[Olpc-open] OLPC G1G1 sales start today

[Nicole Lee]

I see your point, although I didn't mean my comment exactly as a complaint -
simply a measure of relative position. I don't expect an XO to run as fast
as my school computer, considering all the constraints on the design, as
well as the very-much-still-in-development software. I've also played with
various models and the more recent ones (and more recent versions of Sugar)
definitely run more nicely.

-Nikki

On Nov 19, 2007 3:17 AM, Ivan Krstić <krstic at solarsail.hcs.harvard.edu>
wrote:

> On Nov 18, 2007, at 5:22 PM, Nikki Lee wrote:
> > XOs do run slowly, which I personally have found plenty frustrating.
>
>
> The first widely-used desktop x86 CPUs running at approximately the
> speed of the XO's CPU became available with the introduction of the
> Pentium II Deschutes in 1998. With a 100MHz FSB, you could drive the
> core at 450MHz -- at 27.1W TDP. By comparison, our Geode has 3.9% less
> clock speed, four times the L1 cache, a quarter of the L2 cache, and
> consumes about a tenth of the power.
>
> Deschutes-based processors were in wide use in 1999 and 2000. People
> who owned computers with clean OS installs those years by and large
> didn't complain about their computers being too slow or even
> particularly frustratingly slow to use.
>
> For the most part, this means that modern software has become
> incredibly bloated and just awfully bloody stupid. Dave Jones
> elaborated on the latter a bit in his OLS 2006 talk:
>
>     <https://ols2006.108.redhat.com/reprints/jones-reprint.pdf>
>
> We've done a lot of work in trying to cut through the bloat, but we've
> had to -- in parallel -- develop a software platform from scratch,
> deal with hardware unlike anything people have worked with before, and
> actually make this into a shipping product. There's still _a lot_ of
> optimization to be done, both in known paths and in places that we're
> only yet to discover.
>
> So, those of you frustrated by the current state of affairs, I urge
> you to focus your frustration towards getting out your profilers,
> debuggers, torches and pitchforks, and being as mean as you can
> possibly be to the code throughout the stack.
>
> We'll gladly take patches.
>
> --
> Ivan Krstić <krstic at solarsail.hcs.harvard.edu> | http://radian.org
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.laptop.org/pipermail/olpc-open/attachments/20071119/073af4ab/attachment.htm