to be available here:
>>
>> http://www.csm.ornl.gov/torc/C3/
>
> We've been using pdsh:
>
> https://code.google.com/p/pdsh/
I’m a long-time user of pdsh although more recently I’ve been looking at clush
which has a lot more options and a python interface should you need i
e hpc-announce list to be lower than that
of the Beowulf list - and hpc-annouce gets a surprisingly large number of
postings.
Given that it’s easy enough for people to subscribe to hpc-announce and
anything posted would be unlikely to provoke (on-topic) discussion on the
Beowulf list I don
xisting pretty strong connotations is hurting
> more than helping at this point.
I've taken to saying I work in "Computing" as a distinct field from "IT". The
difference being that Computing is about using computers for
calculations/analytics rather than as a
d be, at least in part this is due to a "not invented
here" attitude from both sides but also commercial pressures keep a lot of the
work and algorithms secret. Just look at the number of people from HPC who
have signed on with Amazon/Google and th
rse Nasa are famous for
this but at the end of the day it's something that we've probably all done, I
don't own a DVD player any more and neglected to backup all my DVDs before it
broke. With audio tapes and vinyl I'm not so bad, the challenging one for me
would be all the
ient or you may choose to
have an entire fs tree for each client.
Another option might be to use fuse although I don't have much experience of
that myself, it's basically the same but each client would have a copy-on-write
version of /var and /etc to allow them to write to files in t
earlier on today I ran "pdsh -w [0-25] -R exec
tune2fs -O extents /dev/mapper/ost_%h" to re-tune all the devices in a lustre
filesystem.
Ashley.
--
Ashley Pittman, Bath, UK.
Padb - A parallel job inspection tool for cluster computing
http://padb.pittman.org.uk
_
cluster or traveling over the wire
somewhere to get there.
Ashley.
--
Ashley Pittman, Bath, UK.
Padb - A parallel job inspection tool for cluster computing
http://padb.pittman.org.uk
___
Beowulf mailing list, Beowulf@beowulf.org sponsored by Pe
On 26 Jan 2010, at 19:37, Paul Van Allsburg wrote:
> Ashley Pittman wrote:
>> On 25 Jan 2010, at 15:28, Jonathan Aquilina wrote:
>>> has anyone tried clustering using xen based vm's. what is everyones take on
>>> that? its something that popped into my
are
perspective it's very similar to running real hardware. For my needs
(development) it's perfectly adequate, I've not benchmarked it against running
the same code on the raw hardware though.
Ashley,
--
Ashley Pittman, Bath, UK.
Padb - A parallel job inspection
net driver.
Or the new distro you are trying enumerates the ethernet devices
differently and it's trying to load the getfile from a different
unconnected ethernet port. That's fairly common as well. It could even
be worse that than in that the enumeration could be non-deterministic to
y with the same settings will produce good output on some clusters
and bad on others. The only resource manager which seems to reliably
not mess up output is orte, that and RMS of course.
I believe most people take the route of getting rank[0] to do all the
printing.
Ashley,
--
Ashley Pittman, Bath
nce. I'm sure a case could be made
for running ten Login instances here but I'm not sure of the benefits
myself.
Ashley Pittman.
--
Ashley Pittman, Bath, UK.
Padb - A parallel job inspection tool for cluster computing
http://padb.pittman.org.uk
_
"noatime" these days?
Ashley.
--
Ashley Pittman, Bath, UK.
Padb - A parallel job inspection tool for cluster computing
http://padb.pittman.org.uk
___
Beowulf mailing list, Beowulf@beowulf.org sponsored by Penguin Computing
To change your s
of programmers time and is also likely
to make the application run slower.
Yours,
Ashley Pittman.
--
Ashley Pittman, Bath, UK.
Padb - A parallel job inspection tool for cluster computing
http://padb.pittman.org.uk
___
Beowulf mailing list, Beowulf@beowu
On Thu, 2009-06-25 at 13:09 -0500, Rahul Nabar wrote:
> On Tue, Jun 2, 2009 at 12:39 PM, Ashley Pittman
> wrote:
> Fdupes scans the filesystem looking for files where the size
> matches, if
> it does it md5's them checking for matches and if that
care about underlying performance, the traditional HPC crowd
who, lets be honest, are the ones with the money and the talent anyway.
It's as though HPC has gone or is infiltrating mainstream whilst at the
same time mainstream computing is jumping into the cloud. All of a
sudden HPC doesn
cate files.
There is another test it could do after checking the sizes and before
the full md5, it could compare the first say Kb which should mean it
would run quicker in cases where there are lots of files which match in
size but not content but anyway I digress.
Ashley Pittman.
I can't speak for SiCortex but the Quadrics news as reported on the
register is spot on, I'm surprised however that it's taken so long for
the news to filter through into the main-stream.
Ironically enough SiCortex were one of the first people I sent my CV
to :(
Ashley,
On Thu
u
know the application doesn't fit in memory and can allocate some extra
nodes to host the swapped memory, preferably swapping over the network
to RAM on a remote machine. This doubles the nodes required to run your
job however and makes scheduling it with normal jobs impossible.
Ashley Pi
with that number,
> until somebody in the OpenMPI list told me that
> "anything below 85%" needs improvement. :(
At 24 nodes that's probably a reasonable statement.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.org sponsored
six or seven years old, before then a CPU was just a
CPU and you would refer to "a N CPU cluster".
All in it can be confusing, particularly when dealing with
specifications or software which is more than one generation of hardware
old.
Ashley,
[1] Of course I'm actually referri
ve?
What features do you want? print and gdb are the most common but others
are available if you have specific requirements that these don't meet.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or u
bound and sometimes getting a huge
> load average from failed ipmitool instances hanging around.
Even when it does work running "ipmptool sensor" in-band can often take
30 seconds to complete which isn't great for performance.
Ashley,
___
Beo
nds but is horribly complex and is incredibly difficult to
get 100%. Perhaps we could talk off-list and you can tell me what I've
been doing wrong?
Ashley.
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
en you'll see the
effects at 32 nodes.
Ashley Pittman.
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
re-use a
large part of it if you were so inclined.
> Separately, does anyone here happen to know whether shmem applications
> care about independent progress? That is, if rank A is puting and
> geting
> to rank B, and rank B is off in application code, do the puts and gets
>
of metrics so that it only has to startup and send data
once rather than N times?
Ashley.
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
CPU and some kernel versions are pretty bad,
one version of Red-Hat was effectively un-usable on clusters because of
kscand.
Ashley Pittman.
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
On Mon, 2008-09-22 at 15:44 -0400, Eric Thibodeau wrote:
> Ashley Pittman wrote:
> > On Mon, 2008-09-22 at 14:56 -0400, Eric Thibodeau wrote:
> > If it were up to me I'd turn *everything* possible off except sshd and
> > ntp. The problem however is the maintenanc
erpdfs/pap301.pdf
Also look at "whatelse" from http://www.c3.lanl.gov/pal/software.shtml
Ashley Pittman.
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
but the
> guilty shall remain nameless.
You don't have to buy Dell hardware direct from Dell, there are plenty
of people who will sell you dell nodes with value-add hardware and
software.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
much higher
bandwidth figure, this would have completely defeated the point of the
benchmark in the first place however which was to show that adaptive
routing is necessary for consistent network performance.
Ashley.
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
l number of
processes in a big job sharing a node with a resource hogging job and
slow down the entire big job however I've never seen this happening in
the wild.
> Also, a poorly behaved program can cause the other codes on
> that node to crash (
a version that copies a single file,
slightly harder to do multiple files but still not rocket science.
Basically efficient broadcast isn't as easy to make generic as it seems,
why waste time even trying when you can get MPI to do all tricky bits
like work out toplogy/starting deamons/sec
within the same job
which effectively prevent there being a single optimum "site" algorithm.
AlltoAll *is* the hardest MPI function to implement well and in my view
it makes a good benchmark not just of the network but also of the MPI
stack itself, there is a good chance tha
that openmpi is being
MPI compliant in both cases.
Ashley Pittman.
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
x27;ll point
out if you are doing anything silly with MPI calls, there is enough
flexibility in the standard that you can do something completely illegal
but have it work in 90% of cases, marmot should pick up on these.
http://www.hlrs.de/organization/amt/projects/marmot/
We could take this off-l
ive directory offers.
Unfortunately with Multicast I think network bottle necks are a fact of
life and on network with static hardware configuration it really is
better to have a static software configuration as well.
What problem are you trying to solve?
Ashley Pittman.
On Mon, 2008-07-07 at 16:26 +0200
e most part
constant. In addition it used to be the case there were performance
issues associated with using zeroconf on large networks and the last
thing you want in a cluster is additional network traffic clogging up
the system.
Ashley Pittman.
are thin on the ground which is ironic as HPC
is probably the one industry where Linux has the biggest market share.
Ashley Pittman.
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
ication pattern of the application, at least in the case when you
aren't using all cpu's per node. Your MPI should pick sensible defaults
for you.
Ashley Pittman.
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (diges
oing searches tends to give a few "head in the
> sand" sites but predominantly seem to be oriented for the security
> professional.
It's no different than securing a standard Desktop machine or your
laptop. Disable as many
the be same every run, this probably won't change anything as
all numactl can do is to stabilise the results towards the bottom of the
range observed without it.
Ashley Pittman.
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
problem with this is if the head
node PXE boots on the customers network and gets automatically
re-installed as a windows workstation everybody gets egg on their face.
Yes even "modern" BIOSes are bad but localboot first is a sensible
default.
Ashley Pittman.
_
ing. If you have a specific question
I'm sure either of us could help.
>Pure software HPC is another matter. Are you looking for companies
> with products, or specific markets? Or companies that develop HPC code
> for customers?
I could name several companies in a few d
ve some codes which work better with one and some
which work better with the other.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
ood
enough to do the tests I'd have liked to have done the window was closed
and hardware technology had moved on.
Ashley.
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
we are
still a long way from it being widely used.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
roduct called f1200 so I assume
are related somehow.
The talk is on-line although I'll admit it was about half way through
before I understood what they were talking about.
http://www.cse.scitech.ac.uk/disco/mew18/Presentations/Day2/7th_Session/RobinHarker.pdf
Ashley,
g
within a node consumes CPU cycles and if your code is overlapping comms
and compute to such an extent that latency is not a large factor handing
the comms of to the nic to be handled asynchronously without CPU
intervention can improve performance.
Ashley,
xperience of in-house and then modify that
distro to meet your needs rather than the other way around.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
On Tue, 2007-07-24 at 14:37 +, Patrick Ohly wrote:
> On Tue, 2007-07-24 at 15:15 +0100, Ashley Pittman wrote:
> [examples for the need of a more accurate clock]
> > But none of the ones you list are more than vaguely related to HPC.
> [...]
> > The only thing I've fou
;d use it if it was packaged with the distro and just worked
out the box (without using multicast preferably) but it's absence isn't
enough to cause me problems.
Ashley.
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscr
n the kernel, 5.x versions of
score used a kernel patch, 6.x versions of score patch the network card
drivers themselves as this is easier approach to manage.. Using
different kernels isn't hard but it does require patching code.
Ashley,
___
Beowulf
st and I've seen
procurement contracts which state that any results obtained by the
computer must be 100% repeatable which in effect means the answer to
your question is no, they cannot differ.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.or
is jobs tend to be 32-128p,
> > and run for a week, so it's not ideal to run them under the debugger.
It really shouldn't be that difficult, on a Quadrics cluster at least
you can use the command "padb -x -r " from anywhere in the
clu
On Wed, 2007-04-25 at 07:35 -0700, Christian Bell wrote:
> On Wed, 25 Apr 2007, Ashley Pittman wrote:
>
> > You'd have thought that to be the case but PIO bandwidth is not a patch
> > on DMA bandwidth. On alphas you used to get a performance improvement
> > by evi
On Wed, 2007-04-25 at 11:31 +0200, Håkon Bugge wrote:
> At 17:55 24.04.2007, Ashley Pittman wrote:
> >That would explain why qlogic use PIO for up to 64k messages and we
> >switch to DMA at only a few hundred. For small messages you could best
> >describe what we use as
which seems a little fast to me.
Regardless of how they have done it 1.2 is impressive, what would make
me even more impressed if it was quoted as 1.20 which would, as far as
I'm aware, mean that they had the lowest latency of anybody.
Ashley,
__
On Fri, 2007-04-13 at 01:14 +0900, Naoya Maruyama wrote:
> On 4/12/07, Ashley Pittman <[EMAIL PROTECTED]> wrote:
> > My advice would be first and foremost to look at the core file, I assume
> > your program is receiving a SEGV and exiting? core files can be
> > problem
uld be to set MALLOC_CHECK_=2 to enable integrity
checking in the libc malloc implementation and if using ia64 download
compile gdb from source otherwise you might find it's not all that
accurate at times.
TotalView and DDT are both great if you have a licence for eithe
80 9770.92 10413.19 10318.85
> 1048576 40 18792.78 20762.40 20533.59
> 2097152 20 33849.45 42141.25 41535.32
> 4194304 10 65966.61 81472.99 79850.54
Why is 512 so much quicker than 8? The reduce figures showed the same
issue and it's something I'd want to look at.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
Christian Bell wrote:
> On Wed, 04 Apr 2007, Ashley Pittman wrote:
>> GasNet does however get extra credit for having a asynchronous
>> collective, namely barrier. Unfortunately when you read the spec it's
>> actually a special case asynchronous reduce which is almost
Richard Walsh wrote:
> Ashley Pittman wrote:
>> Patrick Geoffray wrote:
>>
>>> I would bet that UPC could more efficiently leverage a strided or vector
>>> communication primitive instead of message aggregation. I don't know if
>>> GasNet provides
ective, namely barrier. Unfortunately when you read the spec it's
actually a special case asynchronous reduce which is almost impossible
to optimise anything like as well as barrier which is a shame.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf
systems they sell, buy a Quadrics system
from HP and you get HP-MPI by default. I don't think it's because their
MPI is necessarily any better than our MPI on the particular platform,
it's just that it's the HP-MPI and it's a HP platform.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
It's probably a good thing to benchmark to get a
idea of the capability of a given network.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
C..
I was the last year of intake to be taught Ada when I started in 1996,
there was a "C and Unix" module in the second year which I'm fairly sure
was compulsory. These days I program mostly in c although can at least
read fortran. They switched from A
eresting toy cluster, the constraints would certainly
sharpen the mind.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
w you to experiment with
most things. Of course you are only two nodes away from having 16 cpus
and 16 is the next step up the list, what was your budget again...
Of course when you start doing real things rather than just interesting
things it really does depend, not least on what the real thi
s not a new idea, IIRC PSC were doing this six or seven
years ago, I'd be interested to see if hyperthreading helps the
situation, it's almost always turned of and any cluster over 32 CPU's
but it might be advantageous to enable it and use something like cpuset
> And got it. The title is:
>
>"The Case of the Missing Supercomputing Performance"
I wondered if you were talking about that paper but it's from lanl not sandia,
it should be essential reading for everyone working with
On Thu, 2007-01-18 at 08:17 -0600, Richard Walsh wrote:
> Ashley Pittman wrote:
> > On Wed, 2007-01-17 at 08:50 +0100, Mikael Fredriksson wrote
> >> Yes, it is. And more so if this cluster/LAN can also utilize som type
> >> of "MOSIX" system. This will su
their parts the
parts are mostly that of a bog standard Linux distribution.
I suppose it could be true that changing to OS to Windows would make
them less specialised however that probably says more about Windows than
it does about "hard-core&qu
;t imagine hot liquid would be any different however.
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
out
interrupts is 2.72uSec, using interrupts it is 7.20uSec. One
fundamental difference between these two measurements is that when you
use interrupts the kernel has to get involved, without interrupts it
doesn't so you don't just have the interrupt but also a extra syscall.
I know of p
mpich2 header file) anybody can
create a DLL which is binary compatible and the app won't notice. For
this to work of course Microsoft would probably need to release their
source changes to mpich2 under the BSD Licence.
Intel use a dynamic layer underneath or inside MPI which allow
ll create/mandate the model by
> supplying the MPI. Some standards are defined after the fact.
>
> I am not advocating mimicing the Microsoft ABI. I am advocating getting
> a single MPI ABI per ISA ABI. The question of course is, which one.
Aren't these two statements contra
On Mon, 2006-06-12 at 11:18 -0400, Joe Landman wrote:
>
> Ashley Pittman wrote:
>
> >> More to the point, this dynamic binding allows you to write to the API,
> >> present a consistent ABI, and handle the hardware details elsewhere in a
> >> driver which can
On Mon, 2006-06-12 at 10:49 -0400, Joe Landman wrote:
>
> Ashley Pittman wrote:
> > On Mon, 2006-06-12 at 00:02 -0400, Joe Landman wrote:
> >
> >> What Microsoft will do is to take away as much of this as they can. I
> >> haven't seen it yet, but I beli
s at runtime, and just have it work. This is a nice idea.
Perhaps I've missed something here, what do windows DLLs provide that a
linux .so doesn't?
Ashley,
___
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode
82 matches
Mail list logo
