Author: K. Burcham
Date: 10:22:50 02/11/02
Go up one level in this thread
On February 11, 2002 at 12:30:45, Dann Corbit wrote: >On February 11, 2002 at 12:25:25, K. Burcham wrote: > >>On February 11, 2002 at 11:31:35, Dann Corbit wrote: >> >>>On February 11, 2002 at 11:24:28, K. Burcham wrote: >>> >>>> >>>> >>>>Please write out the number for these nodes that i took from program analysis. >>>>This was taken from three pc's analyzing same position. >>>>These numbers dont seem correct to me. The large number fritz7 posts, using >>>>smaller mhz and only 5 hours. do all programs post in kn's? Is shredder6 kn? >>>>fritz7 posts in kn's, so i assume i can write this 13,204,724,000. >>>>Do you just add three 0's to chess tiger for each period, and then three more >>>>because it is actually a kn value? if true then that value would be >>>>7,089,194,000,000? this number now is 13 didgits, DS is only 11 didgits. >>>>I know i am making this more complicated then it is. >>>>I just worked out the math and it seems S6 does not post in kn. >>>>kburcham >>>> >>>>3100 mhz >>>>Deep Shredder6 Depth 18 9 hours 15,884,648,840 nodes >>>> >>>>1507mhz >>>>Chess Tiger 14.0 Depth 19 10 hours 7089194... >>>> >>>>1000mhz >>>>Fritz7 Depth 16 5 hours 13204724kn >>> >>>It does not make any sense to compare node counts between two different >>>programs. Some programs visit a lot of nodes and others far less so. >>> >>>At very deep levels of analysis, it is not uncommon for bad numbers to be >>>reported anyway. >> >>based on your comments Dan, can I assume that the top people here pay no >>attention to nodes per move because the numbers are not reliable? > >No. You can assume that it is not logical to compare the nodes used by one >program compared to the nodes examined by a different one. > >>also it seems your understanding is that when some programs analyze a legal move >>it may/may not be added to the nodes per move post? > >I don't even know what you might mean by this. > >>If your understanding is accurate Dan, then i put to much attention in nodes per >>move. > >Compare MChess or Hiarcs to Goliath. You will conclude that Goliath is more >than ten times faster than they are. And the value of that information is? > >>I have been trying to compare differences when running in analysis mode >>in: mhz/hash/program/time. I was trying to establish some kind of pattern or >>relationship, with an increase in mhz, with time per move being equal, with >>different programs in the same test position. It would seem to me that the total >>legal nodes per move would be one of the many important factors related to >>program strength. > >Except that programs do not examine the total legal number of nodes. If they >did, they would be so supremely weak that patzers would beat them > >>ok here is another thought Dan that i have asked here before, your comment >>reminds me of this question. >> >>Program A searches 300kns. >>Program B searches 1000kns. >>Both programs are running on 1500mhz machines. >>same hash, same ram, same os. >>In match play these programs with this hardware have scored even in 200 games. >>Cpu usage is shown to be equal on both machines. >> >>Now the question: If we say that both programs continually post same amount of >>nodes per move, then could you give some examples what the larger mhz machine is >>doing to consume the kns speed? I assume that some here could make this answer >>very complicated. I still feel that some here could comment enough to give >>others some idea of what is going on that consumes processor but does not add to >>posted nodes per move. also, i am refering to before the egtb are accessed. > >Why do you care about the nodes visited? The quality of the answer is what is >important. You can make a comparison within the same program. So if Hiarcs >visits 90 million nodes on ply 17 and a later run visits 180 million nodes, then >the later run is better, because it was more complete. But you simply cannot >compare the node numbers of two different programs. It's apples to oranges. Dan "Except that programs do not examine the total legal number of nodes. If they did, they would be so supremely weak that patzers would beat them" I was not talking about this at all. I think most here are aware of this statement. I have a feeling that when you said "apples and oranges", you were restating my question. That is what i am curious about. What are some of the things in the programs that we are comparing. I am not saying that we can compare these nodes. (sometimes I think some like me are not capable of clearly stating a question because of our limited program knowledge). so I will try to ask again. Like I said before, if we monitor the cpu usage, and it is 100% on both machines, and both programs have same amount of nodes per move posted in analysis, then I can conclude that the processor in the larger mhz machine is busy doing something---the question was what are some of the tasks that this larger mhz machine is doing while running a chess program that do not allow it to post more nodes per move? (of course assuming that both systems are not burdened with any background software running, or non chess related tasks consuming cpu) kburcham
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.