[Date Index] [Thread Index] [Author Index]
Re: laptop recommendation to run mathematica fast?
On Jun 23, 12:05 pm, John Fultz <jfu... at wolfram.com> wrote: > Most people wouldn't consider the Eee PC a laptop. Wikipedia lists it as a > "subnotebook/netbook". If you were looking for speed first and portability > second, you're at the wrong end of the hardware scale. The design of netbooks > factors portability and battery life as strong primary and secondary goals, withperformance as a distant tertiary goal. . . . i begin to think that those most people are right! before all this started i had just assumed that these little netbooks were just little laptops you could lift with one hand, ran for ages and fit inside a handbag. (so, ok, i admit it, i have a fondness for large tapestry handbags, with fringes, does this necessarily make me a frivolous person? -- no, do not answer that!) but in one of the emails people most kindly sent on this matter was a pointer to http://www.cpubenchmark.net/ , a website that collates lots of cpu's benchmarks and shows that the little intel atom N270 in my asus eee pc scores just 305 compared to typical numbers of 2000 and higher for those core 2 duo things, so with the benefit of the hindsight it is unsurprising perhaps the rather poor showing of 0.32 in the internal Mathematica v7.0 benchmark the asus gets. . . . and yet, and yet . . . the now ancient 2.4 GHz pentium 4 scores only 231 in those same cpubenchmark tests, slower than the atom, but yet still has a Mathematica benchmark of 1.0, or 3 times more speedy! comparisons are odious to be sure, but for me at least this performance still is most puzzling and makes the choosing of a real laptop that much harder, so if anyone else would care to post their laptop Mathematica benchmark here i think that could be most helpful but, never mind! may i extend a big hug to everyone who has posted here or emailed to me on this subject? it was most kind of you all to expend time and effort on this matter. a big thankyou f.c. P.S as a complete aside i should perhaps add that by using that OMP_NUM_THREADS = %NUMBER_OF_PROCESSORS% environmental variables trick mentioned in this group many moons ago, i can tweak the eee pc benchmark for Mathematica v5.2 up to 0.44 and use both threads for the matrix multiplication process etc. with cpu usage maxing at 100%. for Mathematica v7.0 however this trick has no effect and benchmark stubbornly remains fixed at 0.32 with maximum cpu usage 50%.