*****
Offline


‘You’re never alone with
a doppelgänger.’

Posts: 1947
Location: durham
Joined: 04/19/03
Gender: Male
Re: The best analysis program?
Reply #106 - 06/23/14 at 09:53:26
Post Tools
I think I’m nearly sorted! (I have the 3-piece Gaviota tablebase supplied with Arena, and just have to download the 4- and 5-piece bases.)

Yes, I remember playing blitz games against loads of engines years ago and as Black they all met 1 d4 stuff with …d5 and …Nc6 (but still usually whupped me!). But all depends on the position, doesn’t it? I’ve recently been looking at 1 e4 d6 2 d4 g6 3 Nc3 a6 4 f4 b5 5 Nf3 Bb7 6 Bd3 Nd7 7 e5 c5 8 Be4 Be4!? (N) 9 Ne4 cd – and here I’ve found engines have made useful suggestions. But in a position like this one – 1 d4 Nf6 2 c4 d6 3 Nc3 Nbd7 4 e4 e5 5 Nf3 g6 6 Be2 Bg7 7 0-0 0-0 8 Be3 Qe7 9 d5 c5 – there’s surely just no point in turning the engine on!
Your help’s been invaluable, Vass – thanks a lot! I think you should write an online guide on the subject … Smiley
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #105 - 06/21/14 at 19:49:49
Post Tools
Configuring the number of cores, hash size, endgame tablebases support and...pondering on/off and setting the proper use of opening books (if you do match tests between engines) is enough for a start.
For infinite analysis you have to configure maximum hash (irrespective of the engine you use). It means that if your PC has 2048 MB RAM, then your maximum hash size is 1024 MB. For matches between engines it depends on what time control they play. If it's the classical 5' blitz control - 128MB for each engine is enough (reading a big-sized hash file while playing blitz is counter-productive). 256MB hash for each engine is good for matches on 60'+15" time control for example...and so on..
"Large pages" can improve the speed of analysis with 10 to 15 per cents, but not for long (and not every engine can handle them right, too). And besides, you have to be experienced in order to set properly the "large pages" parameters on your PC (if it supports them, of course).
Usually, engines are not good with openings or opening-related positions. That's why their developpers recommend using human-made small opening books (up to 8-15 moves only). Do not rely on proper evaluation of the openings by engines up to the 15th move!
If you let them play in matches on their own, they'll probably start with 1.d4 d5 2. Bf4 Nf6 3.Nc3 Nc6, or similar, all the time.  Grin
...And yes, analysing on one core for two hours is better than analysing on two cores for one hour only.
  
Back to top
 
IP Logged
 
Michael Ayton
God Member
*****
Offline


‘You’re never alone with
a doppelgänger.’

Posts: 1947
Location: durham
Joined: 04/19/03
Gender: Male
Re: The best analysis program?
Reply #104 - 06/21/14 at 10:45:05
Post Tools
Great news that I don't have to change lots of things! But now I'm a bit confused! First I learn that GUIs like Arena are fine as long as you configure engines properly; next I learn that in most cases I should just leave the defaults as they are! So I'm left wondering what it is I should configure? Is it just cores and pondering, or if not, what else? (What about hash? According to the Houdini site my optimum hash is 1024 MB, but is this for all engines, and how would I change it anyway?)

Under 'Tournament Conditions" I have Hash 128MB, Tablebases On, Ponder On. On individual engines I see all sorts of other things, like pawn hash, PV hash, eval hash, SMP mode, large pages, LMR, null move pruning, and just hope and trust I can leave them all well alone!!

I know what you mean about personal experience in analysis. With opening or opening-related positions especially, mine tells me that intervening to explore a range of plausible moves using human knowledge of what kinds of moves are typically played in a given position is often much better than just switching on Infinite or Demo analysis and sitting back (though you have to be 'objective', which can be hard!). But in these latter cases, are you saying that even on a dual-core PC it's good to try a single-core analysis as well because it actually might be just as good? (I've noticed that on longer time limits, an engine in Demo mode analysing an opening-related position will sometimes play moves that look a lot less 'human' and may not be so good as the moves it plays on a shorter time limit!?)

  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #103 - 06/20/14 at 23:05:16
Post Tools
From: http://www.cruxis.com/chess/manual/index.html?cores_and_threads_management.htm
"When using multiple threads, the Split Depth parameter defines the minimum depth at which work will be split between cores."
Imho, you have to "autotune" Houdini, but leave the other engines you have as they are. Nobody knows the settings of these engines (except their developpers, of course). One can expect that the default settings are best, except when told otherwise.
When using multiple threads, it is supposed, logically, the eficiency of the engines to be multiplied automatically by a per cent equal to 100 (i.e. one core = 100 per cent, two cores = 2x100 per cent). But no, it's not like that. Adding a second core doesn't mean you add another 100 per cent efficiency to your analysis power. It means you add efficiency of about 75 per cent to your previous 100. And these per cents diminish with every core you add to your PC configuration - all because of the fact that the cores work in parallel and can often try to work simultaneously on the analysis' lines. So, maybe for analysis it's better to have one powerful core than two weaker ones.
Well, the last is valid when trying infinite analysis. While if you use Aquarium IDeA's methods for analysis, you'll definitely need more cores even if they are weaker. But, that's another story..
As for "how long" when using 'infinite analysis' - mhmm, I think the answer is in the name of the subject.  Grin
Now seriously.. Theoretically, the infinite analysis is better when using one powerful core for two days (with maximum hash) than using two cores for one day only. And it depends on the position you analyse for how long you would allow your engine to dig in. There is a "horizon" that engines can't pass, because of the pruning. I.e. they tend to stay solid in their decisions and don't change their evaluations too easy, even they spot something 'interesting' on a very big distance from the root. The pruning at depth, say, 32 plys is enormous - so, it's not an easy decision to change the first move. A year ago, the developper of Houdini suggested (and even shared) the idea to write a special code in the engines which turn the engines to diminish the plys' depth when spotting a new opportunity and to continue from a firm ground.
Some say they never saw an engine to change its decision after 32 plys depth, but I've seen it.. So, as I already mentioned, it depends on the postion you analyse.
Besides, something you don't have to forget - engines are not designed to work in infinite mode. They are made to fight each other in a head to head matches on different time controls.. So, when you use them for infinite analysis, you definitely don't use them as expected. There are other methods that are more efficient, when trying to achieve a good analysis.
I can talk about engines all night long...so it's time to stop, I think. All the more, the self experience is better than reading someone's "how to" writings.  Wink
  
Back to top
 
IP Logged
 
Michael Ayton
God Member
*****
Offline


‘You’re never alone with
a doppelgänger.’

Posts: 1947
Location: durham
Joined: 04/19/03
Gender: Male
Re: The best analysis program?
Reply #102 - 06/20/14 at 16:42:20
Post Tools
Thanks for all this info and the links Vass! Since I was last here I've had lots of fun, and also managed to answer most of my questions myself!  Wink Just one or two teasers below! ...

Split depth. The Houdini 4 website suggests using ‘autotune’ to determine whether a non-10 value is best for non-i5 and -i7 PCs. Houdini 1.5a also allows this test, according to which my SD is best set as 12. But is this true for all engines not just Houdini, or should I leave my existing defaults? At present I have these wildly divergent defaults: Black Mamba 10, Critter 1.6a 5(!), Hannibal 4(!), Robbolito 14, Stockfish 5 modern (and DD) 0(!!). Ivanhoe has three different types of SD! – CUT (16), ALL (14), and PV (14).

‘Max number of threads per split point.’ I have defaults of 4 for Hannibal, 5 for Sting SF3 and 5 for Stockfish DD (which also has ‘Idle threads sleep’, unticked). Should I leave all of these, because maximums is all they are?

Infinite analysis. After roughly how long, on a 2-core PC and on a 1-core PC, is it probably not worth continuing with it? (How long is a piece of string?!)
  
Back to top
 
IP Logged
 
Michael Ayton
God Member
*****
Offline


‘You’re never alone with
a doppelgänger.’

Posts: 1947
Location: durham
Joined: 04/19/03
Gender: Male
Re: The best analysis program?
Reply #101 - 06/17/14 at 15:41:53
Post Tools
Thanks for both these Vass! I am currently seeing off that old toad work, grrr, but will get with all of this v. soon!
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #100 - 06/17/14 at 11:12:43
Post Tools
By the way, Robert Houdart, the author of Houdini has an excellent manual which can be find online at:
http://www.cruxis.com/chess/manual/index.html?cores_and_threads_management.htm
Under "3.Houdini Configuration", he explains about hash, cores, threads, split depth and other parameters of the engine. A must-read for everyone!  Wink
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #99 - 06/17/14 at 10:53:54
Post Tools
I had to run Arena 3.5 GUI in order to be able to answer your questions.  Smiley
Black screens, after double-clicking on the engine-exe file in your Windows Explorer, are only for receiving information about the engine - whether it is recognized by your PC and if this engine will run on your PC, who is the author, what does this engine recognize about your computer configuration and so on..
All your parameters' changing has to be done on the settings' menu of the GUI.
Under the Arena menu Engine/Manage/UCI there is an option "Common max. CPU Cores Setting" with value "1" (in my case). This way the GUI tells the engine (when installing) that this engine will use one processor core only (when running under this GUI). So you can install two engines (they'll use one core each) and run them one after another in the GUI to analyse a certain position. Still, you can click with your mouse and browse whatever you want on your PC - it won't freeze because every core has two threads and one of this threads will handle your new activity (for example, opening Internet Explorer and so on) for a moment, while the other thread will keep running the analysis of the engine. After stopping your activity, the thread you use for your browsing will return to its usual job - i.e. joining the other thread of the core to prolong the analysis. This feature distinguish the new "modern" processors from the old ones.
As for how to run two engines simultaneously in your Arena GUI: 1. Make sure they use one core each!; 2. Under Engine/Manage/Select there is a box called "loaded engines" - load two of your engines (from the right box with the installed engines -> to the left one) and press "Start" button on the down left side!
I'm sure there are other ways, too...but first start with this one!
Enjoy!  Wink
  
Back to top
 
IP Logged
 
Michael Ayton
God Member
*****
Offline


‘You’re never alone with
a doppelgänger.’

Posts: 1947
Location: durham
Joined: 04/19/03
Gender: Male
Re: The best analysis program?
Reply #98 - 06/17/14 at 09:00:21
Post Tools
Hi Vass and thanks again!
I am a little worried we may be talking at 'cross purposes'!?  Embarrassed Sure, Arena can give troubles but the point is I am perverse! -- I actually enjoy the challenge of getting the best from engines in Arena and WinBoard and the 'learning' that goes with it! I am actually a good 'googler' (my work requires it!) but there's very little good online help with these things (and Arena's help file is useless!). So I may be slow but what I still don't get is how I can change an engine's 'thread' (or other) parameters except by doing it within the GUI. For example, I understand that when Robbolito's black screen says '4 CPUs found, using 4 threads' it means '4 threads found', but should I be changing this via the black screen (then saving that engine somehow)?? I can't find anything on this, nor on how to load two engines simultaneously in Arena to analyse the same position.

Thanks for the POPCYNT info! I couldn't find precisely that online either! How did you acquire your knowledge? Probably not chiefly by googling! I think you are a 'natural' ... (And I freely admit, I am a 'clot' ... Sad)
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #97 - 06/16/14 at 20:06:56
Post Tools
As for launching two times the GUIs, no - it's not serious. Every decent GUI (like Fritz) can run as many engines you want simultaneously to analyse one position, for example...supposedly the number of engines you run is not exceeding the number of the available processor cores.
I think you can run two engines (one core each) simultaneously under the Arena GUI, too..
"Ponder on" means the engine is pondering its next move while it is not its turn to move.
"Ponder off" means this engine is stop pondering (and not using the processor's core) while the other one is pondering its next move.
So, "ponder on" increases the performance of an engine while playing a game in a match between two engines. You can try to run a match between Houdini 4 (with ponder off) and say Gull (with ponder on) and you'll see that Gull will have better performance, though objectively is weaker than Houdini.  Wink
And yes, "popcnt" is the other name of the SSE... instruction.
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #96 - 06/16/14 at 19:45:27
Post Tools
Well, well, well - too many questions, indeed!  Cheesy
Michael, most of the questions can be answered by Uncle Google. For example, look at the processor you have: Intel(R) Core(TM) i3 CPU 540 @ 3.07GHz
http://ark.intel.com/products/46473/Intel-Core-i3-540-Processor-4M-Cache-3_06-GH...
Pointng your mouse on "# of Cores" or "# of Threads" and so on...there and clicking on them you can see the definitions, such as: "A Thread, or thread of execution, is a software term for the basic ordered sequence of instructions that can be passed through or processed by a single CPU core."
Althouh the language is somewhat technical, you have to browse through most of these categories in order to be able to explain for yourself all unclear points you have up to now.
I see you have black screens when you click on Strelka 6 x32 and Protector x64 only.
I never used Protector, so I can tell you nothing about it.. As for Strelka 6, I'm sure this engine doesn't have parameters (it's a raw version of Houdini, after all), so it cannot return you even a single piece of information. (Suppose, the case with Protector is the same.)
Anyway, it seems your processor is a "modern" one and supports all SSE.., AVX and so on versions of the engines. From the abovementioned site when clicking on "Instruction Set Extensions" for your processor (SSE4.2): "Instruction Set Extensions are additional instructions which can increase performance when the same operations are performed on multiple data objects. These can include SSE (Streaming SIMD Extensions) and AVX (Advanced Vector Extensions)."
Some engines (when you click on their exe files in your Windows Explorer) read your PC configuration and return pieces of information, which can be very different at first sight. Well, don't worry - they cannot read different things on your PC! You know you have one processor with two cores (which have two threads each - summing up to 4 thread in common). As for how each engine will return information to you (2 processores, 4 threads or two cores...) is another story. When you know what you have you'll understand what they mean..
I definitely don't use Arena although I have it on my PC. I recommend you to buy (at a discount) an old version of Fritz (say 11, 12 or 13 version - the last version is 14 and has some new features which are not important for you in connection with installing engines and so on) and to forget all your troubles with Arena.
As for Arena 3.5 - it's a free GUI which can do many things but is somehow poor in details when you have to do something different than just "kick and rush". My major tip here can be: First see how the engine-exe returns information about your PC when clicking on it! Then go to the major Settings menu (right-clicking cannot be a serious attempt to fix something Smiley ) and try to "untick" all the common settings. Install engines one by one (after knowing how the given engine reads the configuration of your PC) and configure every engine for itself cosidering its specific features (some of them "say" cores, some "say" threads...but you'll know what they really mean), configure the number of cores and threads they should use as you want.
Let me say one important thing here: Your processor has 2 cores, which have 4 threads in common. Most of the chess engines are designed to use real cores in the manner [one core (two threads) - one engine]. Don't try to fix and change parameters in order to use 4 engines on all 4 threads simoultaneously. It will definitely freeze your computer...and the engines won't run. Every engine, when started, is trying to use one core (and not one thread out of this core). Threads can help for other tasks, such as....when you run a chess engine on two cores simultaneously and trying at the same time to open a program or browse the Internet, one of these 4 threads will fulfil the task preventing the freezing of your PC (and then will return to fulfil the tasks of the chess engine). So, the maximum of your processor is to run an engine on 2 cores (say, for faster analysis), or to run two engines simultaneously on one core each. Every try to run two engines that are set to use two cores each will bring the performance of your PC to a freezing. And that's because they both try to use two cores (2 x 2 = 4 cores), but you have only two, so they'll try to use two threads each instead of two cores each (because there are no others available) and therefore all your 4 threads will be taken by the two engines (and not a single thread will be left untaken) - therefore even a click with your mouse will struggle for the resources of your PC...and the engines are usually very hungry for resources, especially Houdini....  Wink
  
Back to top
 
IP Logged
 
Michael Ayton
God Member
*****
Offline


‘You’re never alone with
a doppelgänger.’

Posts: 1947
Location: durham
Joined: 04/19/03
Gender: Male
Re: The best analysis program?
Reply #95 - 06/16/14 at 15:54:13
Post Tools
Hi Vass and many thanks, incl. for the test suites!

I am learning, but I have lots of questions! – so please only respond to any of this if you have time …

I’ve listed my ‘black screens’ below. I am perplexed as to why some engines say ‘4 CPUs found’, when (so I believe) I have 1 CPU, 2 cores, and 4 threads. Also, as you can see, only two engines give me a totally blank screen, but these engines seem to run fine – in what way does my CPU not support them? (And what is POPCNT – is it also an instruction?)

My next issue is changing engine parameters! In Arena, I can see only two ways of doing this. With the engine open in the GUI you can right-click on the engine analysis, but although this brings up a screen showing a lot of parameters, cores (‘threads’) is not among them (except with Stockfish DD)! The other way is via the pulldown menu Engines—Manage—Uci, and this (which I presume applies to all engines*) shows ‘Common max CPU cores setting’ at the bottom. At present this shows ‘1’ and is ticked (so I guess I have only ever been using 1 core per engine!). So if I change it to ‘2’, does that mean that while it is changed, all my engines that can use 2 cores will do so, while those that can't will continue to use 1 without problems? The screen also says ‘For values not set as “common” here, the engine-specific values are in effect (Ctrl +1/Ctrl +2/Ctrl +3)’, so presumably if I untick it (whatever number it says), most of my engines will use both of my cores? So to sum up: is this pulldown menu a perfectly satisfactory/the best/the only way of changing the ‘core’ parameters of UCI engines installed in Arena, or is there a better way?

It seems to be similar in WinBoard. ‘Engine—Engine #1 Settings’ brings up an engine-specific list of parameters which does not include cores (threads), but ‘Options—Common Engine Settings’ brings up a ‘common’ menu which does. Except that it says ‘Max nr of CPUs’ (set to ‘1’) – so does ‘CPUs’ here mean ‘threads’, parallel perhaps with what those ‘black screens’ mean when they say ‘CPUs’?

Finally: when you speak of running two engines, each having a 1-core parameter, ‘simultaneously’, do you mean I can do this without problems just by launching the GUI (Arena or WinBoard) twice? Also, presumably, if I am conducting a match between two engines each using 1 core, I should put Ponder On for maximum strength? And similarly, if I am analysing using ‘Demo’ mode (where the engine actually plays itself), presumably I can do this in two ways: 2 core + Ponder Off or 1 core + Ponder On – but if I am right, which would be stronger/best?

* Strangely(?), Arena 3.5 has (I notice) introduced 'Common max. CPU cores setting for UCI engines', whereas in Arena 3.0 it was possible to alter the thread settings of far more engines individually by right-clicking on the analysis pane!

*****

Stockfish_14053109_x64_modern --- 5 64 SSE4.2
Stockfish_14053109_x64 --- 5 64
Stockfish dd_x64_modern --- DD 64 SSE4.2
Houdini_15a_x64 --- Info string POPCNT enabled; info string 128 MB hash
Gull 3 x64 --- Gull 3 x64
Strelka6 --- [Completely blank]
Komodo-5-64bit --- Using hardware POPCNT
Critter_1.6a_64bit --- Hardware POPCNT enabled
PanChess 00.537.x64 --- HardPOPCNT : Yes
Fire_3.0_SSE42 --- Fire 3.0 x64
Bouquet 1.8 x64 --- 4 CPUs found, using 1 CPU; POPCNT
Ivanhoe999946h Mode_Analysis --- 4 CPUs detected; Prefetch + PopCnt + multicore + hyper hash enabled
Ivanhoe999946h Mode_Game_Play ---4 CPUs detected; Prefetch + PopCnt + multicore enabled
Robbolito 0.21Q x64 --- 4 CPUs found, using 4 threads
Black Mamba_MP_x64 --- 4 CPUs found, using 1 thread; POPCNT supported
DeepSaros_416_popcnt_w64 --- Compiled with POPCNT, compiled with Prefetch
DeepSaros_416_w64 --- Compiled with Prefetch      
Sting SF 3 VE_pop64 --- SF 3 VE_pop64; CPU has hardware POPCNT
Sting SF 3 VEx64 --- SF 3 VEx64
Protector_Win64 --- [Completely blank]
Hannibal1.4bx64 --- Hannibal1.4bx64
« Last Edit: 06/16/14 at 17:16:34 by Michael Ayton »  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #94 - 06/14/14 at 18:19:20
Post Tools
Well, I'm not an expert, but I'll try to give you some tips.
As long as I know all Intel(R) Core(TM) i3 CPU-s have 2 cores, which means you can use both for analysis simultaneously or separately. You can play matches between engines with "ponder-on"giving one core to each engine...and so on.
When you install an engine on your PC, you'll have to go to the parameters of the engine and give it the number of cores you wish to use. Most of them automatically read your PC and take all the cores of your PC, but not all of them. The number of cores is under the parameter "threads". Don't try to give all your 4 threads on an engine, because your computer can become irresponsible to commands and even if it doesn't happen, it is counter-productive for the analysis. So, under parameters' "threads" you always have to understand real CPU-cores.
For example, you can install an engine twice - the first time on 2 cores, while changing the threads' parameter to 2 and use this engine for analysis on two cores....and the second time to install it on one core (changing the name of the engine like this: Stockfish 5_1core) to use it for games between engines or for analysis on one core. And if you have two engines installed on one core you can easily run them both simultaneously to analyse a position and see the differences they show.
As for how to determine which version of the engine you can install on your PC, say Stockfish, you can easily see if your CPU supports the instructions that are put into the engine from the CPU-Z screen. And even if you are not sure about it, you can double-click on any engine exe in your Windows Explorer - if you see a black screen only, with no information, then your CPU doesn't support this version of the engine, while if you see the same black screen, but with the information for your PC about cores, threads and so on, then your CPU supports this version of the engine.
Of course, if you can use the modern versions of an engine it will be better, because you will gain speed - the specific SSE.., AVX and other CPU instructions help the modern engines to perform faster.
And yes, running analysis on Multi PV is slightly weakening the performance of the engine, because the engines are created for being "warriors" - to fight and win against alikes.. No one created an engine to be good for analysis.
Here I uploaded some tests in pgn, cbv and epd format that some friends of mine gave to me:
http://www19.zippyshare.com/v/89246204/file.html - tactical positions tests
http://www19.zippyshare.com/v/97036595/file.html - endgame tests
Feel free to use them! My friends tell me that a standart test for an engine is to give it a set of 100 tactical positions, giving the engine 2 minutes on each position on one CPU core. Therefore the performance of every engine is counted in per cents automatically - say, Stockfish's performance is 35 out of 100 (35%), Houdini's one 38 out of 100 (38%) and so on..
I myself, never ran a test on my own..  Roll Eyes
  
Back to top
 
IP Logged
 
Michael Ayton
God Member
*****
Offline


‘You’re never alone with
a doppelgänger.’

Posts: 1947
Location: durham
Joined: 04/19/03
Gender: Male
Re: The best analysis program?
Reply #93 - 06/14/14 at 12:10:00
Post Tools
Hi Vass and thank you for a very helpful reply, as ever!

I ran CPU-Z and it tells me 'Cores 2, Threads 4', and under 'Instructions' it gives MMX; SSE, plus SSE2, 3, 4.1, 4.2; SSSE3; EM64T; VT-x. What is it here that tells me (or doesn't) that I can use Stockfish 'Modern'? (I never knew I had a dual-core machine!)

Re engine tests, it sounds as if my small tournaments are virtually useless, even if they are good fun! (In short blitz matches between SF5 and SF5 Modern I got the weird results of 3-1(!), 1.5-2.5, 2-2, 2-2!) Yes, if you could post a test suite or two that would be great -- I imagine other people might be interested too.

I have never played with parameters and almost do not dare! But tell me if there is anything simple (with any engine) I should try changing/tweaking. When I look at my own games I put on Multi PV (to see if that brilliant sacrifice had one refutation, two refutations or no refutation! Cheesy), but otherwise I believe this weakens an engine slightly -- is that true?

  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #92 - 06/13/14 at 13:57:12
Post Tools
Hi, Michael!
As Jury Osipov himself said, Strelka 6 x32 is a decompiled one-core x32 only engine by reversed engineering from Houdini 4 with some minor changes that he mentioned, too. Here comes the question: why use an x32 version of Houdini 4 on one core, when you can use Houdini 4 on four cores, for example?
The GUI you use doesn't count in testing/analysing if you properly put the parameters of your engines.
Hardware is an important factor, yes, because the best engines are usually made for high-end hardware.
Well, your Intel(R) Core(TM) i3 CPU 540 @ 3.07GHz is not a high-end processor, but is still "modern" and you don't have to have big deviations in your engine tests.
The "professional testers" say that if you want to make a valuable test you have to make a big match between two engines in order to see which one is better. They say, that even 5000 games on a fixed time control are not enough and give some high-sophisticated mathematical values which show that even a match from 10000 games can be only 95% reliable.
Anyway, there are several test matches between Strelka 6 x32 and Stockfish 5 that show that Strelka 6 is designed for blitz time control and can hold against Stockfish for 200-300 games, but when you prolong the test you'll see that Stockfish 5 will win convincingly after all. As for the long time controls, Strelka 6 will not perform well against Stockfish 5 if you let the last to use more than one processor's core.
There are several suites of tactical (and other) positions which are used by the testers to test the engines. And I've never been interested in, but if you want, I'll try to supply you with some of them.
I think you can use Stockfish 5 modern version - you can see what kind of instructions (SSE.. AVX and so on..) supports your Intel(R) Core(TM) i3 CPU 540 @ 3.07GHz by downloading a nice tiny program CPU-Z, completely free from the Internet (I think it's a little exe-program that doesn't install on your computer). Of course, if your processor supports these instructions, then the "modern" engines can perform far better on your PC than the "old" ones.
That's it!  Wink
  
Back to top
 
IP Logged
 
Michael Ayton
God Member
*****
Offline


‘You’re never alone with
a doppelgänger.’

Posts: 1947
Location: durham
Joined: 04/19/03
Gender: Male
Re: The best analysis program?
Reply #91 - 06/13/14 at 12:13:37
Post Tools
Interesting thread. I was meaning to write a post here a few weeks ago when … Strelka 6 came out! And then, blow me down, before I’d got round to it Stockfish 5 appeared!

I'm a bit confused, because I understood that an engine’s relative strength might be influenced by (1) the GUI you use it in and (2) the specific characteristics of your PC, but even so, the results of at least one of the engine matches I’ve conducted on my PC seem a bit radical! (I use Arena 3.5 because I have only an older version of ChessBase.) I got these results:

Strelka 6, 4; Houdini 1.5a, nil!! (5 min. games)

Stockfish 5, 2; Strelka 6, 2 (5 min. games)

Stockfish 5, 2.5; Strelka 6, 1.5 (10 min. games; also, same result at 15-min. time limit)

Strelka 6, 4.5
Stockfish 5, 4
Gull 3.0, 3.5
Houdini 1.5a, nil!! (12 x 5 min. games)

The first result seems appreciable since Strelka 6 is after all essentially a 32-bit version of Houdini 4, and maybe one would not expect Stockfish to beat Strelka by a wider margin than it did, but what of the final engine match? What accounts for Houdini flopping as it did? (For info, I’ve included my PC spec and ArenaMark, etc. below.)

On the subject of different engines’ differential analysis and evaluation of the same position, maybe someone expert should devise a test suite of positions geared to allowing ‘style’ (and not just ‘strength’) conclusions to be drawn from this. Anyone know of such a suite? Anyway, in the opening stages of a game at least, rightly or wrongly I tend to trust engines’ actual move suggestions rather more than I do their numerical evaluations, which as we all know often get revised after some subsequent moves are inputted …

PS. Is it best to use Stockfish 5 x64, or Stockfish 5 x64 modern? What is the difference? And what about engines like Sting or DeepSaros that also have a 'pop(cnt)' version, or Fire, that has an 'AVX' and a 'SSE' version, both with or without 'FRC'? -- how is it possible to know which version is best?

*********

Hardware: Intel(R) Core(TM) i3 CPU 540 @ 3.07GHz with 2.7 GB Memory.
Operating system: Windows 7 Home Premium Home Edition (Build 7600) 64 bit.
CPU speed: 3149.69 MHz
ArenaMark: 26.59
CPUEfficiency: 1.27
  
Back to top
 
IP Logged
 
katar
Senior Member
****
Offline


look another year went
by

Posts: 454
Location: LA
Joined: 09/21/05
Gender: Male
Re: The best analysis program?
Reply #90 - 06/11/14 at 16:24:31
Post Tools
Smyslov_Fan wrote on 06/11/14 at 15:47:43:
I'm gonna take a guess about post #83. Regardless of the numeric evaluation, I'm gonna guess that the different engines all agree on the top three candidate moves from that position. Is that true?

I agree with that.
IMO a centipawn is a different "currency" depending on the engine. 
Stockfish may show -1 ; Critter may show -1.5.  Despite the different numbers, I don't think we can say that Critter necessarily gives a greater advantage to Black in that case.  Certain engines may not even be based on the 1.00 = 1 pawn standard.  And that is not even considering the variable relative strength of pawns (e.g., a protected passed pawn on e6 could be worth a rook).

We need a conversion table in order to compare a debt of one Euro (currency) to a debt of two US dollars.  Same with evaluations of two chess engines IMO.

Different engines may also have different thresholds for what is considered a "clear advantage" versus a "decisive advantage."

Of course the real answer to the OP query is, "unless you are a  professional, it doesn't matter, just use one of the top engines," but that sounds cynical.
  

2078 uscf
Back to top
WWW  
IP Logged
 
Smyslov_Fan
YaBB Moderator
Correspondence fan
*****
Offline


Progress depends on the
unreasonable man. ~GBS

Posts: 6902
Joined: 06/15/05
Re: The best analysis program?
Reply #89 - 06/11/14 at 15:47:43
Post Tools
I'm gonna take a guess about post #83. Regardless of the numeric evaluation, I'm gonna guess that the different engines all agree on the top three candidate moves from that position. Is that true?

The advice that the seasoned correspondence players give makes great sense. Since you seem to have access to multiple engines, you don't have to choose just one. Use them all and use your judgement to decide which lines and which evaluations you prefer.
  
Back to top
 
IP Logged
 
Alias
God Member
*****
Offline



Posts: 1509
Location: East of the river Svartĺn
Joined: 11/19/04
Re: The best analysis program?
Reply #88 - 06/11/14 at 05:35:25
Post Tools
As a chess book reader, I sometimes feel that the authors could have suggested moves/plans in positions with unbalanced material. I recall the queen sac line in the Safest sicilian. Playing such positions can be tricky since one does not practise them often.

I've also played CC. (I have 2370 on ICCF I think.) It's of course true that even after much analysis, the lines can be difficult to evaluate, but trying them is of course better than just stopping at the said position. If all positions could be easily evaluated after some computer analysis, we wouldn't have CC.
  

Don't check me with no lightweight stuff.
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #87 - 06/10/14 at 22:14:55
Post Tools
Well, as a correspondence chess player I think I know a thing or two about engines.
And I can speak for hours, but I don't think it's necessary.
Just some hints:
There is no engine which can give you "most accurate evaluations" in infinite analysis mode. You can analyse a position for days reaching 44 plys depth and even more...and still don't have a clue about the real evaluation of the position you analyse. It's all because the so-called "horizon" of the engines. The lines they give may be 'true' up to the 8th-10th ply or so (out of these 44 plys), so leaving them counting for hours is counter-productive. (As for the best engine in the background while browsing through your repertoire, just use Houdini! It's more reliable than Stockfish and doesn't scale too fast while heavy pruning as Stockfish does.. While Komodo is too slow to be reliable for being in the background.)
So, if a long (infinite) analysis doesn't help - then what?
I would suggest two methods: the first one callled Aquarium IDeA and the second one - starting matches (at least 300 games on 5min time control) between engines from the given position.
Aah, and the third one is analysing with a GM behind your left shoulder..  Wink
  
Back to top
 
IP Logged
 
tony37
God Member
*****
Offline


I Love ChessPublishing!

Posts: 619
Joined: 10/16/10
Re: The best analysis program?
Reply #86 - 06/10/14 at 21:35:29
Post Tools
TheDragon wrote on 06/10/14 at 15:39:51:
So much for Larry Kaufman and Komdo's "the other engines overvalue the queen...".

From the brief analysis I have conducted it seems to me more that it is Komodo which undervalues the queen.

Can anyone confirm?

I think it really depends on the position, sometimes all engines overvalue the side with the queen, sometimes they all undervalue the side with the queen
  
Back to top
 
IP Logged
 
TheDragon
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 17
Joined: 04/27/14
Re: The best analysis program?
Reply #85 - 06/10/14 at 15:39:51
Post Tools
So much for Larry Kaufman and Komdo's "the other engines overvalue the queen...".

From the brief analysis I have conducted it seems to me more that it is Komodo which undervalues the queen.

Can anyone confirm?
  
Back to top
 
IP Logged
 
Alias
God Member
*****
Offline



Posts: 1509
Location: East of the river Svartĺn
Joined: 11/19/04
Re: The best analysis program?
Reply #84 - 06/10/14 at 14:34:51
Post Tools
Check the ideas/moves of the different programs. It should become clear after a while which program has evaluated the position correctly.

Never stop at a position! Always check the lines that the programs suggest are the best.
  

Don't check me with no lightweight stuff.
Back to top
 
IP Logged
 
TheDragon
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 17
Joined: 04/27/14
Re: The best analysis program?
Reply #83 - 06/10/14 at 13:56:33
Post Tools
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
*

Hi. Now this is the kind of position which really confuses me. I came across it during my work on the 3.f3 Grunfeld with 3...d5.

Komodo says: Clear Advantage to White.
Houdini says: Very slight Advantage to Black/Equal.
Stockfish says: Very slight Advantage to White/Equal
Larry Kaufman says: White is much better (agrees with Komodo).
Svetushkin says: White has nothing here.
I say: The position looks better and easier to play for white.

The question is: which to believe? Smiley

Regards,
TheDragon
  
Back to top
 
IP Logged
 
Smyslov_Fan
YaBB Moderator
Correspondence fan
*****
Offline


Progress depends on the
unreasonable man. ~GBS

Posts: 6902
Joined: 06/15/05
Re: The best analysis program?
Reply #82 - 06/10/14 at 13:24:30
Post Tools
I think only correspondence chess players would care about the differences because they are so minute.

Both engines are excellent and tend to agree with each other in most positions. The answer you are seeking really depends on the positions. Take a look at the games the two engines played and see if you can come up with patterns that demonstrate strengths or weaknesses in the two programs.

One trait of Stockfish is that it tends to prefer having the Q over various piece combinations more than other engines. In some cases that is a strength, in others that's a weakness.

Analyse the games played between Stockfish and Komodo and see if you can come to conclusions about the types of positions each engine won.

  
Back to top
 
IP Logged
 
TheDragon
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 17
Joined: 04/27/14
Re: The best analysis program?
Reply #81 - 06/10/14 at 13:04:15
Post Tools
Vass wrote on 06/09/14 at 22:18:26:
Well, responding to your first question:
Yes, 1.Stockfish; 2-3 Komodo/Houdini (or Houdini/Komodo depending on the time control).
For 5-10 minutes on the key positions and 30sec-1min for the rest...you can use Houdini. I think Stockfish & Komodo are better than Houdini when using more than 30sec-1min for a position. (Stockfish is better than Houdini even on blitz time controls, but this is all because of its fast scaling and pruning.)


Thanks for the reply. The program I am using allows me to do multiple analysis (so I am going to go into 5-10 minutes for leafnodes and + - a minute for the rest).

If you had to choose, which one is it for the long analysis? I am looking for the most accurate evaluations. Komodo or Stockfish? Tongue Cheesy

Sorry, I know I am pushing quite a bit but anyway.

P.S. If I am going through my repertoire/game with an engine in the background, would you suggest Stockfish?


Thanks and Regards,
TheDragon
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #80 - 06/09/14 at 22:18:26
Post Tools
Well, responding to your first question:
Yes, 1.Stockfish; 2-3 Komodo/Houdini (or Houdini/Komodo depending on the time control).
For 5-10 minutes on the key positions and 30sec-1min for the rest...you can use Houdini. I think Stockfish & Komodo are better than Houdini when using more than 30sec-1min for a position. (Stockfish is better than Houdini even on blitz time controls, but this is all because of its fast scaling and pruning.)
  
Back to top
 
IP Logged
 
TheDragon
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 17
Joined: 04/27/14
Re: The best analysis program?
Reply #79 - 06/09/14 at 12:12:58
Post Tools
Hi there.

So it seems that Stockfish 5 is the strongest playing engine. Followed by Houdini and then Komodo. Am I right?

The question is which to use for analysis and to analyse my repertoire. Larry Kaufman (Komodo developer) says that Komodo is best for long-term analysis and more positionally focused (with more time spent on evaluations). Houdini is apparently better when it comes to blitz and quick tactics. And Stockfish, well, I don't know.

Basically I don't know which one to use to analyse my repertoire. The engine will spend around 5-10 minutes on the key positions each and maybe 30 seconds - 1 minute for the rest (I am using Chess Position Trainer 5, CPT, by the way).

So basically, I don't know which to use. Should I use a combination, and if so, which one should take preference?

Regards,
TheDragon
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #78 - 06/04/14 at 13:57:06
Post Tools
A couple of days ago Stockfish 5 was released.
A free chess engine which is definitely better than the last versions of Houdini, Komodo and alike.. Several independent tests at different time controls prove it. And what is more, this engine won the last TCEC - the unofficial world chess engines' championship on 190' + 30" time control.
http://stockfishchess.org/
As for the best analysis program, well...I don't know.  Wink
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #77 - 12/14/13 at 20:24:15
Post Tools
tony37 wrote on 12/14/13 at 18:40:16:
I think the real curse is that there will be even more draws at the top level and that just isn't very interesting

I'm not sure about it.
Three different almost equal in strength engines can mess things up.
Besides, draws at the top correspondence chess level happen because top players do not want to take a big risk. They usually play some checked lines in the well known openings - such as Sicilian Najdorf, Ruy Lopez or Semi-Slav, Catalan and alike. In fact, no one wants to play King's Indian, for example, when meeting another top player.  Roll Eyes
  
Back to top
 
IP Logged
 
tony37
God Member
*****
Offline


I Love ChessPublishing!

Posts: 619
Joined: 10/16/10
Re: The best analysis program?
Reply #76 - 12/14/13 at 18:40:16
Post Tools
I think the real curse is that there will be even more draws at the top level and that just isn't very interesting
  
Back to top
 
IP Logged
 
tony37
God Member
*****
Offline


I Love ChessPublishing!

Posts: 619
Joined: 10/16/10
Re: The best analysis program?
Reply #75 - 12/12/13 at 19:33:24
Post Tools
brabo wrote on 12/12/13 at 19:26:08:
Vass wrote on 12/12/13 at 13:57:36:
Several tests recently show that Houdini 4 has some big competitors for the title "Best Analysis Program" in the face of its rivals: last versions of Komodo and Stockfish. While at short time control matches Houdini 4 keeps scaling better and wins over the last mentioned, they seem to prevail in long time control matches.
A true heaven for correspondence chess players! They can use several engines depending on the position they analyse and can choose between three different evaluations in order to choose the best move.  Wink

At the same time it is also a curse. Now you have to make a choice which is for the operator (in most cases 1000 or more points weaker than the engines) often just guessing.

??
it is still allowed to actually analyse these moves
  
Back to top
 
IP Logged
 
brabo
God Member
*****
Offline


Welcome chessfriend

Posts: 1064
Joined: 02/02/07
Re: The best analysis program?
Reply #74 - 12/12/13 at 19:26:08
Post Tools
Vass wrote on 12/12/13 at 13:57:36:
Several tests recently show that Houdini 4 has some big competitors for the title "Best Analysis Program" in the face of its rivals: last versions of Komodo and Stockfish. While at short time control matches Houdini 4 keeps scaling better and wins over the last mentioned, they seem to prevail in long time control matches.
A true heaven for correspondence chess players! They can use several engines depending on the position they analyse and can choose between three different evaluations in order to choose the best move.  Wink

At the same time it is also a curse. Now you have to make a choice which is for the operator (in most cases 1000 or more points weaker than the engines) often just guessing.

Just one other remark about to choose between the different evaluations. It is well possible that the 3 engines show different evaluations but still show the same move and even recommended first line (at least a number of moves). I want to say that different evaluations don't always mean different moves.
  
Back to top
WWW  
IP Logged
 
Matemax
God Member
*****
Offline


Chesspub gives you strength!

Posts: 1302
Joined: 11/04/07
Re: The best analysis program?
Reply #73 - 12/12/13 at 15:19:12
Post Tools
Vass wrote on 12/12/13 at 13:57:36:
Several tests recently show that Houdini 4 has some big competitors for the title "Best Analysis Program" in the face of its rivals: last versions of Komodo and Stockfish. While at short time control matches Houdini 4 keeps scaling better and wins over the last mentioned, they seem to prevail in long time control matches.
A true heaven for correspondence chess players! They can use several engines depending on the position they analyse and can choose between three different evaluations in order to choose the best move.  Wink

Playing correspondence myself I still have fun using my own engine (in ancient times called "brain"), too.
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #72 - 12/12/13 at 13:57:36
Post Tools
Several tests recently show that Houdini 4 has some big competitors for the title "Best Analysis Program" in the face of its rivals: last versions of Komodo and Stockfish. While at short time control matches Houdini 4 keeps scaling better and wins over the last mentioned, they seem to prevail in long time control matches.
A true heaven for correspondence chess players! They can use several engines depending on the position they analyse and can choose between three different evaluations in order to choose the best move.  Wink
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #71 - 09/01/13 at 14:39:38
Post Tools
Sandman wrote on 09/01/13 at 13:25:54:
It seems to me chess engines spit out the same moves as all the others. ...


Well, it depends on the position you analyse.  Roll Eyes

Sandman wrote on 09/01/13 at 13:25:54:
... Are there any good engines that give different analysis possibilities or standout by giving unique choices?


The problem is that they all are calculators. We don't have to forget that. And the main things that usually differ in chess engines are the speed and the evaluation function. The faster the better...as we all know, while the evaluation function makes the big difference between them. So among the best ones, Houdini, Stockfish, Komodo and Rybka have completely different evaluation algorithms. Well enough...to not complain, I think.  Wink
  
Back to top
 
IP Logged
 
Sandman
Full Member
***
Offline



Posts: 134
Joined: 05/10/07
Gender: Male
Re: The best analysis program?
Reply #70 - 09/01/13 at 13:25:54
Post Tools
It seems to me chess engines spit out the same moves as all the others. Are there any good engines that give different analysis possibilities or standout by giving unique choices?
  

“All it takes is one bad day to reduce the sanest man alive to lunacy.
That's how far the world is from where I am.
Just one bad day.”
Back to top
 
IP Logged
 
scchess
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 16
Joined: 08/09/13
Re: The best analysis program?
Reply #69 - 08/09/13 at 08:26:20
Post Tools
On the mobile platform, I think SmallFish (Stockfish) is your best friend.
  
Back to top
 
IP Logged
 
Master Om
Full Member
***
Offline


I Love ChessPublishing!

Posts: 187
Joined: 02/20/10
Re: The best analysis program?
Reply #68 - 06/26/13 at 02:55:23
Post Tools
Still Komodo CCT is strongest and better than Komodo 5.1 at single core. It is too poor at tactics. hence cannot be recommended as a general Analysis tool. Still Houdini 3 Tactical is best. But as 3rd opinion is still good.
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #67 - 06/17/13 at 21:22:24
Post Tools
A brand new candidate for the best analysis engine:

Komodo 5.1 multi-processor (commercial) - at last  Grin
Released 17 of June, 2013
http://komodochess.com/komodo51mp-32a

Roll Eyes
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #66 - 05/10/13 at 11:13:25
Post Tools
Stockfish 3: http://stockfishchess.org/download/ (free) and
Komodo CCT: http://komodochess.com/komodoCCT-31a (commercial)
are the new candidates for the title: "The Best Analysis Program".  Wink

IPON Rating list (ponder on):
http://www.inwoba.de/index.html (Latest update: 2013.05.05)
  
Back to top
 
IP Logged
 
Justin John Hay
YaBB Newbies
*
Offline


I Love ChessPublishing!

Posts: 4
Joined: 02/09/13
Re: The best analysis program?
Reply #65 - 02/09/13 at 18:25:33
Post Tools
Houdini would be most peeps top vote
  
Back to top
 
IP Logged
 
fling
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1591
Joined: 01/21/11
Gender: Male
Re: The best analysis program?
Reply #64 - 01/09/13 at 11:50:53
Post Tools
Thanks Vass! I have the free ones, I guess since I want to try Aquarium, I'll go for it with Houdini 3. As for the rest of the commercial ones, I think I have enough anyway!  Cheesy
  
Back to top
 
IP Logged
 
GMTonyKosten
YaBB Administrator
*****
Offline


Mr Dynamic?

Posts: 3149
Location: Clermont-Ferrand
Joined: 12/19/02
Gender: Male
Re: The best analysis program?
Reply #63 - 01/09/13 at 10:26:29
Post Tools
Markovich wrote on 01/09/13 at 03:04:31:
Well maybe someone will correct me, but White's win in #54 looks to me like something that most strong players would consider.

Maybe, you would have to see a3 (threatening to trap the queen) to force the queen back, and then Qe1, and then evaluate the position correctly.
However, like most tactical positions it is a lot easier to see when you know there is a tactic to be found!
  
Back to top
IP Logged
 
Girkassa
Senior Member
****
Offline


I Love ChessPublishing!

Posts: 433
Joined: 04/07/07
Re: The best analysis program?
Reply #62 - 01/09/13 at 10:08:21
Post Tools
Markovich wrote on 01/09/13 at 03:04:31:
Well maybe someone will correct me, but White's win in #54 looks to me like something that most strong players would consider. Not having Houdini 3.0 turned on is hardly an excuse for not spotting that particular sequence, I opine.


I agree. It is a natural sacrifice, and while I think few players would be able to determine it was winning OTB, I think many strong players would seriously consider it.

Also, checking my Rybka 4, after 13.Rxe7 c6 it needs about 3 seconds to put 14.Rxg7+ as its first choice. Although it initially evaluates it as equal, just seeing that move on top of the list would have made me nervous if I was intending to recommend this for Black.
  
Back to top
 
IP Logged
 
Michael Ayton
God Member
*****
Offline


‘You’re never alone with
a doppelgänger.’

Posts: 1947
Location: durham
Joined: 04/19/03
Gender: Male
Re: The best analysis program?
Reply #61 - 01/09/13 at 09:01:11
Post Tools
Quote:
Michael Ayton wrote on 20. Dec 2012 at 14:30:
That's impressive! My humble Houdini 1a takes fully 5.75 seconds!

Why don't you download the free 1.5a, it's pretty good?


Sorry Tony, that's the one I meant! My PC is pretty fast, so I guess we can conclude H3 is significantly better ...

  
Back to top
 
IP Logged
 
Markovich
God Member
*****
Offline



Posts: 6099
Location: Columbus, Ohio
Joined: 09/17/04
Re: The best analysis program?
Reply #60 - 01/09/13 at 03:04:31
Post Tools
Well maybe someone will correct me, but White's win in #54 looks to me like something that most strong players would consider. Not having Houdini 3.0 turned on is hardly an excuse for not spotting that particular sequence, I opine.
  

The Great Oz has spoken!
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #59 - 01/08/13 at 23:30:37
Post Tools
fling wrote on 01/08/13 at 22:20:24:
So, if I were to choose only one engine, I take it that you would recommend Houdini 3. And if I have a choice of a second one, which would it be, Rybka 4? Something else? Deep Hiarcs?


Well, let's see!

One engine? - Houdini 3 (commercial).
Two engines? - ... + Stockfish 2.3.1 (free) - sometimes better in closed positions..
Three engines? - ... + Houdini 1.5a (free) - more reliable for an analysis in depth.
Four engines? - ... + Critter 1.6a (free) - sometimes better in endgames' analysis and the second best after Houdini, in fact.
Five engines? - ... + Komodo 5 (commercial) - sometimes giving some good analysis lines (ideas), different from others, because of its unique method of evaluation.
Six engines? - ... + Ivanhoe B46 (free) - just like Komodo 5.
Seven engines? - ... + Deep Rybka 4.1 (commercial) - because of pure nostalgia, maybe.. Or, just because it's good.
Eight engines? - ... + Zappa Mexico II (commercial) - because of its unique "dissident aggressor" options for solving tactical positions that Houdini 3 "tactical mode" won't solve.
Nine engines? - ... + Critter 0.90 (free) - good for some endgame analysis and closed position analysis, too.
Ten engines? - ... + Deep Fritz 10 (commercial) - once it was a great program, you know. And the most "human" engine I have ever seen..  Smiley

But, let me stop here! Otherwise I would be tempted to mention Deep Hiarcs 14, Vitruvius, Fire 2.2 xTreme and so on...but for no reason.  Undecided
  
Back to top
 
IP Logged
 
fling
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1591
Joined: 01/21/11
Gender: Male
Re: The best analysis program?
Reply #58 - 01/08/13 at 22:20:24
Post Tools
So, if I were to choose only one engine, I take it that you would recommend Houdini 3. And if I have a choice of a second one, which would it be, Rybka 4? Something else? Deep Hiarcs?
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #57 - 12/21/12 at 15:40:23
Post Tools
By the way, one big improvement in Houdini 3 is the so-called "tactical mode".
What it means?
It means that one can prepare Houdini 3 for highly tactical positions just by checking the option "tactical mode" in the engine parameters menu before doing an analysis.
According to some testers, it solves tactical studies up to 30 per cents better (and faster) than the Houdini 3 in classical mode. These results rank the engine among the best ones in tactics.
I often try this option in my correspondence chess tries. And I can say I'm very impressed by the results, too..  Wink
Meanwhile, some testers ran matches between Houdini 3 and Houdini 3 "tactical mode". What is amazing is that Houdini 3 "tactical mode" is just a bit weaker in playing strength. Usually, engines don't play very well in tactical mode, because they're somehow "twisted" in parameters. Not Houdini 3 "tactical mode" though. Still it can win a match against any of the best engines available.  Cheesy
  
Back to top
 
IP Logged
 
GMTonyKosten
YaBB Administrator
*****
Offline


Mr Dynamic?

Posts: 3149
Location: Clermont-Ferrand
Joined: 12/19/02
Gender: Male
Re: The best analysis program?
Reply #56 - 12/21/12 at 12:47:53
Post Tools
Michael Ayton wrote on 12/20/12 at 14:30:31:
That's impressive! My humble Houdini 1a takes fully 5.75 seconds!

Why don't you download the free 1.5a, it's pretty good?
  
Back to top
IP Logged
 
Michael Ayton
God Member
*****
Offline


‘You’re never alone with
a doppelgänger.’

Posts: 1947
Location: durham
Joined: 04/19/03
Gender: Male
Re: The best analysis program?
Reply #55 - 12/20/12 at 14:30:31
Post Tools
That's impressive! My humble Houdini 1a takes fully 5.75 seconds!
  
Back to top
 
IP Logged
 
GMTonyKosten
YaBB Administrator
*****
Offline


Mr Dynamic?

Posts: 3149
Location: Clermont-Ferrand
Joined: 12/19/02
Gender: Male
Re: The best analysis program?
Reply #54 - 12/20/12 at 13:26:48
Post Tools
In Glenn's December update (on the Dutch) he originally said that 12...d5 might be a good solution to Black's problems in the 7...Qe8 8 Re1 Leningrad:

* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
* * * * * * * *
*
Apparently he was using Rybka4, which likes Black's position. However, a subscriber pointed out that White has almost a forced win so Glenn had to quickly change his analysis!
Indeed, when I checked with Houdini3 it gives 13 Rxe7 c6 14 Rxg7+ as winning in less than a second!
  
Back to top
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #53 - 12/17/12 at 22:21:52
Post Tools
GMTonyKosten wrote on 12/17/12 at 21:20:34:
I just looked at a position I had already analysed in my new book (DW-Spanish) with Houdini 3, and to my surprise it didn't find the draw that Houdini 1.5 had discovered! Roll Eyes


I'm not surprised at all..
Maybe, as a correspondence chess player, I'm addicted too much to my beloved engine ...but, if Houdini 1.5a is as faster as Houdini 3 and can reach the same ply's depth for the same amount of time...I would never even try to use the new toy called Houdini 3.  Wink
The problem is that a good analysis is taking time, too much time.. One too many sleepless nights..
And I have my own conclusions I can share. For example, I never believe Houdini 3's evaluation in a complex position if not reached at least 27-28th ply's depth.. While 1.5a version's evaluation on 26th ply's depth is almost definitive. The only problem I find in Houdini 1.5a is that its style of play is not as dynamic as Houdini 3's one. Hence the difference in ELO points..  Roll Eyes
  
Back to top
 
IP Logged
 
GMTonyKosten
YaBB Administrator
*****
Offline


Mr Dynamic?

Posts: 3149
Location: Clermont-Ferrand
Joined: 12/19/02
Gender: Male
Re: The best analysis program?
Reply #52 - 12/17/12 at 21:20:34
Post Tools
I just looked at a position I had already analysed in my new book (DW-Spanish) with Houdini 3, and to my surprise it didn't find the draw that Houdini 1.5 had discovered! Roll Eyes
  
Back to top
IP Logged
 
Master Om
Full Member
***
Offline


I Love ChessPublishing!

Posts: 187
Joined: 02/20/10
Re: The best analysis program?
Reply #51 - 12/05/12 at 03:07:14
Post Tools
GMTonyKosten wrote on 12/01/12 at 14:09:29:
Vass wrote on 12/01/12 at 14:06:31:
H3 is better, faster...and so on..

OK, thanks. Smiley



Houdini is not better always.
  
Back to top
 
IP Logged
 
GMTonyKosten
YaBB Administrator
*****
Offline


Mr Dynamic?

Posts: 3149
Location: Clermont-Ferrand
Joined: 12/19/02
Gender: Male
Re: The best analysis program?
Reply #50 - 12/01/12 at 14:09:29
Post Tools
Vass wrote on 12/01/12 at 14:06:31:
H3 is better, faster...and so on..

OK, thanks. Smiley
  
Back to top
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #49 - 12/01/12 at 14:06:31
Post Tools
GMTonyKosten wrote on 12/01/12 at 12:14:50:
How does Deep Hiarcs 14 rate against Houdini 3?

You can see the rating lists from the most respectable testing groups here:
http://www.computerchess.org.uk/ccrl/
http://www.husvankempen.de/nunn/
http://inwoba.de/index.html

...and a little private blitz test under the following (hardware/software/settings) conditions:

Intel Core i7-3930K 3.20Ghz @ 4.77 GHz 16.0 GB RAM
Windows 7 Professional SP 1
Fritz Benchmark:
Speed: 42.96
KNS: 20619
GUI: CB-Fritz 13
Book: Perfect 2012 - 7 moves
Hash:128
GRT-RB-TB: ON
Ponder: OFF
Blitz 5m 0

...finished as listed here:
1   Houdini 3 Pro x64    +308  +75/=21/-4 85.50%   85.5/100
2   Deep HIARCS 14      -308  +4/=21/-75 14.50%   14.5/100

...and the games can be downloaded from:
http://www.mediafire.com/?b9nimawf8nigrgz

The results speak for themselves.. H3 is better, faster...and so on.. As for the analysis purpose (and here we go again) I think the margin is not such wide.. Almost every high level engine can do a proper analysis with a little help of the device that is controlling the mouse buttons..  Wink
  
Back to top
 
IP Logged
 
GMTonyKosten
YaBB Administrator
*****
Offline


Mr Dynamic?

Posts: 3149
Location: Clermont-Ferrand
Joined: 12/19/02
Gender: Male
Re: The best analysis program?
Reply #48 - 12/01/12 at 12:14:50
Post Tools
How does Deep Hiarcs 14 rate against Houdini 3?
  
Back to top
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #47 - 09/28/12 at 06:43:13
Post Tools
Robert Houdart:
"The 3 matches have now finished.

Houdini 3 - Komodo 5 match: +43 -19 =58
72-48 (+68 Elo ą 42 Elo).

Houdini 3 - Stockfish 2.3.1 match: +44 -16 =60
74-46 (+82 Elo ą 42 Elo).

Houdini 3 - Houdini 2.0c match: +48 -15 =57
76.5-43.5 (+94 Elo ą 42 Elo).

Download All the Games (single zip containing the 3 matches) at: http://www.cruxis.com/download/Houdini3_LongTC_Matches.zip

Over-all result: Houdini 3 scored 61.8% (+81 Elo ą 24 Elo) against the average of Houdini 2.0c, Komodo 5 and Stockfish 2.3.1.
The individual match results have very large confidence intervals, please consider these when discussing results."

Lips Sealed
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #46 - 09/27/12 at 20:34:21
Post Tools
Another pre-release test at 90mins time control of Houdini 3, made by the author, Robert Houdart:
"The second match just finished.
Final result of the Houdini 3 - Stockfish 2.3.1 match: +44 -16 =60
74-46 (+82 Elo ą 42 Elo).
Download Games at http://www.cruxis.com/download/Houdini3_Stockfish231_90min.zip "
Cheesy
More info at: http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=25731;pg=3

That's it!  Cool
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #45 - 09/26/12 at 13:46:40
Post Tools
tony37 wrote on 09/26/12 at 13:08:45:
Vass wrote on 09/26/12 at 08:53:59:
Houdini 3
Robert Houdart:
"The engine development is completely finished, but there's still a lot of work in packaging and documentation (e.g. write the User's Guide). Hopefully around October 10.
Some info about it: http://talkchess.com/forum/viewtopic.php?t=45303&start=10

Mr Houdart ran some tests on long time controls (90 min+30 sec/move starting from 60 Noomen suite positions, one core each)
http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=25731
Very impressive results:
Final result of the Houdini 3 - Houdini 2.0c match: +48 -15 =57
You can download all of the games from http://www.cruxis.com/download/Houdini3_Houdini20c_90min.zip

Well, obviously that's the engine for my future analysis!  Wink
I'll post some of these games here - hope you don't mind!

I wonder if they've fixed the queen vs. minor pieces problem where Houdini typically overrates the side with the queen (I have version 1.5 so I don't know if this was fixed in version 2 already)

It was fixed in 2.0 version.  Smiley
  
Back to top
 
IP Logged
 
tony37
God Member
*****
Offline


I Love ChessPublishing!

Posts: 619
Joined: 10/16/10
Re: The best analysis program?
Reply #44 - 09/26/12 at 13:08:45
Post Tools
Vass wrote on 09/26/12 at 08:53:59:
Houdini 3
Robert Houdart:
"The engine development is completely finished, but there's still a lot of work in packaging and documentation (e.g. write the User's Guide). Hopefully around October 10.
Some info about it: http://talkchess.com/forum/viewtopic.php?t=45303&start=10

Mr Houdart ran some tests on long time controls (90 min+30 sec/move starting from 60 Noomen suite positions, one core each)
http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=25731
Very impressive results:
Final result of the Houdini 3 - Houdini 2.0c match: +48 -15 =57
You can download all of the games from http://www.cruxis.com/download/Houdini3_Houdini20c_90min.zip

Well, obviously that's the engine for my future analysis!  Wink
I'll post some of these games here - hope you don't mind!

I wonder if they've fixed the queen vs. minor pieces problem where Houdini typically overrates the side with the queen (I have version 1.5 so I don't know if this was fixed in version 2 already)
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #43 - 09/26/12 at 09:45:00
Post Tools
Enough with it!

I just don't know what to say..

I've seen many computer games, but never seen such a complete control all over the board. Running on one core each and achieving 19-24 plys depth is obviously not enough for general conclusions, but still..
H3 lost some games, too.. But most of them because of some inferior book moves that were given to it. Of course, H2.0c played under the same conditions and was fed by some inferior book moves, too..
But, the results are very, very impressive!  Shocked
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #42 - 09/26/12 at 09:35:18
Post Tools
And now Queen's Indian...with both colours:



  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #41 - 09/26/12 at 09:31:00
Post Tools
How to play Nimzo indian?
Let's see:



  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #40 - 09/26/12 at 09:16:43
Post Tools
IsaVulpes wrote on 09/26/12 at 09:11:09:
Mildly related, is there a commonly seen as "best" GUI? So far I've always just loaded engines in whatever Fritz version I happened to own, but I always thought of getting a dedicated, better GUI that comes without an Engine on its own.
Are you guys using anything specific?

E: Is this even halfway the right place or should I open a new thread for this or something

It depends on what you particularly want from the GUI.
Yes, you have to open a new thread here...and tell us what you want.  Wink
  
Back to top
 
IP Logged
 
Vass
God Member
*****
Offline


I Love ChessPublishing!

Posts: 1105
Joined: 06/22/11
Re: The best analysis program?
Reply #39 - 09/26/12 at 09:13:11
Post Tools
Some say Gruenfeld is one of the best openings for black..
Mhmm, maybe...but not for H3!