[prev in list] [next in list] [prev in thread] [next in thread] 

List:       pcc-list
Subject:    Re: [Pcc] Things to add for next release
From:       Alan Cox <alan () lxorguk ! ukuu ! org ! uk>
Date:       2014-10-18 18:56:27
Message-ID: 20141018195627.10f5f777 () www ! etchedpixels ! co ! uk
[Download RAW message or body]

> things are easier now (register usage has, on the whole, gotten much more
> uniform) and some things are worse (the timing models in today's CPU's are
> baroque and bizarre beyond reasonable description...).
> 

They reflect the way the hardware really works that's all. On the other
hand increasingly the input doesn't matter as much. If you throw a bunch
of instructions at the CPU in a reasonably sane order it'll probably do a
reasonable job at extracting parallelism. It has to because the rules
change each CPU and nobody is going to recompile their OS for each
specific core unless they are running a really huge HPC setup or they run
Gentoo Linux.

Alan
_______________________________________________
Pcc mailing list
Pcc@lists.ludd.ltu.se
http://lists.ludd.ltu.se/cgi-bin/mailman/listinfo/pcc
[prev in list] [next in list] [prev in thread] [next in thread]