Originally posted by: Idontcare
Originally posted by: pm
So I wasn't actually involved in the design of any of these CPU's and so I know them more as a user than I do as an Intel employee. But I'd be interested enough to collect any Unix based benchmarks that you want on all three. Just tell me what you want to see and then give me a few days to get the data. I think it's as simple as dragging a USB key from computer to computer and running a few commands... if the Pentium III can boot from USB... I can even grab data from the Via C3 900MHz if anyone is curious.
I have a Kill-o-Watt too. It'd be a fairly interesting project in an of itself, but if it helps someone, then I'll do it for the fun of it.
I believe the summary that says that the Pentium M is a heavily modified Pentium III is more or less correct. But adding a new FSB, an improved branch-predictor, a heavily modified and increased cache, and new instruction sets among other changes makes a pretty big improvement in my mind. As I recall, at the same clock frequency in something like Drystone MIPS, a Dothan is about 1.8x faster than Pentium III, and then in FPU Ops, using SSE2 code, more than 5x faster.
Well now that would be an intriguing set of data to have :thumbsup:
I appreciate the enthusiasm and offer but I don't really have any *nix apps of interest these days. Gaussian98 and Guassian03 were about my peak of using linux for HPC efforts and I wouldn't really put much merit into the floating-point results such a bench would generate with atom and the likes.
I'm more interested in knowing the superficial performance stuff...as in "compared to my 800MHz P3, a 1.6GHz Atom would prolly feel (1) on par in windows office apps, (2) slightly snappier in windows apps, or (3) prepare to have your friggen mind blown by the performance increase of Atom over P3!
Likewise I'd be interested in understanding the relative battery life improvement. Although this seems to be a moving target because the OEM's like to take lower power platforms and just pair them with commensurately smaller/cheaper/lighter batteries so the consumer is perpetually locked into this 2-3hr battery life. (Same with cellphones, argh)
But a 3-way review of P3 vs. Atom vs. C3 would be nice just to start building a working database of relative performance and performance/watt of these new netbook-class compute stations.