Author: Robert Hyatt
Date: 14:53:19 04/14/03
Go up one level in this thread
On April 14, 2003 at 16:06:28, Tom Kerrigan wrote: >On April 13, 2003 at 11:15:15, Robert Hyatt wrote: > >>On April 13, 2003 at 02:38:00, Tom Kerrigan wrote: >> >>>On April 13, 2003 at 01:03:17, Robert Hyatt wrote: >>> >>>>>You have been criticizing people for "bad math" this entire thread. You rejected >>>>>the notion of a 50%-50% division: >>>> >>>>I didn't "reject" anything. I simply said it didn't match _my_ results and >>>>it doesn't. I've already posted some results with wild variance. Separate >>> >>>Nonsense. Just read the quote I posted and you seemingly ignored: >>> >>>"But I don't buy the 50% stuff, the cpu is not that simple internally. One >>>thread will run at nearly full speed and the other gets slipped into the gaps" >> >>And? That is based on _my_ testing with Crafty, as I said. IE it doesn't >>happen that way for my _threaded_ application. I agreed that it _did_ happen >>for totally independent processes, after I tested, because I don't normally >>run that kind of test... > >Face it, you were making statements about the processor's architecture that were >incorrect (remember the Cray YMP drivel you tried to use as some sort of >supporting argument?). You obviously didn't realize that the CPU's resources >were cleanly divided and now you're trying to pin this on some semantic "thread" >vs. "processor" mumbo jumbo and saying that your numbers back you up, except >that when you wrote the statements, you hadn't run any experiements at all If you look back at my _initial_ post, I _clearly_ said I could not find any reference to how SMT divided stuff up. I then _clearly_ said that for Crafty, it did _not_ divide things evenly. That's _all_ I have said repeatedly. And that hasn't changed a bit, because both are still true. Although someone did post something they found that says that the two logical cpus are "evenly divided". And that seems true for threads that are not doing what I am doing with locks, and so forth. What is causing this is unknown at present. >(remember all the explanations about how running an experiment would be nearly >impossible?) and didn't even know what your numbers were. And now that you have >numbers, the most disparate being 70-30, they _don't_ back you up. I don't follow. I said "I can't run the test without fixing up a linux kernel that lets me lock a thread to a logical cpu." I finally got around to testing it. And I got that 70-30 performance split between the two logical CPUs. I also got almost 50 50 with two separate processes. So I don't see what "doesn't back me up". 70-30 is _not_ nearly equal in my book... If you get a >30% speedup from HT and one logical processor runs at full speed, that's 77-23 >right there. Sure, you said "nearly" full speed, but you also didn't see a range >as wide as 70-30 in all cases. No I didn't. But I also didn't see 50-50 either. So there is a variable, as there seems to _always_ be when SMP code is run whether the hardware is SMP or SMT seems irrelevant. > >>But it doesn't happen for the particular program that was the subject of this >>thread, namely does SMT produce any _real_ speedup in Crafty. Data said > >How do you figure that? The root of this thread is 3.06 Xeon Test Results, the >post that I referred you to, and that post doesn't even have the word Crafty in >it. Did you see _my_ response? Did you find the word "crafty" in it? Sure you did.. Vincent then started with the "It doesn't work, you are running _slower_ with SMT than without... etc." Talking explicitly about Crafty. I then posted _crafty_ data showing that SMT _does_ improve search time performance by up to 20% on the positions he gave... Again, _all_ about crafty. > >>>You go on to say that your assertion is supported by your results, but there is >>>a big difference between saying "this is how things work and my results agree" >>>and "my results indicate this is how things work." >> >>Again, look at the context of the entire thread. Not just the post you read. >>It was about Crafty and SMT. That's why I ran the tests and posted the results. > >Again, no it wasn't. > >-Tom
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.