Author: Uri Blass
Date: 08:11:46 04/29/01
Go up one level in this thread
On April 29, 2001 at 11:03:36, Robert Hyatt wrote: >On April 29, 2001 at 05:42:42, Uri Blass wrote: > >>On April 29, 2001 at 05:08:37, Chessfun wrote: >> >>>On April 29, 2001 at 04:01:32, Uri Blass wrote: >>> >>>>On April 29, 2001 at 03:39:53, Bruce Moreland wrote: >>>> >>>>>It seems clear why it wasn't invited to participate: >>>>> >>>>>1) The organizer is going to use a multiprocessor machine. >>>>>2) Fritz and Junior run on a multiprocessor machine. >>>>>3) Tiger is known to not run on such a machine. >>>>>4) Tiger is very strong, but if it is stronger than Junior or Fritz, it's >>>>>probably not stronger by much. >>>>>5) A multiprocessor machine should produce a significant performance boost. >>>>>6) It is hoped that the event will produce an "accurate" winner. >>>>> >>>>>If you allow these points, you can make a case that Tiger on a single processor >>>>>can't be stronger than Junior or Fritz on a multi. >>> >>>>I understood that tiger can use more than one processor. >>> >>> >>>I never understood that. I took it that one could be produced. >>> >>> >>>>I guess that Tiger could earn less than Fritz or Junior because of the fact that >>>>Christophe and Ed had not enough time to optimize tiger for more than one >>>>processor but it still can earn something from more than one processor. >>> >>> >>>Assuming they are within say 20 SSDF rating points of each other on >>>equal machines which seems likely. It seems highly unlikely without >>>proper debugging and testing that the above would be true. And an >>>event such as this IMO is not the place for that testing/debugging. >> >>I believe that testing debugging can be done in a few days if the target is only >>to be practically sure that it is better than the one processor version. > >Sorry, but you are badly wrong. It takes _months_. The bugs are hard to >find. They will only show up infrequently (not reproducible easily) and >so forth. > > >> >>You only need to test it in games to see that there is no problem and to test it >>in test position to see if the 2 processor version is faster than the one >>processor version. > >Don't make statements until you _write_ one. I finished Cray Blitz (the most >recent parallel search algorithm) in 1988. I found bugs for two years at >_least_. When I went from 2 to 4 processors new bugs showed up. 4 to 8? >same thing. 8 to 16? Yet again. > > > > > >> >>You only needs to give the parallel version to play to see if there are problems >>in games and few days are enough to get a enough games to be sure that bugs are >>not common and usually do not happen in games. > > > >Suppose I told you I had bugs that would show up in a given position. But >only one of every 10,000 times the position was ran for 3 minutes? > >Your debugging idea makes me (as a software engineer) cringe. You could play >all day, every day, for a month, and still have major and serious bugs that you >had not _seen_. My point is not that you can avoid bugs but that I believe that you can have a better version in a few weeks. I do not say that the better version is optimized for using more than one processor. I read here that patzer did something productive about parallelizing(in that case I understood that 2 processors were 1.2 times faster than 1 processor in relatively a short time). If you are 1.2 times faster and the bug happens only in 1 out of 100 games then the new program earns something from 2 processors. Uri
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.