On Fri, Mar 24, 2006 at 07:02:53AM -0700, Maurice Hilarius wrote:
> Don't confuse bandwidth with latency.
> GbE has latency that is much greater than Myrinet, old or new.
Also a good example of why comparing peak bandwidth can be misleading:
at medium-sized messages, Myrinet has much higher bandw
"Dietrich Heese-Boutin" <[EMAIL PROTECTED]> wrote:
> Any one have success running gig-e through the pci-e bus?
>
Yes, of course, However be careful as quite a few do not yet support "channel
bonding" aka "teaming" aka "Link aggregation"
Intel recalled their dual port fibre GbE NICs over this issu
Hi John,
If the application is sufficiently coarsed-grained, i.e. sufficiently
embarassingly parallel, it will work imo. It will work more like a
grid environment than a cluster, but for coursegrained work that is enough.
The main problem might be maintenance and installation: with a
wireless
On Thu, 23 Mar 2006 Karen Shaeffer wrote:
>I can say that really large earthquakes do create
> a lot of very chaotic motion.
During college I worked some summers at the HP optoelectronics
division on Page Mill road in Palo Alto. One year(1979?) we
had an earthquake that was around a 5. The groun
At 06:35 AM 3/24/2006, Eugen Leitl wrote:
From: [EMAIL PROTECTED]
- Forwarded message from "John McKelvey jmmckel%x%attglobal.net"
<[EMAIL PROTECTED]> -
From: "John McKelvey jmmckel%x%attglobal.net" <[EMAIL PROTECTED]>
Date: Fri, 24 Mar 2006 09:28:57 -0500
To: "Leitl, Eugen -id#3h6-"
It might be interesting to use one of the accellerated 802.11G systems
but it'd only give you UP TO 108 Mb/s half-duplex.
There are other options: 802.16 has some non-standard implementations
ranging up toward 80 Mb/s full-duplex and there are some gigabit
half-duplex (but *why*?) wireless ser
"Andrew D. Fant" <[EMAIL PROTECTED]> writes:
> John Hearns wrote:
>
>> Eh? Install the libraries in /usr/local/x on the head node.
>> Export this as an NFS share. Mount on all nodes, and put it into your
>> LD_LIBRARY_PATH.
>> IMHO, there is very little reason for installing APPLICATION level
From: [EMAIL PROTECTED]
- Forwarded message from "John McKelvey jmmckel%x%attglobal.net" <[EMAIL
PROTECTED]> -
From: "John McKelvey jmmckel%x%attglobal.net" <[EMAIL PROTECTED]>
Date: Fri, 24 Mar 2006 09:28:57 -0500
To: "Leitl, Eugen -id#3h6-" <[EMAIL PROTECTED]>
Subject: CCL: Wireless c
> Any one have success running gig-e through the pci-e bus?
why would you expect it to be different?
pci-e exists mainly to provide headroom (ie, 16x), as well as some
improvements to configurability (vendors can balance lanes per slot).
for low-latency NICs, pci-e shaves off some time, but that
On Fri, 24 Mar 2006, [EMAIL PROTECTED] wrote:
> In a message dated 3/23/2006 6:32:45 A.M. Pacific Standard Time,
> [EMAIL PROTECTED] writes:
> Robert Brown said everything he says is False"
>
> Nothing like the hint of paradox in morning.
>
> --
> Doug
Robert Brown me
"Robert G. Brown" <[EMAIL PROTECTED]> writes:
> Do pigeons have Buddha nature?
Coo.
--
Leif Nixon -Systems expert
National Supercomputer Centre- Linkoping University
-
11 matches
Mail list logo
