Re: Green Compiler ?

"Jonathan Thornburg" <jthorn@astro.indiana.edu>
2 Jan 2013 04:09:15 GMT

          From comp.compilers
Related articles
[12 earlier articles]
Re: Green Compiler ? anton@mips.complang.tuwien.ac.at (2012-12-28)
Re: Green Compiler ? gneuner2@comcast.net (George Neuner) (2012-12-28)
Re: Green Compiler ? z80eu@arcor.de (Peter Dassow) (2012-12-29)
Re: Green Compiler ? DrDiettrich1@aol.com (Hans-Peter Diettrich) (2012-12-30)
Re: Green Compiler ? mailbox@dmitry-kazakov.de (Dmitry A. Kazakov) (2012-12-30)
Re: Green Compiler ? gneuner2@comcast.net (George Neuner) (2012-12-31)
Re: Green Compiler ? jthorn@astro.indiana.edu (Jonathan Thornburg) (2013-01-02)
Re: Green Compiler ? gah@ugcs.caltech.edu (glen herrmannsfeldt) (2013-01-02)
Re: Green Compiler ? DrDiettrich1@aol.com (Hans-Peter Diettrich) (2013-01-02)
Re: Green Compiler ? gah@ugcs.caltech.edu (glen herrmannsfeldt) (2013-01-02)
Re: Green Compiler ? gah@ugcs.caltech.edu (glen herrmannsfeldt) (2013-01-02)
Re: Green Compiler ? numerist@aquaporin4.com (Charles Richmond) (2013-01-04)
| List of all articles for this month |
From: "Jonathan Thornburg" <jthorn@astro.indiana.edu>
Newsgroups: comp.compilers
Date: 2 Jan 2013 04:09:15 GMT
Organization: Compilers Central
References: 12-12-010 12-12-012 12-12-022 12-12-028 12-12-034 12-12-037 13-01-002
Keywords: architecture, performance
Posted-Date: 02 Jan 2013 12:07:50 EST

George Neuner <gneuner2@comcast.net> wrote:
>>But registers are a very scarce resource, so that frequent loading from
>>memory is hardly avoidable.
>
> My opinions are colored by experience with DSPs, but I have long
> thought that it would be helpful to have a few K-words of non-cache
> scratchpad memory very close (1..2 cycles) to the CPU.


The Cray-2 and Cray-3 had this (I think they called it "local memory").
Alas, compilers had a lot of trouble making good use of it. As John
McCalpin put it in 1998 (at the time he was at SGI):


# Local memories work fine if an experience performance person gets
# to rewrite the code. It gets a lot more difficult if you have
# to educate a compiler to do the right thing....


I don't know if compilers have improved significantly in this area.


And on the hardware side, you'd have to figure out what to do on
context switch...
--
-- "Jonathan Thornburg [remove -animal to reply]" <jthorn@astro.indiana-zebra.edu>
      Dept of Astronomy, Indiana University, Bloomington, Indiana, USA
      "C++ is to programming as sex is to reproduction. Better ways might
        technically exist but they're not nearly as much fun." -- Nikolai Irgens
[It sounds like if we keep working, we might reinvent register windows. -John]


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.