Dann Corbit wrote:No need to bother with AsmFish anymore.
It's dead or dying, and now it's hit the floor.
That's just nonsense. It is a lot faster than SF and therefore stronger unless a bug has crept in. If a bug has crept in, it is just a matter of finding and quashing it.
IMHO there is no need for such a dramatic title as you have used here, which in my perception does not show much respect for the incredible amount of work that has gone into creating and maintaining asmFish.
Well, of course you are right.
But I mentioned it a month ago nicely, and nobody paid any attention to it.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
Dann Corbit wrote:No need to bother with AsmFish anymore.
It's dead or dying, and now it's hit the floor.
That's just nonsense. It is a lot faster than SF and therefore stronger unless a bug has crept in. If a bug has crept in, it is just a matter of finding and quashing it.
IMHO there is no need for such a dramatic title as you have used here, which in my perception does not show much respect for the incredible amount of work that has gone into creating and maintaining asmFish.
Well, of course you are right.
But I mentioned it a month ago nicely, and nobody paid any attention to it.
ernest wrote:
Amazing that this is due to the new test conditions (mainly: longer Time Control).
Correct. The average game-length raised from 190'' to 430'' with the new time-control. And on the new machine, around +30% more nodes per second are crunched. So, we talk about around 3x more nodes, the engines can crunch in a game. And the more nodes are calculated, the Elo-advantage of faster calculation (asmFish) gets lower.
So, it would make definitly no sense to use asmFish in TCEC (instead of Stockfish) - as some people suggested...
But asmFish is an amazing project. Especially on smartphones or other not-so-fast machines, asmFish is really a step forward.
When the BrainFish-testrun is finished (weekend), I will start asmBrainFish-testrun (latest asmFish, using Cerebellum-Library) - the best of the best of Stockfish. Both projects (assembler-rewrite of Stockfish and Cerebellum (Stockfish-calculated and backwards minimaxed opening-moves)) are really cool developments based on Stockfish. But it is much cooler, that both projects can be used together as one and that so much ELO can be gained out of this!
Stefan (SPCC)
There is one other effect that I can think of: by introducing all the "Brainy" versions into the database, at the moment there are fortunately only two left, but suddenly asmFish was no longer the highest rated engine. In a ratingpool, the top engine (or engines that are clearly stronger than the rest) will 'run away' and the other ratings will be compressed (bottom engine will freefall, I experienced this effect personally from my days at the chessclub ) So that would push the asmFish rating downwards as well. It is not a full rating pool because all the engines are running a gauntlet only, against non stockfish entities. I don't know if the effect is still there because asmFish does not have to compete directly with Brainfish and asmBrainfish. But there might be some effect?
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan
Werewolf wrote:I thought asmfish was +20% or something faster than SF Dev.
It is even more than that (25 to 30%) depending on the position...
But it is clear that the Elo advantage of speed only (double speed for instance) decreases significantly when Time Control is increased.
Andreas Strangmuller has convincingly shown this effect in his tests
(see http://www.fastgm.de/).
Still, at any time scale, for SF to achieve the same move quality as asmFish, the user must give it 30% more time.
Henk wrote:I was surprised they did not quit with AsmFish earlier. Already in 1987 they told me you should only write the small time critical parts of a system in assembler.
Back in 1988 I took an assembler course. It marked the end of my interest in computer programming. Either that, or the LISP class afterwards. Those were the only two courses I ever quit on. It didn't help that I had the most boring and unqualified teachers.
Werewolf wrote:I thought asmfish was +20% or something faster than SF Dev.
It is even more than that (25 to 30%) depending on the position...
But it is clear that the Elo advantage of speed only (double speed for instance) decreases significantly when Time Control is increased.
Andreas Strangmuller has convincingly shown this effect in his tests
(see http://www.fastgm.de/).
Still, at any time scale, for SF to achieve the same move quality as asmFish, the user must give it 30% more time.
exactly, +1, people go to great lengths to overclock their machine by 5 or 10%, here you get SF almost 30% faster - it's the same as overclocking you machine by almost 30% ... it's amazing how negative this board can get...
Henk wrote:I was surprised they did not quit with AsmFish earlier. Already in 1987 they told me you should only write the small time critical parts of a system in assembler.
Back in 1988 I took an assembler course. It marked the end of my interest in computer programming. Either that, or the LISP class afterwards. Those were the only two courses I ever quit on. It didn't help that I had the most boring and unqualified teachers.
[Course was operating systems. Context switch being time critical.
Teachers told me that if they were good they would not be there for they would be underpaid]
Asmfish is an excellent piece of work and it speeds up the stockfish a lot.
It is not necessary to determine the Elo difference at all, but the only thing that matters is the speed advantage.
If I have the choice to wait 1 hour for a certain result or only 40 minutes - then I always prefer 40 minutes and it doesn't matter how much Elo that is.
Henk wrote:I was surprised they did not quit with AsmFish earlier. Already in 1987 they told me you should only write the small time critical parts of a system in assembler.
Back in 1988 I took an assembler course. It marked the end of my interest in computer programming. Either that, or the LISP class afterwards. Those were the only two courses I ever quit on. It didn't help that I had the most boring and unqualified teachers.
Found the introduction movie. But it is in Dutch. They say I should spend most of the time on problem analysis and only a small part on coding. Already knew I was doing something stupid. https://www.youtube.com/watch?v=YfIMCABYcC4&t=912s
Henk wrote:They say I should spend most of the time on problem analysis and only a small part on coding.
“Would you tell me, please, which way I ought to go from here?”
“That depends a good deal on where you want to get to,” said the Cat.
“I don’t much care where–” said Alice.
“Then it doesn’t matter which way you go,” said the Cat.
“–so long as I get SOMEWHERE,” Alice added as an explanation.
“Oh, you’re sure to do that,” said the Cat, “if you only walk long enough.”
) So that would push the asmFish rating downwards as well. It is not a full rating pool because all the engines are running a gauntlet only, against non stockfish entities. I don't know if the effect is still there because asmFish does not have to compete directly with Brainfish and asmBrainfish. But there might be some effect?
Those were the only two courses I ever quit on. It didn't help that I had the most boring and unqualified teachers. 
