Re: AMD vs. Intel Floating Point
- To: mathgroup at smc.vnet.net
- Subject: [mg29481] Re: AMD vs. Intel Floating Point
- From: "Orestis Vantzos" <atelesforos at hotmail.com>
- Date: Fri, 22 Jun 2001 02:20:09 -0400 (EDT)
- Organization: National Technical University of Athens, Greece
- References: <9gcman$364$1@smc.vnet.net> <9gn7i2$ere$1@smc.vnet.net> <9gpog0$hg4$1@smc.vnet.net> <9gs2ic$k3p$1@smc.vnet.net>
- Sender: owner-wri-mathgroup at wolfram.com
I am not a hardware spec, I write what I know to be true. Seti@home runs better on Intel processors and I have seen that with my own eyes. Why I consider it a quasi-valid benchmark for floating-point operations? Well, all it really does is certain mathematical transformations (Fast Fourier among them)...it has to depend on floating -point operations. Could the difference be in the math-coprocessor? Errors in AMD floating-point operations that force them to repeat a portion of the operations? I can't really tell, hardware is not my field... Orestis PS. Jens loosen up... "Jens-Peer Kuska" <kuska at informatik.uni-leipzig.de> wrote in message news:9gs2ic$k3p$1 at smc.vnet.net... > Hi, > > I have never seen an benchmark where a Intel CPU of the same > clock speed beats an Athlon. > > You may look at: > > http://fampm201.tu-graz.ac.at/karl/timings40.html > > and see that the fastest 5 (five !) entries > are Athlon CPU's. Since an Athlon has one floating point > pipeline more than an Intel CPU it is foolish to ask > > "Work three workers more than two ?" > > I would realy like to see why Seti@home is slower on an > Athlon -- but it is definetly *not* the floating point > performance. > > BTW since when where *screen saver* used as floating point > benchmarks ? > > Regards > Jens > > Orestis Vantzos wrote: > > > > In what sense is it foolish? Seti@home for instance, which relies heavily on > > floating point operations, does work slower on AMD chips... > > Orestis >