On Tue, 15 Aug 2006, Mike Davis wrote:
Mark Hahn wrote:
huh? what value does big-A have to add here? the correct queueing system
is
the one that is cheap, low-maintenance, efficient, easy to use, etc. those
are things that users and sysadmins know, not behind-desk-sitters...
Difference of
Mike Davis wrote:
> I'm not 100% sure about that Mark. I care about big-A administration. I
> care about showing departments what resources are actually available. I
> care about what is the most efficient use of limited University
> resources. When I meet with researchers they often say that the
Mark Hahn wrote:
huh? what value does big-A have to add here? the correct queueing
system is
the one that is cheap, low-maintenance, efficient, easy to use, etc.
those are things that users and sysadmins know, not behind-desk-sitters...
Difference of definition here. I believe that Big-A ad
Jim Lux wrote:
> At 08:41 AM 8/11/2006, Geoff Jacobs wrote:
>
> Sure you would.. Actually, these days, you might use an IGBT, depending
> on the switch rate..
Well, it was just a guess based on the bare pcb of PicoPSU supplies.
> Switching PSUs all rely on rectifying the AC supply to generate a
I'm not 100% sure about that Mark. I care about big-A administration. I care
about showing departments what resources are actually available. I care about
what is the most efficient use of limited University resources. When I meet
with researchers they often say that they had no idea that there
At 02:00 PM 8/15/2006, Geoff Jacobs wrote:
Jim Lux wrote:
> At 08:41 AM 8/11/2006, Geoff Jacobs wrote:
>
> Sure you would.. Actually, these days, you might use an IGBT, depending
> on the switch rate..
>
> Switching PSUs all rely on rectifying the AC supply to generate a DC bus
> voltage, that i
I'm not 100% sure about that Mark. I care about big-A administration. I
care about showing departments what resources are actually available. I
care about what is the most efficient use of limited University
resources. When I meet with researchers they often say that they had no
idea that there
I've sent Bill a link to this Educause group.
http://www.educause.edu/content.asp?page_id=6673&bhcp=1
It is one of the IT management lists and its misssion covers many of the
ideas that we are discussing.
Personally, I'm OK no matter what the ultimate decision is
but some readers may not be
Mark Hahn wrote:
> IMO, centralization breeds contempt ;)
[virtual coffee splatters real screen]
[]
> that said, it's entirely possible to sustain a "rolling cluster": start
> with one generation, and incrementally move it forward. this is easiest
> if you have standard parts (plain old
beowulf traffic itself is "noise"? If you are thinking of a "list for
university deans" or members of research support offices or departmental
...
administerable and accountable should they get audited) -- then yeah, I
think a new list or other venue would be very useful.
yes. the overlap is
Jim Lux wrote:
> At 08:41 AM 8/11/2006, Geoff Jacobs wrote:
>
> Sure you would.. Actually, these days, you might use an IGBT, depending
> on the switch rate..
>
> Switching PSUs all rely on rectifying the AC supply to generate a DC bus
> voltage, that is then converted to the DC voltage you want
On Tue, 15 Aug 2006, Joe Landman wrote:
Now that RGB has written a thesis on this ... :)
No, the thesis went to Bill offline. You guys just got the
synopsis...;-)
And I'm too tired/busy from my "vacating" to do a proper job, sorry...
Work starts in earnest next week, and a whole lot of it st
- integration of a cluster into a larger University IT infrastructure
(storage, authentication, policies, et. al.)
just say no. we consciously avoid taking any integration steps that
would involve the host institution trusting us or vice versa. well,
not quite - we managed to live for ~5 yea
Now that RGB has written a thesis on this ... :)
Robert G. Brown wrote:
> On Tue, 15 Aug 2006, Bill Rankin wrote:
>
>>> Are there problems so specific to the higher education realm
>>> that you think they'd benefit from their own forum?
>>
>> Not so much that would benefit from their own forum,
On Tue, 15 Aug 2006, Bill Rankin wrote:
Are there problems so specific to the higher education realm
that you think they'd benefit from their own forum?
Not so much that would benefit from their own forum, but there are a lot of
issues that are not directly related to the construction and te
On Aug 14, 2006, at 12:28 AM, Brian Dobbins wrote:
Echoing Mark's reply, it seems to me that a lot of the volume IS from
people who are in an academic environment and wrestle with the issues
therein. My two cents would be that I think the wide variety of
experience and ideas here only helps whe
The best pathscale fortran stream results for Opteron, at least for 1
core, about 1 year ago were obtained for
-CG:use_prefetchnta -LNO:prefetch_ahead=4 -O3 -mp
keys of compilation.
May be there is some sense to play w/this parameters for Woodcrest
also ?
Yours
Mikhail Kuzminsky
Zelinsky Inst
Peter Kjellstrom wrote:
[EMAIL PROTECTED] streamd]# hostname ; date ; for i in 1 2 3 4 5 ; do export
OMP_NUM_THREADS=$i ; ./streamd | egrep "Total memory re|Number of Th|Function
|Copy:|Scale:|Add:|Triad:"; done
tbox3
Fri Aug 11 17:59:22 CEST 2006
Total memory required = 457.8 MB.
Number of Th
Interesting...
Given that Add and Triad are virtually the same
it's surprising that Copy and Scale are so different.
IMHO Scale should be more like Copy. Compiler effect?
> here you go (dell 2950 with 8 modules and streams compiled with icc-9.1 -O3:
>
> [EMAIL PROTECTED] streamd]# hostname ; dat
On Tuesday 15 August 2006 17:25, Richard Walsh wrote:
> Mark Hahn wrote:
> >>> Good point which makes perfect sense to me.
> >>> Given that the theoretical maximum is actually 21.3 GB/s
> >>> the real maximum Triad number must be 21.3/3 = 7.1 GB/s.
> >
> > I don't get this - triad does two reads an
Oops. I misinterpreted what Keith was saying.
I thought he was trying to justify the "one third" empirical
rule by referring that every read needs to check against
the other socket thus effectively reducing the bandwidth by 3.
It's obvious that the "three quarters" rule which worked perfectly
for
Mark Hahn wrote:
Good point which makes perfect sense to me.
Given that the theoretical maximum is actually 21.3 GB/s
the real maximum Triad number must be 21.3/3 = 7.1 GB/s.
I don't get this - triad does two reads and one write.
if you don't use store-through ('nt' versions of mov),
then the w
Good point which makes perfect sense to me.
Given that the theoretical maximum is actually 21.3 GB/s
the real maximum Triad number must be 21.3/3 = 7.1 GB/s.
I don't get this - triad does two reads and one write.
if you don't use store-through ('nt' versions of mov),
then the write also implies
On Tue, Aug 15, 2006 at 12:29:02PM +0100, Kozin, I (Igor) wrote:
>
> Good point which makes perfect sense to me.
> Given that the theoretical maximum is actually 21.3 GB/s
> the real maximum Triad number must be 21.3/3 = 7.1 GB/s.
> And that's the best number I've heard of.
Then how do you explai
Kozin, I (Igor) wrote:
[...]
> Here is a pointer to some measured latencies
> http://www.anandtech.com/IT/showdoc.aspx?i=2772&p=4
Hmmm. The text was littered with the fluff marketing bits, about how
this is devastating to Opteron and all that. Sounded quite a bit like
it was written by Intel
Good point which makes perfect sense to me.
Given that the theoretical maximum is actually 21.3 GB/s
the real maximum Triad number must be 21.3/3 = 7.1 GB/s.
And that's the best number I've heard of.
Here is a pointer to some measured latencies
http://www.anandtech.com/IT/showdoc.aspx?i=2772&p=4
26 matches
Mail list logo
