Re: Eman 6.50 Versionshinweise (17. Okt. 2020)
Unread post by KhalidOmar » Sat Nov 21, 2020 4:11 pm
....
....
....
I'm no longer distributing licenses. Starting from version 6.60 it will be open for everyone without a license. [Since its StockFish based he is required to also include the SOURCE or a pointer to.]
Has Eman 6.60 appeared or will it appear? ...
On Khalid's page he writes:
"Download
Current Stable Version: 5.60
Find the author on InfinityChess
or Infinitychess Users Forum "
Re: Eman 6.50 Versionshinweise (17. Okt. 2020)
Unread post by KhalidOmar » Sat Nov 21, 2020 4:11 pm
....
....
....
I'm no longer distributing licenses. Starting from version 6.60 it will be open for everyone without a license. [Since its StockFish based he is required to also include the SOURCE or a pointer to.]
Has Eman 6.60 appeared or will it appear? ...
On Khalid's page he writes:
"Download
Current Stable Version: 5.60
Find the author on InfinityChess
or Infinitychess Users Forum "
PS: If anyone has Eman 6.60 to post it here.
The current stable version is 6.52.
I usually visit his website daily - I'll upload here when it appears.
Its now available but regrettably Khalid has not uploaded the SOURCE so in keeping with the Forum Rules I will NOT be uploading the binary here.
My regrets .. rules are rules.
A reminder : please .. do NOT upload the binary or links to it here without HIS source .. you will face a PERMANENT ban.
Let's hope it uploads on the site ...
I personally really liked Eman - and I played tens of thousands of games with him - which I also sent to Khalid.
Unfortunately - he decided to make Eman work only if you are online - which affected me, because for me the connection with the net is very unstable and so you can't do matches, tests, etc.
So from my point of view it was an unfortunate decision by Khalid ...
Speaking of unfortunate decisions:
1 - I also consider that there were also the decisions in the case of BrainLearn and ShashChess to give up the "MCTS" option.
2 - ChessMan's give up of the "Tactical" option in the case of CF_EXT.
But it's their job and I do what I want with it ...
3 - At the moment I really like Massimiliano's SF Poly.
If he introduced the "Learn" and "Full threads" option, it would be perfect from my point of view.
About "Full threads". I suggested this option to Marco in 2019 in the summer.
Marco liked it and then implemented it on all his engines.
It has also been implemented in Raubfisch and ShashChess since then.
unsigned int max_threads = std::thread::hardware_concurrency();
...
o["Threads"] << Option(max_threads, 1, 512, on_threads); // sets the maximum number of threads as default
About the learning, I'm not sure on how Stockfish uses the nnue networks but there could be a possibility that it's in reading-writing mode. If that's the case, the learning function is already implemented within the nnue usage.
ZamChess wrote:About "Full threads". I suggested this option to Marco in 2019 in the summer.
Marco liked it and then implemented it on all his engines.
It has also been implemented in Raubfisch and ShashChess since then.[/color][/glow]
unsigned int max_threads = std::thread::hardware_concurrency();
...
o["Threads"] << Option(max_threads, 1, 512, on_threads); // sets the maximum number of threads as default
About the learning, I'm not sure on how Stockfish uses the nnue networks but there could be a possibility that it's in reading-writing mode. If that's the case, the learning function is already implemented within the nnue usage.
ZamChess wrote:About "Full threads". I suggested this option to Marco in 2019 in the summer.
Marco liked it and then implemented it on all his engines.
It has also been implemented in Raubfisch and ShashChess since then.[/color][/glow]
Oh, you misunderstood both two subjects I guess. Full Search Threads option sets threads for brute-force, for example if you set threads to 2 and set fullsearch threads to 1 then one thread searchs normally while one thread searchs without pruning.
Learn doesn't mean "changing NNUE evals" which SF and others cannot. Learn stands for BrainLEARN, position learning.
Best Regards...
unsigned int max_threads = std::thread::hardware_concurrency();
...
o["Threads"] << Option(max_threads, 1, 512, on_threads); // sets the maximum number of threads as default
About the learning, I'm not sure on how Stockfish uses the nnue networks but there could be a possibility that it's in reading-writing mode. If that's the case, the learning function is already implemented within the nnue usage.
ZamChess wrote:About "Full threads". I suggested this option to Marco in 2019 in the summer.
Marco liked it and then implemented it on all his engines.
It has also been implemented in Raubfisch and ShashChess since then.[/color][/glow]
Oh, you misunderstood both two subjects I guess. Full Search Threads option sets threads for brute-force, for example if you set threads to 2 and set fullsearch threads to 1 then one thread searchs normally while one thread searchs without pruning.
Learn doesn't mean "changing NNUE evals" which SF and others cannot. Learn stands for BrainLEARN, position learning.
Best Regards...
unsigned int max_threads = std::thread::hardware_concurrency();
...
o["Threads"] << Option(max_threads, 1, 512, on_threads); // sets the maximum number of threads as default
About the learning, I'm not sure on how Stockfish uses the nnue networks but there could be a possibility that it's in reading-writing mode. If that's the case, the learning function is already implemented within the nnue usage.
ZamChess wrote:About "Full threads". I suggested this option to Marco in 2019 in the summer.
Marco liked it and then implemented it on all his engines.
It has also been implemented in Raubfisch and ShashChess since then.[/color][/glow]
Oh, you misunderstood both two subjects I guess. Full Search Threads option sets threads for brute-force, for example if you set threads to 2 and set fullsearch threads to 1 then one thread searchs normally while one thread searchs without pruning.
Learn doesn't mean "changing NNUE evals" which SF and others cannot. Learn stands for BrainLEARN, position learning.
Best Regards...
Yes, you are right moonstonelight.
It can be used both in games and especially in analysis.
How was SF Mate Search before ...
Here you have 2 in 1.
Make Someone Happy Today...
