4 ms·
Being fascinated by the physics of it, it would be really interesting to have a more detailed description of the H/W and S/W involved. I guess they get a few GB
by NKosmatos 9y ago
Being fascinated by the physics of it, it would be really interesting to have a more detailed description of the H/W and S/W involved. I guess they get a few GB per collision which for sure needs some hefty number crunching CPUs, data storage and smart statistical (big data) analysis.
- stonogo 9y agoIt's all in the open. I have an acquaintance on the project, although we haven't talked about it in some time. https://indico.hep.pnnl.gov/event/14/session/1/material/0/0.pdf https://indico.hep.pnnl.gov/event/14/session/1/material/0/0....
- tetha 9y agoThe storage, network and compute needs of these projects are really exciting. You get a couple gigs to a couple dozen gigs per measurement run, and you're running a lot of measurements. As the presentation shows, they are looking at a petabyte per month. That's a lot of stuff to crunch. CERN is also pushing 2 - 3 digit petabyte storages on ceph around for their LHC. Genome analysis has similar interesting requirements. A sequencer for a human genome outputs something like 100 - 200Gb of raw data to store and process, and then you have some computationally tricky analysis to run on that.
- dukwon 9y agoRaw event sizes in modern collider experiments are typically 100s of kB to a few MB. The trigger rate at Belle II is about 30 kHz. It's comparable to an LHC experiment in terms of data output, but it has the advantage of lower particle multiplicity, making reconstruction and simulation a lot easier.