Re: brief olpc call notes

From: Tomas Frydrych <tf_at_o-hand.com>
Date: Thu Dec 07 2006 - 09:20:46 CET

Ivan Krstić wrote:
> Some brief comments inline.
>
> Robert Staudinger wrote:
>> Ivan starting explaining the limited hardware resources of the target
>> device. Clock speed may seem fast but when taking into account limited
>> caches (small L1, no L2) and slow flash memory overall performance
>> comparable to first generation pentium at 75-90MHz.
>
> Actually, not a first generation Pentium, but a Pentium 4 Mobile running
> at roughly 50-90 MHz. I was giving a number scaled to the processor a
> lot of people have in their normal laptops today for comparison, but the
> number is *extremely* unscientific (rough approximation from Python
> benchmarks) and meant only to give people an idea of what we're up against.

From what I have gathered from JG's presentation at Guadec, this is not
so much due to the HW, but because the HW is being throttled rather
significantly to reduce power consumption. One of the implications of
this is that applications need to be biased toward code speed, rather
than memory use.

It would be useful to have some benchmarks for AbiWord itself at some
point, so that we could pin down what bits need most attention
performance-wise specifically from the OLPC's point of view.

Tomas
Received on Thu, 07 Dec 2006 08:20:46 +0000

This archive was generated by hypermail 2.1.8 : Thu Dec 07 2006 - 09:21:32 CET