1 / 5

To MEP or not to MEP

To MEP or not to MEP. Niko Neufeld RTTC meeting, October 26th 2004. The LHCb transport format: MEPs. The online transport format is MEPs: https://edms.cern.ch/document/499933.1

mateja
Download Presentation

To MEP or not to MEP

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. To MEP or not to MEP Niko Neufeld RTTC meeting, October 26th 2004

  2. The LHCb transport format: MEPs • The online transport format is MEPs:https://edms.cern.ch/document/499933.1 • MEP means that event fragments from several consecutive triggers are merged into one MEP and send as a single IPv4 packet • MEPs are used in HLT and L1 to reduce the overhead incurred by sending small packets over the network • The number of fragments in a MEP is called the packing factor (PF) - it is a constant (during a run), adjustable global parameter of the system • The nominal PFs for L1 and HLT are 25 and 10 respectively Niko Neufeld CERN, PH

  3. Sending Events in the RTTC • In the RTTC we will have a single source of data (a disk-server) sending to at least one sub-farm (most likely three) • This means we could send completely build events to the SFC and the SFC would just “forward” them • I think we should not opt for this strategy Niko Neufeld CERN, PH

  4. Fully realistic event-builder in action with all timing relevant overheads, packet handling issues No special version of event-builder code Possibility to much more realistic robustness tests (missing/corrupted MEP from single source will be the most likely error in the final system) Need to reformat raw data into MEPs: not difficult, study in progress, will present results in next meeting Need real event-builder: no problem: will be tested and ready by end of this year (c.f. Benjamin’s talk) sub-farms will not run at 120 MB/s --> even a less performing SFC can do the job The case for MEPs Niko Neufeld CERN, PH

  5. “To MEP or not to MEP, that is the question here” • I think, “’Tis MEPness, but there is a method in it” - or leaving Shakespeare aside: • For the price of a few technical details (re-formatting of raw-data) we get a fully realistic system - we should go for MEPs Niko Neufeld CERN, PH

More Related