Difference: AMC13DebugLog (82 vs. 83)

Revision 8319 Sep 2013 - DavidZou

Line: 1 to 1
 
META TOPICPARENT name="CmsSlhc"
Please log AMC13 test activity below, blog style (new entries at top)
Added:
>
>

2013-09-13, David and Eric

We are experiencing IPBus2 errors during block reading, which has also been experienced by collaborators at both Bristol and CERN.
  • There seems to be a random chance for a block read of the events to fail (for example during a dump for events to file).
  • Failures will occur after a variable amount of block reads
  • Failure do not seem to related to corrupt data. A successful work around seems to be that if a block read fails, rechecking the same event immediately afterwards will succeed.
  • Created a test functions in AMC13Tool to either block read through all event (i.e. block reads followed by page turns) or block reading over the same event over and over.
    • Failures occur in both causes which can occur after variable number of block reads, typically after a few hundred. But can occur as early as after a few tens of reads or 1000+ block reads.
    • If the size of the block read is reduced (e.g. reading only 1/10th of the total number of words per event, for the fake events create by AMC13 that is about 620 words), the rate at which the error occurs seems to increase.
      • Consistently failing after only a few block reads (typically less than 10)
      • With this reduced block read size, errors can be experienced by manually doing successive block reads on the events saved in buffer (e.g. brv 0x4000 620)
  • Introducing a sleep between successive seems to slightly decrease the rate of failures, but only marginally.
    • A one second sleep may only double or triple the average number of block reads before failure for a give block read size.
    • But the significant delay introduced severely slows down the process of reading or dump a multitude of events, and therefore is not a viable work around.
 

2013-09-13, Nic E, David and Eric

Nic E is here with his VT892 crate, PM, MCH and AMC13XG# 34 to diagnose dead AMC13 problems.

 
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback