Winter 0.6 has been released. I am expecting 50 Elo points on one thread and a bit more on more threads.
I might go a bit more in depth at some point, but that was my intention for the last two releases as well, so I'm not making any promises. If I do go into more detail, I will likely make a thread for this version + the previous two.
As was the case last time, my Windows compiles are cross compiles from Linux and I did only a brief confirmation that the binaries are working with WINE. If anyone encounters issues, please contact me as soon as possible, thank you in advance.
jorose wrote: ↑Tue May 28, 2019 12:18 am
Winter 0.6 has been released. I am expecting 50 Elo points on one thread and a bit more on more threads.
I might go a bit more in depth at some point, but that was my intention for the last two releases as well, so I'm not making any promises. If I do go into more detail, I will likely make a thread for this version + the previous two.
As was the case last time, my Windows compiles are cross compiles from Linux and I did only a brief confirmation that the binaries are working with WINE. If anyone encounters issues, please contact me as soon as possible, thank you in advance.
jorose wrote: ↑Tue May 28, 2019 12:18 am
Winter 0.6 has been released. I am expecting 50 Elo points on one thread and a bit more on more threads.
Thanks Jonathan. Unfortunately my first results do not support your optimism, Winter being at approximately 30 % against an average 2980 opposing field. Or I have misunderstood your estimation.
BTW, the files are not exes, I renamed them and run the new CPU one. Is everything all right there?
This version fixes a nasty bug, and there are a few LMR improvements. In self play, it's just over +20 ELO vs 0.23, so may be a little less versus other engines.
jorose wrote: ↑Tue May 28, 2019 12:18 am
Winter 0.6 has been released. I am expecting 50 Elo points on one thread and a bit more on more threads.
Thanks Jonathan. Unfortunately my first results do not support your optimism, Winter being at approximately 30 % against an average 2980 opposing field. Or I have misunderstood your estimation.
BTW, the files are not exes, I renamed them and run the new CPU one. Is everything all right there?
Thank you so much for the reply and feedback!
That is troubling. The extension is a minor issue which I won't fix for now as it is purely cosmetic and much lower priority.
I am assuming this is on 1 core at 4/40 time control? How is the speed compared to vs 0.5? I assume there were no losses on time? What GUI are you using?
When I get home I will run a gauntlet between Winter 0.6 and the previous versions, but if it is a compile specific issue this will be tricky, as I don't have access to any Windows machines at all at the moment. I might also run an old SF version against 0.5 and 0.6, if there is a discrepancy then I will be able to do a binary search over the intermediate versions.
Its a good thing you caught this now, it means I will hopefully be able to find the issue before the TCEC season starts!