Author: Robert Hyatt
Date: 21:47:43 09/24/01
Go up one level in this thread
On September 24, 2001 at 23:31:42, Vincent Diepeveen wrote: > >Have fun running on a 375Mhz processor. > >You're arguing like those Macintosh guys who praise their G4. > >Well my sister has a g4 at 450Mhz. and it's dual. > >Happy for my sister, she's btw a graphics designer and the only >reason she has macintosh is because she can't work with windows (yet). > >I ship to apple/motorola a few questions regarding things of the cpu. > >Simple questions. Still today i have no answer back from them! Vincent, the way you ask questions, it is no wonder they wouldn't answer. IE "question: why do you have such a stupid small branch target buffer?" Who would take the time to answer such? > >Same will happen with the processors you described. > >I mean a 375Mhz processor. How for christ sake is it going to beat >*ever* a 1.4Ghz MP processor or a 2.0Ghz P4 processor? > >Bandwidth, Que? I thought we talked about how fast a chessprogram >runs on it. Not on whether meteorologists are happy as i can answer >that already. Meteorologists aren't happy right now as it's raining >and thundering here! Just try them first. A lot of slow processors will fry a pair of very fast ones, every single time. > >>argue "but it can't do that." Just like you argued for months about the >>speedup numbers in Cray Blitz even though I showed you the raw output. "It > >what i say now is that the speedup you get at cray blitz at 8 processors >is 6.6, and that on a cluster with 8 nodes you'll never get close to that, >not even close to 4.0 that's what i think, because at the cray blitz >you had shared memory, here you do not have shared memory! > >So our old bet still stands! Any bet you'd like that says < 4.0 out of 8 is money in my bank... give me time... > >Actually i think it's going to be pretty hard to get a better than >squareroot speedup first at 8 nodes. sqrt 8 = 2.83 > >Note i wasn't complaining about cray blitz but about APHID being a >not realistic thing because its numbers are based upon 8 ply searches. Only the paper _you_ are looking at. Later aphid results were for Crafty, which was _not_ just 8 ply searches. That year crafty was doing 12 plies in Jakarta on a P6/200 machine. > >>just can't do that" is a common theme. But before you launch into that kind of >>argument, you ought to at _least_ know what you are talking about. If you >>haven't run on an SP, nor studied the tech references on it, it makes you look >>foolish to dismiss it as useless when some _serious_ computer scientists are >>using these machines daily to solve serious computational problems. Just >>because _you_ can't use 'em doesn't mean everyone is so limited. > >Yes i laugh for 375Mhz processors now that it's september 2001! > >To me making a machine existing out of 375Mhz processors >it's like next: > >You design worlds biggest aircraft (superjumbo) >and instead of using gigantic engines like jumbo's use, >you propel the worlds biggest aircrafty using 8192 old >bicycles like you see in our beloved capital Amsterdam so much. Ever heard of the man-driven aircraft that crossed the English Channel? Them bicycle pedals worked just fine. > >> >> >>> >>>No they were not slower than a 266Mhz PII would have been for me. And >>>my code had some things which now would do better at a 64 bits machine >>>but at that time a bit worse so i considered it equally fast to a PII >>>at 300Mhz. >>> >>>But the PII processsor was already years old at that time, whereas the >>>brandnew SUN processor was only clocked 300Mhz!!!!!!!! >> >>trash. so what? >> >> >>> >>>Each workstation (single cpu) was 5 times the price of a PII450 system. >>> >>>Of course that PII450 couldn't be put in a 32 processor shared memory >>>system, which the SUN most likely can be put in. >>> >>>The PII450 isn't hot swappable etcetera. >>> >>>So if you really want to run an application which has been written for >>>a cluster, and then can put it at 8192 processors (which will never >>>be able to get used at the same time i bet. most likely you can at >>>most allocate 1000 processors or so for a single job). >> >> >>How about adopting a new standard for yourself? Before you say something, >>check it out. "I bet" is not going to win friends and influence people in >>the world of computing. "I have shown" is far more convincing. >> >> >> >> >>> >>>In that case there is of course a use in having such a cluster. >>> >>>But the speedup over a dual 1.4 MP will be most likely not >>>even close to a factor 1000. >>> >>>Factor 100 perhaps? >> >>I'll bet that 8K processors can produce a 1000x faster search. But even >>if it was only 500 times faster, that will still cook your goose for Sunday. >> >> >> >> >>> >>>Pay a programmer a bit and it's a factor 30 perhaps? >>> >>>Now if this process runs for a week, then for research institutes there >>>is of course a big advantage, because you are 30 weeks faster! >>> >>>You need 1 week instead of 30. >>> >>>Obviously there is a use here to make a huge system, but i would be >>>pretty amazed if it's getting used like that. >>> >>>Most likely 100 scientists kick on that they get 64 processors >>>from a 8192 processor machine! >> >>Your "most likely" is garbage. Why don't you ask someone at one of the >>SP2 computer sites? I know some I will be happy to put you in touch with. >>Maybe some of the guys up at Oak Ridge will give you _real_ data to erase >>your bad guesses... >> >>Want some names??? >> >> >> >> >>> >>>The only real advantage on this machine is again for the meteorologists, >>>who can use big memory, bit storage, and big bandwidths. >>> >>>But well. They don't need many processors. Just a huge RAM memory! >>> >>>The bottom line is that compared to a 1.4Ghz MP, they already need >>>16 times more processors for each MP you would use! >>> >>>If a scientist allocates 32 processors with an application that's only >>>needing processor power, then a dual 1.4 will be faster for them! >>> >>>If they need its bandwidth, why then create a machine with so many >>>processors? >> >> >>You are getting to the issue. Maybe because they _need_ that much computational >>power. >> >> >> >> >>> >>>>>Still probably optimistic number of nodes a second. >>>>>So at 8192 processors, from which you can perhaps use a 1000 at a time, >>>>>I would get 15M nodes a second. >>> >>>>>Now that looks great, but that's of course on a CLUSTER. Speedup perhaps >>>>>10%. 1.5M nodes a second effectively, but the bigger the depth the less >>>>>the speedup gets as the branching factor will be worse, unless i accept >>>>>that the thing first slows down at each processor (which is a likely >>>>>approach) and pray that the latency is more than fast at this thing. >>>>> >>>>>So you sure outsearch deep blue by many plies, but not if a new deep >>>>>blue would be pressed on a chip using nullmove and DDR-RAM at it. >>>>> >>>>>So you are not faster in NPS, but search improvements would let it >>>>>search deeper. that still wouldn't make my DIEP faster on this machine >>>>>than DB was in nodes a second. >>>>> >>>>>Of course DBs focus upon only getting the maximum number of NPS (that's >>>>>how they advertised the thing. search depths have no commercial value) >>>>>sure made it faster than what i would get on this machine. >>>>> >>>>>>Is this really so for those in the know with hardware and these types of >>>>>>machines?
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.