MoTheG (better@not.tell) on 11/9/09 wrote:
---------------------------
>David Kanter (dkanter@realworldtech.com) on 11/8/09 wrote:
>
>> Or caches for that matter.
>
>hm, but don't GPUs use caches just as well?
>
>> Many workloads are very difficult to parallelize or
>> vectorize with a compiler. Think GCC or most latency
>> sensitive applications.
>
>vectorisation will only be possible for certain media and science applications
>thus GPUs will continue to be only applicable to these. GPUs can react to the data
>nowerdays by going sequential. GCC itself should be parallizeable.
>For GCC I'm thinking parsing/IO could be a thread and many parts can be compiled
>seperatly just as I can precompile parts. Certainly no task for a GPU but many cores.

... it seems to me that you think you can just randomly join arbitrary words together to form a sentence and call the result a profound idea ...

Please stop doing it!