Re: Dust Collection article for gcc.gnu.org

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



John!

On 08.03.21 13:47, John Hamlin wrote:
> ​Hello there, 
>  My name is John and I would like to chat about how I can help gcc.gnu.org. I would like to find out about your company goals and write some informative content for you to put on your website. 
>  I can do it all for free.

What a true blessing!

It's been annoying me for years that the better gcc gets, the more dust
it emits. This is caused by aggressive optimization shaving off
instructions from the code. Where do these instructions go? Well, into
the air of course.

What is exceptionally problematic about this is that recent publications
have shown that especially the incredibly tiny single-cycle opcodes form
respirable instruction dust which can cause both short- and long-term
consequences if the compiler process is not ventilated properly.

My only concern/question would be if the blowers that transport away
these hazardous particles are small enough to be incorporated on-chip
silicon level. For instance, it could be laid out so that only a few
cores on a die are outfitted with nanoscale centrifugal blowers and
through software gcc could be told to run only on these ventilated
cores. My guess would be that an industrial blower with a blade diameter
of 50-100nm should be fully sufficient to vent a core that is 100%
occupied with a gcc process.

In the meantime, I will look into OSHA requirements; unventilated
compiling is something that should have been outlawed years ago.

Stay safe,
All the best,
Johannes

-- 
"A PC without Windows is like a chocolate cake without mustard."



[Index of Archives]     [Linux C Programming]     [Linux Kernel]     [eCos]     [Fedora Development]     [Fedora Announce]     [Autoconf]     [The DWARVES Debugging Tools]     [Yosemite Campsites]     [Yosemite News]     [Linux GCC]

  Powered by Linux