[dpdk-ci] Minutes of DPDK Lab Meeting, August 22nd

[Trishan de Lanerolle]

Hi,
As we work to move the paperwork forward, have you had an opportunity to
connect with your legal teams to review the documents.

The usage agreement in particular is directly applicable to members
contributing hardware.

Please let me know your feedback/red line changes by end of the week if
possible.

Thanks,
Trishan



On Aug 22, 2017 5:07 PM, "O'Driscoll, Tim" <tim.odriscoll at intel.com> wrote:

Distribution list:
- We agreed to use the ci at dpdk.org mailing list for any future discussion
on the lab, so I've used it for these minutes. I've copied those who were
on the previous distribution for the minutes, in case anybody has not yet
subscribed to ci at dpdk.org. If you haven't subscribed yet, you can do so at:
http://dpdk.org/ml.

Lab Contract:
- The draft contract, with changes based on the LF review, is attached.
Others should review this and return comments to Bob (ren at iol.unh.edu) by
Friday
September 1st. We'll discuss any significant issues at our next meeting on
September 5th.
- The UNH-IOL Usage Agreement is also attached, and is also available at:
https://www.iol.unh.edu/sites/default/files/charters/unh-
iol-usage-agreement.pdf. People may want to review this as well.

Hardware:
- NXP and Intel hardware is in procurement. Mellanox hardware is ready but
needs internal approval before shipping to UNH.

CI:
- The script Mellanox have been working on to identify the target
repository for a patch set is being reviewed internally and should be sent
to the mailing list soon.
- Fangfang has sent a patch that includes a reference doc and some other
scripts. Thomas will review this.

Test Results:
- Discussed whether we always need to provide absolute numbers for test
results, or if delta values can be provided in some cases. Providing delta
values may be useful for new hardware or new tests that have not yet been
fully tuned, where vendors do not want to make absolute numbers public.
- Agreed that we should be able to decide per vendor and per test case
whether absolute or delta numbers should be provided. This should be
configured as part of the test script, so that the result is either the
absolute performance number or a delta comparing this to a previous
reference value.

Future meetings:
- Agreed to continue these meetings on a bi-weekly basis for now. If
anybody else on this mailing list wants to join, let me know and I'll
forward the meeting details.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://dpdk.org/ml/archives/ci/attachments/20170830/368d3749/attachment.html>