Highlights from the CHEP Conference

mlp68 writes "This is from the floor of the 'Computing in High-Energy Physics' conference in Interlaken, Switzerland, which has been going on this week and closes today. Computer experts and scientists from large physics experiments around the world discussed their work and progress on various fronts. Data logging rates of 100's of MB/s and Linux farms with 100's or 1000's of machines are standard fare. Most of the talks presented, as well as today's summary talks, are available online. An interesting tidbit is that one experiment is now using P2P technology to configure their data acquisition components faster than was possible before."

Astrophysics implications

[Pi_0's don't shower]

In addition to competition for funding and notoriety in the scientific public, Astrophysicists and High-Energy Physicists often compete for computer resources, such as for the Earth Simulator. (Especially with the QCD High-Energy Physicists.)

If there are significant advances in that sort of computing, perhaps the resources currently taken up by QCD calculations could be used for things like numerical relativity?

Re:Astrophysics implications

[menscher]

I'd rather see the resources currently taken up by numerical relativity used for things like QCD calculations.

Disclaimer: I do lattice QCD.

Re:Astrophysics implications

[going_the_2Rpi_way]

As the article says, it seems to me that HPC and clusters in general are now standard fare for researchers in every field. The cost and availability of hardware is no longer really the issue (although there will always be the escalation) so much as the human resources component. I suspect the ability to program, QA/QC and interpret the massive amounts of inputs, code and results required to use these clusters to their potential is really becoming the limiting factor.

QCD getting their dedicated computers

[b2u]

A chip optimized to do QCD calculations (QCDOC QCD on Chip) [columbia.edu] has been developed and parallel computers using that chip are being built. More details [arxiv.org] can be found here. Machines are being built by the U.S., U.K. and Japan. The U.S. and Japaneese machines will be located at Brookhaven National Lab Computing Facilty [bnl.gov] which is close to Columbia University where the QCDOC was developed.

Just think..

[dj245]

..of a beowulf cluster of these in Soviet Russia where the data logs YOU!

oh forget it.

Wow!

[Ed_Moyse]

I just got back from CHEP (I gave a talk on my experiment's event data model) and find that we're on slashdot!

It was a hectic week and I'm still recovering ... interesting things I learnt:
- Interlaken has a Hooters bar (the only one in switzerland, next to the conference centre)
- no matter where you plan to go for a drink, you end up in Buddys bar.
- .... oh, and some stuff about computing as well. ;-)

What *was* amazing though was how the LHC experiments totally dominated the conference. I wasn't expecting this.