Re: 1teraflops cell processor possible?

Russell Wallace <[EMAIL PROTECTED]> wrote:

> Nor have we any clue about whether we might find them next year, next
> decade or next century. So by all means let's keep looking for better
> algorithms, but in the meantime we have two options:

> 1) Let people keep dying by the millions from diseases like cancer,
> AIDS and malaria that we might be able to find improved treatments for
> if we know more about the molecular machinery involved.

> 2) Spend a few billion out of our civilization's vast economic surplus
> cranking more speed out of the algorithms we do have.

> Option 2 strikes me as vastly superior.

I don't think there are any current issues in cancer, AIDS, or malaria
treatment that would be solvable by throwing more cycles at these
problems.  These are not fundamentally computational problems, or, at
least, our understanding is not currently limited by our computational
resources.  A few billion dollars well spent on AIDS and malaria
treatment or even on basic research might well save millions of lives.
Spending that money on a faster computer, today, is not going to save
anyone.

-- Dave