Daniel Shawul wrote:show me your data so that I can tell you your bugs
It would be a wonderful experience to have you in a helpdesk replying to customers!
Real data are what I told you above: overhead in Syzygy appears to be an order of magnitude smaller than with other table base solutions (including egbb). The node speed drop with Syzygy probing is an order of magnitude smaller than with the 3 other probing solutions I've implemented in Houdini, and allows for probing very deep in the tree with multiple threads.
If you want, you can use Houdini 3 and soon Houdini 4 to generate and compare data yourself. Drop me an e-mail at houdini@cruxis.be if you want access to a Houdini license.
Can some decompile Houndini 3 and see if Houdart changed the egbb codes searchers from 8 to 32. I just can't believe it , the same guy who botched up basic EGBB implementation, actually looked in to the code ...
[MODERATION]
No directed to anyone in particular, I am replying to the last post.
There are complaints about this thread. Maybe I missed something, but I cannot see a direct violation of the charter. What I do see is a typical rudeness when hot topics are contested. However, I cannot understand why it needs to be so hot. Regardless, I respectfully ask to tone it down a notch. Even if no violation was committed, it is bound to happen and the current tone discourage members to participate.
Miguel
PS: Please, people, do not answer to this message and try to argue/fight with me. That will be off topic and I will axe it. I am just trying to calm down the waters here, for the rest of the members at least.
I apologize for suggesting that someone decompiles it. But it is just that he is is saying he compiled for 32 cores, when in reality scorpio egbb probe code has hard coded numbers that work only for 8 cores. No one ever compiled for more than 8 cores before. I will spend a few minutes with hexrays and see how he botched it up yet again...
Daniel Shawul wrote:Haha it is very likely that I have found yet another problem of yours with scorpio EGBBS ,besides your botched up implementation ofcourse. Scorpio egbbs are not compiled for more than 8 processors. The number 8 is hard coded so even Jim never compiled for more than 8. No wonder you see orders of magnitudes decrease in speed, and may even hang, this pretty much explains it. I repeat No one ever compiled egbbdll for more than 8 processors Look in the code if you don't beleive me. Infact I know so because TCEC's Martin told me Scorpio can't run with 16 cores and I told him exactly that.
And I also looked in the cache and realized how good I implemented it! To begin with every single EGBB has its own cache that would only require locking when two threads probe KBNKB simultanesouly. Even with that a 16Mb cache is like 2000 entries which is fast.
You fail to realize that Houdini doesn't use egbbdll, it compiles the Scorpio code in the engine itself. The Scorpio code in Houdini works with up to 32 threads.
It must be gratifying for you to boast the efficiency of your own code, but your reply also shows that you've never actually used it with large number of threads probing simultaneously. I suggest that you start doing that, compare the performance to the Syzygy bases, and then we can continue the discussion.
The implantation is on you if you want more than 8 threads. If your code does not scale well find out why and fix it. The system is not at fault. The comrades wont help you, you have to do it yourself.
It is very much the case here. I found the relevant part of the probing code and saw critical parts gone!! To be sure I have to use the multi-threaded version of the decompiler but it is very very unlikely that one would find those hard coded numbers, unless you authored it yourself. Even I forgot about it until Martin started playing 16 cores and reported bugs. It is just very very odd for a guy who didn't even implement probing correctly, to find out such details buried in the code, unbelievooble. More tomorrow once I find out how to decompile multi-threaded code properly.
As a sidenote, I want to mention that this thread was intended to be about the Syzygy bases.
Thanks your for your Scorpio bitbases Daniel, as a user I think they were the first bitbases I knew and used, IIRC with the Bright engine, mainly. Interesting experience of endgame strength, many years ago already.