[Date Index] [Thread Index] [Author Index]
Re: Help! Mathematica on my 500MHz outperforms my GHz machine!
There are serious differences in performance between windows NT 4.0 and 9x for Mathematica 2.x and 3.x, typically a factor of 4 on the same machine in favor of NT. The Mathematica kernel is sleeping in the background in win 9x most of the time and graphics has no speed at all. Mathematica 4.0 again is faster a factor of 3 to 10 for many routines, especially graphics and numerics. r.f. terryis at my-deja.com wrote: > > Help! > > I just took delivery of a 1GHz machine which I > bought expressly for the purpose of running > Mathematica. After installation and setup I ran > a couple of simulations only to discover that it's > performing at about the same level as my 500MHz > Pentium III!! Can anyone explain to me why this is > the case? > > Additional configuration information: > > Pentium III vs. AMD Athlon (K7 overclocked) > 500MHz vs. 1000MHz > 128MB RAM vs. 128MB RAM > 100MHz Front Side Bus vs. 200MHz system interface > bus based on the Alpha EV6 bus protocol > > Windows98 vs. Windows NT 4 > > Both running the same version of Mathematica > (18.104.22.168), the same simulation (90% of which > is simple floating point calculations), both > dedicated only to running Mathematica at the time > of testing. Disk speed is irrelevant as the > simulation completes without writing to disk. Swap > space doesn't seem to be the issue. > > I don't have the exact clock speed of the AMD > processor but I would imagine it's at least 600MHz > which in my mind means that it should be faster > even without over-clocking to 1GHz but it isn't! > > Simple tests of floating point speed (with for > example a QBASIC program) and other processes > perform as expected so it would appear that > Mathematica is doing something different on the > AMD machine then on the Pentium, but what? > > I've already gone to Wolfram Tech Support. They > suggested I post here. > > Any and all help will be appreciated. > > Please reply to tchung at us.oracle.com as I do not > regularly read this newsgroup. > > Thanks in advance! > > Sent via Deja.com http://www.deja.com/ > Before you buy. -- Dr. Roland Franzius Theor. Physik FB Physik, Univ. Osnabrueck +++ exactly <<n>> lines of this message have value <<FALSE>> +++