Hardware only takes you so far. Scalability comes largely from the efficiency of the software. Poor software results in large amounts of communication between nodes, slowing down a cluster.
Agree. But sometimes, no matter how good your software is, the problem is inherently hard to parallelize, and you end up with humongous communication requirements.
$./bubrawatch.py -v
ns1.bubra.biz is 81.203.73.17 (81-203-73-17.user.ono.com)
ns2.bubra.biz is 80.138.221.95 (p508ADD5F.dip.t-dialin.net)
ns3.bubra.biz is 80.11.243.45 (AMarseille-102-1-2-45.w80-11.abo.wanadoo.fr)
ns4.bubra.biz is 80.46.141.109 (dsl-80-46-141-109.access.uk.tiscali.com)
ns5.bubra.biz is 82.65.110.228 (lns-p19-16-82-65-110-228.adsl.proxad.net)
$
bubra.biz just handles the nameserver stuff, not web hosting.
> 4. Hack the trojan - harvest the addresses of the spammers' web sites from the data feed.
Actually, I've done that. I have in my hand the IP number of the originating server. I have tried to get the interest of CERT, FBI and my local police authorities. No luck.
Well, they claim 9.2 teraflops theoretical peak performance, and that will probably still yield something like 4-5 teraflops actual sustained performance, as measured by standard LINPACK benchmark.
My guess is that the phone was killed by software
glitches; that they finally realized that the
phone was too buggy to release.
A quick google shows that the Z100 project has
been a long and painful journey.
The phone was to have been released last year,
but was delayed because of trouble with GPRS. It
was then supposed to have been released Q1 this
year, but in March it was announced that it would
be delayed until summer because of software glitches. And now they finally killed their darling.
Yeah. There's actually quite a lot of research going into this currently. It's called the Grid (think "power grid", ubiquitous, simple to use), and I predict it will be the next big buzzword.
The problem of buying and selling computation power on some sort of broker basis is a quite interesting problem in itself. Exactly what are you selling? Hardly CPU hours, since the value of those depends on the hardware.
Slashdot Mirror
Agree. But sometimes, no matter how good your software is, the problem is inherently hard to parallelize, and you end up with humongous communication requirements.
Then you need clever hardware engineers.
NEWater looks like any other glacier-clear bottled H20.
Ehm. Even seen meltwater from a glacier? It's blue-green and opaque from silt.
You can't compare those figures. The 65 Tflops is estimated peak speed, the 36 Tflops is measured sustained speed on LINPACK.
Actually, bubra.biz seem to be doing fine;
$ ./bubrawatch.py -v
ns1.bubra.biz is 81.203.73.17 (81-203-73-17.user.ono.com)
ns2.bubra.biz is 80.138.221.95 (p508ADD5F.dip.t-dialin.net)
ns3.bubra.biz is 80.11.243.45 (AMarseille-102-1-2-45.w80-11.abo.wanadoo.fr)
ns4.bubra.biz is 80.46.141.109 (dsl-80-46-141-109.access.uk.tiscali.com)
ns5.bubra.biz is 82.65.110.228 (lns-p19-16-82-65-110-228.adsl.proxad.net)
$
bubra.biz just handles the nameserver stuff, not web hosting.
> 4. Hack the trojan - harvest the addresses of the spammers' web sites from the data feed.
Actually, I've done that. I have in my hand the IP number of the originating server. I have tried to get the interest of CERT, FBI and my local police authorities. No luck.
Meanwhile, the spamming continues.
Hmm, are they really going for OGSA? Wonder if they'll write their own implementation or build on the Globus toolkit like everybody else.
Problem is, Globus 2 is notoriously buggy, and Globus 3, which supports OGSA, is written in Java and is really slow.
Somehow I don't see millions of users on a platform like that.
Ah, well, no. I recently installed a Beowulf
cluster where the nodes had neither floppy drives,
nor CD:s.
Just a matter of PXE booting them into a network
installation. No sweat.
Even if this laptop might not support PXE, I'm sure
it would be rather trivial to reinstall it.
Well, they claim 9.2 teraflops theoretical peak
performance, and that will probably still yield
something like 4-5 teraflops actual sustained
performance, as measured by standard LINPACK
benchmark.
My guess is that the phone was killed by software glitches; that they finally realized that the phone was too buggy to release.
A quick google shows that the Z100 project has been a long and painful journey.
The phone was to have been released last year, but was delayed because of trouble with GPRS. It was then supposed to have been released Q1 this year, but in March it was announced that it would be delayed until summer because of software glitches. And now they finally killed their darling.
Yeah. There's actually quite a lot of research going into this currently. It's called the Grid (think "power grid", ubiquitous, simple to use), and I predict it will be the next big buzzword.
See Global Grid Forum, Grid Today and the Globus project for starters.
The problem of buying and selling computation power on some sort of broker basis is a quite interesting problem in itself. Exactly what are you selling? Hardly CPU hours, since the value of those depends on the hardware.