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

Re: [adam@flounder.net: Re: testing and no release schedule]

On Tue, Mar 30, 2004 at 09:22:38PM +1100, Russell Coker wrote:
> On Tue, 30 Mar 2004 20:03, Ingo Juergensmann <ij@2004.bluespice.org> wrote:
> > > I wouldn't run two compiles at the same time on one of the original
> > > Pentium machines either.
> > Why not? It's fast enough?
> Old machines simply can't manage enough memory to cache two large source 
> trees.  The original pentium machines had limitations of about 64M (if you 
> wanted the memory to be cached that is).

Well, be happy that you have L2 Cache at all. m68k has no L2. :)

> > Granted, arrakis has 7 SCSI disks on two different hostadapters with 5
> > distributed swap partitions that are making swapping wonderful smooth, even
> > if some evil perl tasks use >590 MB of memory.
> Quite impressive.  Such a machine should be able to sustain several builds at 
> the same time.

At least I have no problems with parallel builds on arrakis, although it is
supposed to be way slower than any mips build by nature... ;)

> > Sure, parallel builds don't speed up things, but they don't necessarily
> > need to slow down things either, when you do it right.
> True.

Oh, parallel builds *can* speed up things, like it was explained before with
that blocking I/O argument. Just for the logs, that I'm aware of this... ;)

> > Well, if someone would ask me, I'd vote for removing this strange "this is
> > a buildd, please don't use it for manual builds"-motd policy. It does not
> > make any sense to me. When the buildd admin thinks, it will slow down the
> > machine, he could get some more machines to broaden the load. And I don't
> > think that f.e. mips has a hardware problem at all...
> Fair points.

So, who files the GR then? It is GR season now, isn't it? (-:

Ciao...              // 
      Ingo         \X/

Reply to: