Sean Evans wrote:I do not believe a chess program is 2700 ELO at 500MHZ. In addition, the opening book will be small and there are no endgame tablebases.
This would make a good Hyatt question!
Quite possible
the SSDF has rated the Resurrection II(Revelation)- Rybka 2.2n ..2634..
the Resurrection II runs on a Xscale PXA255 processor-500 Mhz-32MB Dram
in fact there are several engines rated by the SSDF running on 400-500 Mhz processors with ratings near or higher then 2700
swami wrote:Could somebody play some games with Glaurung for Ipod against Hiarcs for Ipod? It should be interesting.
I don't think it would be particularly interesting with the current iPod version of Glaurung. It will be a little more interesting with the upcoming Stockfish-based version, which is a lot stronger (in a match between the two on my first generation iPod Touch, Stockfish won by 77-23, with 66 wins, 28 draws and only 8 losses). I still expect Hiarcs to be clearly stronger, though. Stockfish 1.4 seems to be a little stronger than Hiarcs 12.1 on fast 64-bit desktop computers, but on the ARM CPU it is crippled by its bitboard infrastructure, and runs extremely slowly.
I'm a little confused by Hiarcs' results in Argentina. Based on a combination of test suites, own games and games against humans on the ICC, I estimate the strength of Stockfish (current development version, which is stronger than 1.4) on the iPod Touch against humans in blitz games to be around 2200 or at most 2300. In slow games, it is certainly weaker, perhaps around 2100. It is not surprising that Hiarcs is stronger than Stockfish on the iPod Touch, but how can it be as much as 500-600 Elo points stronger?
swami wrote:Could somebody play some games with Glaurung for Ipod against Hiarcs for Ipod? It should be interesting.
I don't think it would be particularly interesting with the current iPod version of Glaurung. It will be a little more interesting with the upcoming Stockfish-based version, which is a lot stronger (in a match between the two on my first generation iPod Touch, Stockfish won by 77-23, with 66 wins, 28 draws and only 8 losses). I still expect Hiarcs to be clearly stronger, though. Stockfish 1.4 seems to be a little stronger than Hiarcs 12.1 on fast 64-bit desktop computers, but on the ARM CPU it is crippled by its bitboard infrastructure, and runs extremely slowly.
I'm a little confused by Hiarcs' results in Argentina. Based on a combination of test suites, own games and games against humans on the ICC, I estimate the strength of Stockfish (current development version, which is stronger than 1.4) on the iPod Touch against humans in blitz games to be around 2200 or at most 2300. In slow games, it is certainly weaker, perhaps around 2100. It is not surprising that Hiarcs is stronger than Stockfish on the iPod Touch, but how can it be as much as 500-600 Elo points stronger?
Well, I think you certainly underestimate the strength of stockfish on Ipod, Tord. Maybe it's the result of your usual self-deprecating modesty
It might be better if you do some tests of Stockfish on IPod against 2500 rated engines (on PC) and see the stats. I believe most IM's play at this level. I recommend Nejmet, Brutus, Rotor, The Crazy Bishop, Phalanx Romi chess.... as some of the consistent 2400 - 2500 rated engines. I'm sure that Stockfish on Ipod will beat these engines majority of the time.
swami wrote:Well, I think you certainly underestimate the strength of stockfish on Ipod, Tord. Maybe it's the result of your usual self-deprecating modesty
I think it's more likely that you underestimate the importance of a 100x difference in speed.
It might be better if you do some tests of Stockfish on IPod against 2500 rated engines (on PC) and see the stats. I believe most IM's play at this level. I recommend Nejmet, Brutus, Rotor, The Crazy Bishop, Phalanx Romi chess.... as some of the consistent 2400 - 2500 rated engines. I'm sure that Stockfish on Ipod will beat these engines majority of the time.
I just started a 100 game match against Phalanx. The first game was a draw, Phalanx has an advantage in the endgame in the second. I played another match against Phalanx a few months back, and at that time, Phalanx was crushingly superior (I don't remember the exact score). It will probably be closer this time, but I still expect Phalanx to win comfortably.
swami wrote:Well, I think you certainly underestimate the strength of stockfish on Ipod, Tord. Maybe it's the result of your usual self-deprecating modesty
I think it's more likely that you underestimate the importance of a 100x difference in speed.
It might be better if you do some tests of Stockfish on IPod against 2500 rated engines (on PC) and see the stats. I believe most IM's play at this level. I recommend Nejmet, Brutus, Rotor, The Crazy Bishop, Phalanx Romi chess.... as some of the consistent 2400 - 2500 rated engines. I'm sure that Stockfish on Ipod will beat these engines majority of the time.
I just started a 100 game match against Phalanx. The first game was a draw, Phalanx has an advantage in the endgame in the second. I played another match against Phalanx a few months back, and at that time, Phalanx was crushingly superior (I don't remember the exact score). It will probably be closer this time, but I still expect Phalanx to win comfortably.
Results tomorrow.
I'm looking forward to the results. I think maybe Human IM's and GM norm players are over-rated. Maybe 2200 rated engines on today hardware play at their level. I remember Danasah scoring positive results against IM's/GM's of 2500 rating range in some tournaments that Pedro reported, I don't remember the actual scores but I guess it was certainly high for Danasah, which is a 2500 rated engine btw. So maybe humans are actually at around the level of a 2300 engine on today's PC.
The ipod hardware is not comparable to the best PocketPC hardware.For comparision I ran the latest versions of Chess Genius on both a ipaq 312mhz and ipod touch 1st generation.The ipaq had faster performance.Let alone a faster ipaq that runs at 624mhz.I think the cpu's in ipods are Samsung which perform much slower than there inte counterparts.
Tord Romstad wrote:I just started a 100 game match against Phalanx. The first game was a draw, Phalanx has an advantage in the endgame in the second. I played another match against Phalanx a few months back, and at that time, Phalanx was crushingly superior (I don't remember the exact score). It will probably be closer this time, but I still expect Phalanx to win comfortably.
More or less as I expected. Stockfish simply gets terribly outsearched. It often holds its own into the late endgame, but then falls apart tactically in time trouble.
It is likely that Stockfish's results could have been slightly better with some minor timing adjustments. I noticed that 11 of the games were lost on time, at least one of them in an easily won endgame. Some of the 11 time losses could probably have been avoided by checking for timeout a little more often, but Phalanx would still have won by a huge amount.
This result indicates that Stockfish on the 1st generation iPod Touch (with a 416 MHz ARM CPU) has a strength of about 2300 on the CCRL scale.
I think maybe Human IM's and GM norm players are over-rated. Maybe 2200 rated engines on today hardware play at their level. I remember Danasah scoring positive results against IM's/GM's of 2500 rating range in some tournaments that Pedro reported, I don't remember the actual scores but I guess it was certainly high for Danasah, which is a 2500 rated engine btw. So maybe humans are actually at around the level of a 2300 engine on today's PC.
We don't have enough data to be sure, but I suspect that the opposite is true. The CEGT and CCRL lists are around 200 points too high compared to FIDE ratings.
Tord Romstad wrote:I just started a 100 game match against Phalanx. The first game was a draw, Phalanx has an advantage in the endgame in the second. I played another match against Phalanx a few months back, and at that time, Phalanx was crushingly superior (I don't remember the exact score). It will probably be closer this time, but I still expect Phalanx to win comfortably.
More or less as I expected. Stockfish simply gets terribly outsearched. It often holds its own into the late endgame, but then falls apart tactically in time trouble.
It is likely that Stockfish's results could have been slightly better with some minor timing adjustments. I noticed that 11 of the games were lost on time, at least one of them in an easily won endgame. Some of the 11 time losses could probably have been avoided by checking for timeout a little more often, but Phalanx would still have won by a huge amount.
This result indicates that Stockfish on the 1st generation iPod Touch (with a 416 MHz ARM CPU) has a strength of about 2300 on the CCRL scale.
I think maybe Human IM's and GM norm players are over-rated. Maybe 2200 rated engines on today hardware play at their level. I remember Danasah scoring positive results against IM's/GM's of 2500 rating range in some tournaments that Pedro reported, I don't remember the actual scores but I guess it was certainly high for Danasah, which is a 2500 rated engine btw. So maybe humans are actually at around the level of a 2300 engine on today's PC.
We don't have enough data to be sure, but I suspect that the opposite is true. The CEGT and CCRL lists are around 200 points too high compared to FIDE ratings.
Hi Tord,
Do you have a automated way to setup these matches? (Aka: So forum of autoplayer)
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers
AdminX wrote:Do you have a automated way to setup these matches? (Aka: So forum of autoplayer)
Most Chess GUIs (and a few command line utilities) have some sort of support for running automated matches between chess engines. The only little problem with Stockfish and Phalanx is that they use two different protocols (Stockfish is a UCI engine, and Phalanx an XBoard engine), and most GUIs don't support both. I think Arena supports both, but as a Mac user, I can't use it.
I used XBoard (the Unix version of Winboard) with Stockfish running through Fabien Letouzey's excellent PolyGlot adapter, which makes a UCI engine look and behave like an XBoard engine from the GUI's point of view. XBoard, Phalanx and PolyGlot were running on my iMac, and Stockfish (just the UCI engine) on my iPod Touch. I use a similar setup when running Stockfish on the ICC.
