Anon (no@email.com) on 1/10/11 wrote:
>
>If we judged all technologies on the first quick-and-dirty hack that got to market,
>the world would not be a better place, dont you think?

Umm. The whole transcoding-with-CUDA thing isn't exactly
new, is it?

So please do show the numbers for the good transcoders.
Because all I ever see is the ones that generate clearly
inferior end results. And those are the performance numbers
that people then quote..

Because it's entirely possible that the performance of the
better-quality encoders aren't all that stunning.

Anyway, my own personal gripe with parallel computing ends
up just being "it's hard". And I've seen too many of the
"it's too hard to do it right, so I'll quote numbers for
the code that cuts all the corners" to be all that surprised
any more.

So instead of trying to dispute the indisputable (did you
even read the performance articles that were quoted
in this thread? Did you even read about the bad
quality of the CUDA engine?), why don't you point people
to the numbers and quality comparisons for those good
engines that you claim exist?

Linus