A technical question: Are these tables single-sided or double-sided?
And a follow-up in case of single-sided: What is the expected performance when using them on a, say, 8-16GB, 6-8 core system, with optimised probe depth?
Marcel, you have asked the same damn question months ago when we debated about 'super syzygy bitbases'. Now that karma did the job with that awful Stockfish-syzygy 16-core TCEC I won't say no more. That was hilarious btw. I will not pretend that your questions here are honest so better take a hike
A technical question: Are these tables single-sided or double-sided?
And a follow-up in case of single-sided: What is the expected performance when using them on a, say, 8-16GB, 6-8 core system, with optimised probe depth?
Marcel, you have asked the same damn question months ago when we debated about 'super syzygy bitbases'. Now that karma did the job with that awful Stockfish-syzygy 16-core TCEC I won't say no more. That was hilarious btw. I will not pretend that your questions here are honest so better take a hike
At that time the performance was unknown to you. Besides, I was asking about probe time at the time. Now I ask about strength, which seems completely fair to me. So, is the performance unknown to you, or do you want to keep the results for yourself? Or do we need to wait for somebody else to ask the question? I will adopt them if they provide an advantage over the Syzygy bases.
Angrim wrote:Also they are only bitbases(won/loss/draw) so they fit much less information into the reduced space that they take up.
And half-sided.
Yes they also don't solve chess unlike some recent claims of EGTBs... Why don't you go back to your over-hyped syzg threads and improve on that embarassing 160G figure?
As you very well know, it makes no sense at all to compare your 48.1 GB figure with my 150.1 GB figure.
If you insist too much I'll generate single-sided plain WDL myself and then we can compare...
A technical question: Are these tables single-sided or double-sided?
And a follow-up in case of single-sided: What is the expected performance when using them on a, say, 8-16GB, 6-8 core system, with optimised probe depth?
Marcel, you have asked the same damn question months ago when we debated about 'super syzygy bitbases'. Now that karma did the job with that awful Stockfish-syzygy 16-core TCEC I won't say no more. That was hilarious btw. I will not pretend that your questions here are honest so better take a hike
At that time the performance was unknown to you. Besides, I was asking about probe time at the time. Now I ask about strength, which seems completely fair to me. So, is the performance unknown to you, or do you want to keep the results for yourself? Or do we need to wait for somebody else to ask the question? I will adopt them if they provide an advantage over the Syzygy bases.
What performance results? I didn't do any tests, only one I know who did that is Kai, and he concluded all bitbases perform equally more or less. This is not the thread to discuss syzygy issues but butthurt fanboys keep popping up ...
A technical question: Are these tables single-sided or double-sided?
And a follow-up in case of single-sided: What is the expected performance when using them on a, say, 8-16GB, 6-8 core system, with optimised probe depth?
Marcel, you have asked the same damn question months ago when we debated about 'super syzygy bitbases'. Now that karma did the job with that awful Stockfish-syzygy 16-core TCEC I won't say no more. That was hilarious btw. I will not pretend that your questions here are honest so better take a hike
At that time the performance was unknown to you. Besides, I was asking about probe time at the time. Now I ask about strength, which seems completely fair to me. So, is the performance unknown to you, or do you want to keep the results for yourself? Or do we need to wait for somebody else to ask the question? I will adopt them if they provide an advantage over the Syzygy bases.
What performance results? I didn't do any tests, only one I know who did that is Kai, and he concluded all bitbases perform equally more or less. This is not the thread to discuss syzygy issues but butthurt fanboys keep popping up ...
My thinking is: 48.1GB single-sided seems perfect for 64GB systems, and probably performing the same on 32GB systems. Very soon those will be standard. Even without need for SSD then. You reach that point earlier than the double-sided system.
Below 32GB, one can't look at the 'data sheet' and tell what happens (at least I can't), but needs a measurement to decide if it hurts or wins. It can go either way. I was hoping you have some own results with Scorpio already.
Ok now your question seem honest, sorry I didn't do a test because I wasn't even aware of existence of a 64Gb flash drive until I went to Walmart to buy DVDs for egbbs about a week ago. Incidentally the sizes of the 6-men fit perfectly, which will avoid the need for SSD as you mentioned. My laptop has some SSD but not enough to hold them all. That is probably the case for most people as well, but I didn't do any tests on it but share them with Joshua.
jshriver wrote:It breaks my heart to see good people fight.
With other board games Daniel would not even need to make an argument for Bitbases, the argument has already long been settled in his favor. The same facts apply to chess but often times hype wins the argument over facts.
jshriver wrote:It breaks my heart to see good people fight.
Agree, and I don't see the point really, since different EGTB (or BB) formats have their advantages and disadvantages. Also, the more choice the better.
With that said, nice to see the 6-piece Scorpio available. Thanks for hosting them, and thanks to Daniel Shawul for generating and sharing.
