Wichert Akkerman [EMAIL PROTECTED] writes:
There is more out there than just i386 system. I would love to see
you attach a 100 gigabyte disk to a m68k system.
Looking at several m68k's here, they have SCSI-2 connectors. Wouldn't
it be feasible to connect an IDE-RAID system to them?
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On Tuesday 02 April 2002 06:17, Ben Collins wrote:
On Tue, Apr 02, 2002 at 03:48:49AM +0200, Paul Russell wrote:
Looking at my testing PPC box with grep-available, we have only about
8GB total Installed-Size. So I would expect a ccache of 1GB
Paul Russell [EMAIL PROTECTED] cum veritate scripsit:
If this is indicative, a complete debian build would use 24GB, and I
would say that a cache of a few GB would be a win (given not all
packages get built as frequently).
Also, a build farm could be optimized to usually build the same
On Mon, 2002-04-01 at 23:17, Ben Collins wrote:
Looking at my testing PPC box with grep-available, we have only about
8GB total Installed-Size.
glibc packages total installed size is only a few dozen megs. However,
the source builds takes up about 600megs. XFree86, about 1.6gigs.
glibc's
On Tue, Apr 02, 2002 at 03:17:45AM -0500, Anthony DeRobertis wrote:
On Mon, 2002-04-01 at 23:17, Ben Collins wrote:
Looking at my testing PPC box with grep-available, we have only about
8GB total Installed-Size.
glibc packages total installed size is only a few dozen megs. However,
On Tue, 2 Apr 2002 10:17, Anthony DeRobertis wrote:
On Mon, 2002-04-01 at 23:17, Ben Collins wrote:
Looking at my testing PPC box with grep-available, we have only about
8GB total Installed-Size.
glibc packages total installed size is only a few dozen megs. However,
the source builds
That's all of three 100GB IDE disks running in RAID 0. Four disks if for
some reason you want redundancy on your cache.
Surely you don't presume that a) All of our autobuilders have enough
bays for 3 IDE disks and b) Can take IDE at all (vore, the sparc buildd,
doesn't have any IDE).
Of
On Tue, 2002-04-02 at 11:17, Russell Coker wrote:
So we firstly need to find a real slow arch which also supports 4 new large
IDE disks (remember that machines 3 years old tend not to have good support
for 32G drives).
That's primarily a BIOS problem, right? Does it matter for Linux?
Previously Anthony DeRobertis wrote:
That's primarily a BIOS problem, right? Does it matter for Linux?
There is more out there than just i386 system. I would love to see
you attach a 100 gigabyte disk to a m68k system.
Wichert.
--
On Tue, Apr 02, 2002 at 02:30:03PM -0500, Anthony DeRobertis wrote:
That's all of three 100GB IDE disks running in RAID 0. Four disks if for
some reason you want redundancy on your cache.
Surely you don't presume that a) All of our autobuilders have enough
bays for 3 IDE disks and b)
On Tue, Apr 02, 2002 at 03:10:14PM -0500, Ben Collins wrote:
The ones that could use it, are the ones that can't take it :) I seriously
doubt the m68k can hold 4 disks. I'm pretty sure the m68k's run scsi too.
Perhaps architectures which are slow enough to really benefit would still
benefit if
On Tue, Apr 02, 2002 at 02:35:06PM -0500, Anthony DeRobertis wrote:
That's primarily a BIOS problem, right? Does it matter for Linux?
Actually, current kernels have a problem with current 160G drives.
Mainstream 48bit ide support is expected Real Soon Now. :)
--
Mike Stone
--
To
On Tue, Apr 02, 2002 at 03:38:18PM -0500, Matt Zimmerman wrote:
Perhaps architectures which are slow enough to really benefit would still
benefit if they had to go out over the network to a cache? Certainly if
there was other storage on the same LAN, maybe even farther.
Unfortunately the
On Tue, 2 Apr 2002 21:35, Anthony DeRobertis wrote:
On Tue, 2002-04-02 at 11:17, Russell Coker wrote:
So we firstly need to find a real slow arch which also supports 4 new
large IDE disks (remember that machines 3 years old tend not to have good
support for 32G drives).
That's primarily a
On Monday 01 April 2002 18:23, Junichi Uekawa wrote:
Martin Schulze [EMAIL PROTECTED] cum veritate scripsit:
The same package: almost never
the same file: often, with every new compile.
Just take into account that a package contains 50 .c files that need
to be compiled. An updated
On Tue, Apr 02, 2002 at 03:48:49AM +0200, Paul Russell wrote:
On Monday 01 April 2002 18:23, Junichi Uekawa wrote:
Martin Schulze [EMAIL PROTECTED] cum veritate scripsit:
The same package: almost never
the same file: often, with every new compile.
Just take into account that a
16 matches
Mail list logo