VEGAS Pulsar Project Meeting - Sprint Review: 2017 February 6th 2:00 - 3:00pm ET

Present

  • Ryan, Joe, Ray, Richard, Natalia, Laura

Sprint Story Overview

  • VE-154: Patch code in pulsar branch (COMPLETE)
  • VE-2: Merge VPM software branches with release branches (COMPLETE)
  • VE-164: Manager stability issues (INCOMPLETE)
  • VE-156: Test new c0800x0512 BOF file (COMPLETE)
  • VE-156: Review high time resolution modes (IN PROGRESS)
  • VE-171: Update SRBS (INCOMPLETE)
  • VE-162: Finish test environment assembly and move to equipment room (INCOMPLETE)
  • VE-163: Review/assess design for LBW software components (INCOMPLETE)
  • VE-170: Document plan/requirements for test environment (INCOMPLETE)
  • VE-166: Deploy instrumented VEGAS HPC code (COMPLETE)
  • VE-159: Review data visualization tools (INCOMPLETE)
  • VE-99: Release notes (IN PROGRESS)

Story Details

VE-154: Patch code in pulsar branch

Shared memory fixes in release branch are now in the pulsar branch.

VE-2: Merge VPM software branches with release branches

Took Ray about 3 days, but VPM is now rebased on BTL branch and can be released if it passes testing.

VE-164: Manager stability issues

Shared memory patches in release branch may help with stability, but this remains TBD.

VE-156: Test new c0800x0512 BOF file

Ryan tested a new c0800x0512 with optimal PFB bin widths. It performed as expected but did not get rid of data artifacts.

VE-156: Review high time resolution modes

Natalia and Ryan met to discuss possible problems with the size of the data blocks and/or the number of packets expected from the block headers. Natalia is meeting with Richard to review how to read in the GUPPI raw data.

VE-171: Update SRBS

No comments

VE-162: Finish test environment assembly and move to equipment room

Needs to moved and tested.

VE-163: Review/assess design for LBW software components

Joe has completed the review and has identified a more detailed list of tasks to work on. These include creating a spectral line PFB, generalizing the HBW accumulator, and creating a component that can read raw data on-disk for testing.

VE-170: Document plan/requirements for test environment

Moved to the next sprint.

VE-166: Deploy instrumented VEGAS HPC code

No comments.

VE-159: Review data visualization tools

Moved to this sprint at lower priority.

VE-99: Release notes

Ryan has started release notes and is waiting on more detail about M&C changes to include in the notes.

Outstanding Issues

Coherent Mode Data Artifacts

Artifacts were still seen in tests using different FPGAs/Banks. There was not enough time to try test tone injection. This is scheduled for Wednesday maintenance.

Ryan investigated historical data in more detail. In particular, I looked much more closely at the acceptance testing data from January to precisely measure the patterns with which the artifacts emerge. Results:

The following formulas for different modes predict channel numbers c0800x0128 8*n +/- 2

c0800x0256 32*n +/- 3

c0800x0512 32*n +/- 6

c0800x1024 16 + (n*32 +/- 3) (seem to be smaller amplitude artifacts, too)

c0800x0248 128*n +/- 26 (also seem to be smaller amplitude artifacts)

A pattern emerges when you map this into frequency space. Some modes seem to have artifacts at more frequencies than others but there are some common ones. Specifically, we always get artifacts at 1100 MHz + (n*50 MHz +/- 10 MHz) (note that 1100 MHz is the bottom of the band)

While the 950 MHz IF module did seem to produce more noise in the incoherent modes, there is no sign of this type of repeating pattern at certain frequencies.

VEGAS Pulsar Project Meeting - Sprint Planning: 2017 February 6th 2:00 - 3:00pm ET

Backlog

  • Document plan/requirements for test environment
  • Release notes
  • Data visualization
  • High time resolution modes

Critical Issues

  • Coherent mode artifacts
    • Ryan really wants to identify the source of these before release.
  • Manager stability
    • Coherent modes are not in a usable state due to manager stability issues. Specifically, during the last run of major tests, if a bank manager crashed a manual shared memory clean-up and ROACH reboot was required to bring it back to a useable state. The shared memory patches that were pulled into the VPM branch may help with this, but more testing is required. If issues persist, they need to be debugged and patched.
  • Test environment
    • Manager still is not functioning properly in the simulator. A functioning test environment would greatly accelerate other trouble shooting.

New Stories

  • VE-164: Manager stability issues
  • VE-177: Fix manager crashes in simulator
  • VE-175: Troubleshoot coherent mode data artifacts
  • VE-170: Document plan/requirements for VEGAS test environment
  • VE-173: Move test environment to equipment room
  • VE-171: Update SRBS
  • VE-174: Release notes
  • VE-156: Continue work on high time resolution modes
  • VE-176: Release M&C and config tool software
  • VE-179: Create spectral line FFT/PFB component for matrix

-- RyanLynch - 2017-02-06
Topic revision: r3 - 2017-02-09, RyanLynch
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback