On 03/22/06 21:54, Gavin Maltby wrote:
On 03/20/06 19:09, [EMAIL PROTECTED] wrote:
FMA looks like it's about 100k/cpu (ERPT_MAX_ERRS * max_ncpu *
ERPT_DATA_SZ)
And they are all preallocated? Sounds like a bug to me.
Yes they are, so that we can fill error reports from any context
without
On 03/20/06 19:09, [EMAIL PROTECTED] wrote:
FMA looks like it's about 100k/cpu (ERPT_MAX_ERRS * max_ncpu * ERPT_DATA_SZ)
And they are all preallocated? Sounds like a bug to me.
Yes they are, so that we can fill error reports from any context
without worrying sbout kmem allocation, pil
Just when you think the big companies have all the
toys (enterprise servers), you find companies like
this making 16-cpu servers:
http://www.kraftway.com/products/g-scale_6016.html
(Kraftway G-Scale 6016)
also the IBM x455
http://www-03.ibm.com/servers/eserver/xseries/x455.html
Also the NEC
Well, if max_ncpus was just properly set to the actual maximum, as Andrei's
follow-on is done correctly, they wouldn't be allocating all this
extra space. Historically, SPARC has always done so, but x86 hasn't known
how many CPUs it had early enough in the boot process to set it correctly.
It
On Sun, Mar 19, 2006 at 10:38:36AM +0100, [EMAIL PROTECTED] wrote:
I didn't try this on a laptop, but here are some numbers from a 2-way
AMD system running in 64-bit mode showing how much memory gets used by
the
kernel in each case.
NCPUmax_ncpus kernel
64 2
On Sun, Mar 19, 2006 at 10:38:36AM +0100, [EMAIL PROTECTED] wrote:
I didn't try this on a laptop, but here are some numbers from a 2-way
AMD system running in 64-bit mode showing how much memory gets used by
the
kernel in each case.
NCPUmax_ncpus kernel
64 2
Ian Collins wrote:
Andrei Dorofeev wrote:
I didn't try this on a laptop, but here are some numbers from a 2-way
AMD system running in 64-bit mode showing how much memory gets used by
the
kernel in each case.
NCPUmax_ncpus kernel
64 2 227MB
21 21
On Mon, Mar 20, 2006 at 08:09:43PM +0100, [EMAIL PROTECTED] wrote:
On Sun, Mar 19, 2006 at 10:38:36AM +0100, [EMAIL PROTECTED] wrote:
I didn't try this on a laptop, but here are some numbers from a 2-way
AMD system running in 64-bit mode showing how much memory gets used by
the
Roland Mainz wrote:
Ian Collins wrote:
Andrei Dorofeev wrote:
I didn't try this on a laptop, but here are some numbers from a 2-way
AMD system running in 64-bit mode showing how much memory gets used by
the
kernel in each case.
NCPUmax_ncpus kernel
64 2
I didn't try this on a laptop, but here are some numbers from a 2-way
AMD system running in 64-bit mode showing how much memory gets used by
the
kernel in each case.
NCPUmax_ncpus kernel
64 2 227MB
21 21 231MB - stock Nevada bits
64 32
Andrei Dorofeev wrote:
Hi Roland,
I didn't try this on a laptop, but here are some numbers from a 2-way
AMD system running in 64-bit mode showing how much memory gets used by
the
kernel in each case.
NCPUmax_ncpus kernel
64 2 227MB
21 21 231MB
What about single processor AMD64-based laptops? I think there will be a
lot of wasted memory in that case.
How much are we talnking about? Quantify it and then we can talk.
Casper
___
opensolaris-discuss mailing list
Roland Mainz wrote:
BTW: How do other version of Unix (like FreeBSD/NetBSD) handle the
problem ?
The theoretical limit for NetBSD is 32, because some bitmasks are used
for IPI handling, etc.
I think it can be bumped to 64 without too much problems, by changing
the type of those bitmasks
Roland Mainz wrote:
BTW: How do other version of Unix (like
FreeBSD/NetBSD) handle the
problem ?
The theoretical limit for NetBSD is 32, because some
bitmasks are used
for IPI handling, etc.
I think it can be bumped to 64 without too much
problems, by changing
the type of those bitmasks to
Yet in research labs (a lot of clustered solutions)
and automotive fields, the Linux kernel has a dominant
role (not necessarily the RH distro either).
The processor count in clustered systems is irrelevant;
clustering is not affected by the single CPU limit.
I think the arguement is still
On Sat, Mar 18, 2006 at 12:38:27PM +0100, Frank van der Linden wrote:
But, touting the limit is kinda pointless if the current limit is well
below it.. I've never run NetBSD/amd64 on more than 4CPU/16G, and it
might blow up spectacularly when run on 32CPU hardware, should it become
[EMAIL PROTECTED] wrote:
Yet in research labs (a lot of clustered solutions)
and automotive fields, the Linux kernel has a dominant
role (not necessarily the RH distro either).
The processor count in clustered systems is irrelevant;
clustering is not affected by the single CPU limit.
Andrei Dorofeev wrote:
Are there any plans to support more than 64 CPUs on AMD64 kernels
anytime soon (e.g. add theoretical support for such configurations) ?
Yes, we have plans to make it possible to raise NCPU above 64. No
specific schedule at this point.
BTW: I just checked qemu - it
Hi Roland,
Yes, we have plans to make it possible to raise NCPU above 64. No
specific schedule at this point.
BTW: I just checked qemu - it seems to support up to 255 x86 CPUs. May
be something worth for testing the boot sequence on very large x86
systems.
Yep, I'm well aware of that.
Andrei Dorofeev wrote:
Yes, we have plans to make it possible to raise NCPU above 64. No
specific schedule at this point.
BTW: I just checked qemu - it seems to support up to 255 x86 CPUs. May
be something worth for testing the boot sequence on very large x86
systems.
Yep, I'm
Hi Roland,
xx@@@!!!... ;-(
Are the qemu people aware of the problem ?
We don't know yet who is at fault here. If it is QEMU, then we'll
tell the qemu people about it.
- Andrei
___
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
Yet in research labs (a lot of clustered solutions)
and automotive fields, the Linux kernel has a
dominant
role (not necessarily the RH distro either).
The processor count in clustered systems is
irrelevant;
clustering is not affected by the single CPU limit.
I think the arguement is still
Andrei Dorofeev wrote:
I remember the discussion very good. I am complaining that Red Hat is
the technology leader now. Sun had half a year to deal with the
problem. Why is it so difficult to change the number from 21 to 64 as
proposed in the previous discussion?
Fair enough. I
Felix Schulte wrote:
On 3/14/06, Gavin Maltby [EMAIL PROTECTED] wrote:
On 03/14/06 12:19, Felix Schulte wrote:
Good morning!
Just saw
https://www.redhat.com/archives/nahant-list/2006-March/msg00049.html
- Red Hat now supports 64 cpus on AMD64, Sun only 21. Why is Sun
lagging
Hi Roland,
Are there any plans to support more than 64 CPUs on AMD64 kernels
anytime soon (e.g. add theoretical support for such configurations) ?
Yes, we have plans to make it possible to raise NCPU above 64. No
specific schedule at this point.
IMO it could be tricky if this manual tuning
25 matches
Mail list logo
