lkaufman wrote: We'll still be below Houdini 4 on IPON due to Houdini's contempt helping against weaker opponents and to IPON using AMD since Komodo does noticeably better compared to Houdini on Intel than on AMD overall.
Apologies for my bad memory Larry... what tests have been run to show that on a comparable basis ?
In the past, when I've noticed that someone running tests on AMD machines (for example Ingo/Ipon) gets worse ratings for Komodo relative to Houdini than on either my Intel tests or others (CEGT at same time limit as IPON or myself), we've checked relative nodes per second and invariably the ratio in favor of Houdini is significantly higher (ballpark of ten percent) on AMD. So roughly running on AMD costs Komodo about ten elo vs. Houdini compared to running on Intel.
I think Ray may be asking for data (proof), not something you may have observed...
There's always seems to be an excuse (especially when results are posted that don't meet your claims):
LC is not optimal, bad SSE performance, incorrect platform, etc.
Sorry to be frank, and in all due respect...
over the years your responses (especially without data to back them up) often appear to be commercial damage control/sugar-coating/spin.
lkaufman wrote: We'll still be below Houdini 4 on IPON due to Houdini's contempt helping against weaker opponents and to IPON using AMD since Komodo does noticeably better compared to Houdini on Intel than on AMD overall.
Apologies for my bad memory Larry... what tests have been run to show that on a comparable basis ?
In the past, when I've noticed that someone running tests on AMD machines (for example Ingo/Ipon) gets worse ratings for Komodo relative to Houdini than on either my Intel tests or others (CEGT at same time limit as IPON or myself), we've checked relative nodes per second and invariably the ratio in favor of Houdini is significantly higher (ballpark of ten percent) on AMD. So roughly running on AMD costs Komodo about ten elo vs. Houdini compared to running on Intel.
I think Ray may be asking for data (proof), not something you may have observed...
There's always seems to be an excuse (especially when results are posted that don't meet your claims):
LC is not optimal, bad SSE performance, incorrect platform, etc.
Sorry to be frank, and in all due respect...
over the years your responses (especially without data to back them up) often appear to be commercial damage control/sugar-coating/spin.
Okay, here is hard data. Compare IPON list with CEGT ponder on 5' +3" list. Conditions are almost identical, except IPON uses only AMD, while CEGT uses mostly Intel. Average the ratings of the two top Houdinis (3 and 4) and compare with the two top Komodos averaged (tcec and 6). Houdini's lead is 8.5 elo greater on IPON. I estimated 10 elo for this, and CEGT does use some AMD, so this is pretty good confirmation. But anyone with access to both hardware types can do the NPS comparison in just a few minutes to confirm this.
lkaufman wrote:I'm running matches on three Windows 7/8 Intel computers with a total of 40 cores between Komodo 7a (released this morning) and Houdini 4, single core games. Most of the games are at 5' + 3" except the games on the quad are at 2' + 1". Five man Syzygy for both, ponder off, book derived from TWIC high-level games, variable length.
Results after 1552 games are a virtual tie; Komodo is just 7 games ahead. Since it is clear that Komodo scales better with more time than Houdini, I think it is fair to say that we have finally surpassed Houdini in strength at most reasonable time controls, at least on Intel machines. Now there remains only Stockfish to surpass, a tall order to be sure.
lkaufman wrote:I'm running matches on three Windows 7/8 Intel computers with a total of 40 cores between Komodo 7a (released this morning) and Houdini 4, single core games. Most of the games are at 5' + 3" except the games on the quad are at 2' + 1". Five man Syzygy for both, ponder off, book derived from TWIC high-level games, variable length.
Results after 1552 games are a virtual tie; Komodo is just 7 games ahead. Since it is clear that Komodo scales better with more time than Houdini, I think it is fair to say that we have finally surpassed Houdini in strength at most reasonable time controls, at least on Intel machines. Now there remains only Stockfish to surpass, a tall order to be sure.
Quite intresting Larry!
I also think Houdini 4 and Komodo 7 are at the same level.
I am doing some kind of "endless" tournament with 14 engines.
All engine use 6cpu and no TBs.
Ponder OFF 10min +10sec
I got similar results with Komodo 1223 after 299 games:
lkaufman wrote: We'll still be below Houdini 4 on IPON due to Houdini's contempt helping against weaker opponents and to IPON using AMD since Komodo does noticeably better compared to Houdini on Intel than on AMD overall.
Apologies for my bad memory Larry... what tests have been run to show that on a comparable basis ?
In the past, when I've noticed that someone running tests on AMD machines (for example Ingo/Ipon) gets worse ratings for Komodo relative to Houdini than on either my Intel tests or others (CEGT at same time limit as IPON or myself), we've checked relative nodes per second and invariably the ratio in favor of Houdini is significantly higher (ballpark of ten percent) on AMD. So roughly running on AMD costs Komodo about ten elo vs. Houdini compared to running on Intel.
I think Ray may be asking for data (proof), not something you may have observed...
There's always seems to be an excuse (especially when results are posted that don't meet your claims):
TC is not optimal, bad SSE performance, incorrect platform, etc.
Sorry to be frank, and in all due respect...
over the years your responses (especially without data to back them up) often appear to be commercial damage control/sugar-coating/spin.
Okay, here is hard data. Compare IPON list with CEGT ponder on 5' +3" list. Conditions are almost identical, except IPON uses only AMD, while CEGT uses mostly Intel. Average the ratings of the two top Houdinis (3 and 4) and compare with the two top Komodos averaged (tcec and 6). Houdini's lead is 8.5 elo greater on IPON. I estimated 10 elo for this, and CEGT does use some AMD, so this is pretty good confirmation. But anyone with access to both hardware types can do the NPS comparison in just a few minutes to confirm this.
IPON is doing private testing for you...
CEGT is now using your recommendations to test Komodo (i.e. TC 5'3")
lkaufman wrote: We'll still be below Houdini 4 on IPON due to Houdini's contempt helping against weaker opponents and to IPON using AMD since Komodo does noticeably better compared to Houdini on Intel than on AMD overall.
Apologies for my bad memory Larry... what tests have been run to show that on a comparable basis ?
In the past, when I've noticed that someone running tests on AMD machines (for example Ingo/Ipon) gets worse ratings for Komodo relative to Houdini than on either my Intel tests or others (CEGT at same time limit as IPON or myself), we've checked relative nodes per second and invariably the ratio in favor of Houdini is significantly higher (ballpark of ten percent) on AMD. So roughly running on AMD costs Komodo about ten elo vs. Houdini compared to running on Intel.
I think Ray may be asking for data (proof), not something you may have observed...
There's always seems to be an excuse (especially when results are posted that don't meet your claims):
TC is not optimal, bad SSE performance, incorrect platform, etc.
Sorry to be frank, and in all due respect...
over the years your responses (especially without data to back them up) often appear to be commercial damage control/sugar-coating/spin.
Okay, here is hard data. Compare IPON list with CEGT ponder on 5' +3" list. Conditions are almost identical, except IPON uses only AMD, while CEGT uses mostly Intel. Average the ratings of the two top Houdinis (3 and 4) and compare with the two top Komodos averaged (tcec and 6). Houdini's lead is 8.5 elo greater on IPON. I estimated 10 elo for this, and CEGT does use some AMD, so this is pretty good confirmation. But anyone with access to both hardware types can do the NPS comparison in just a few minutes to confirm this.
IPON is doing private testing for you...
CEGT is now using your recommendations to test Komodo (i.e. TC 5'3")
Ingo (IPON) also runs private tests of unofficial SF versions and anything else that interests him. I expect he would do the same for Houdini if asked.
CCRL is not relevant to the discussion of 5' plus 3" tests since they don't test at anything close to that. Their 40/4 ratings do show Komodo 7 at +33 over Komodo TCEC (each on 4 cores), close to the IPON result. I would also point out that the gain on IPON is based on only using 4 man TB, which is almost useless as we already play most of the 4 man endings perfectly without TB (I think Q vs R is the only exception). I estimated +30 for the engine and other ten for the 5 man tb; at least the first figure looks completely justified now.. Maybe ten elo for 5 man tb is a bit too high.
lkaufman wrote: I think it is fair to say that we have finally surpassed Houdini in strength at most reasonable time controls, at least on Intel machines.