[Ofa_remotepm] First pass at a presentation slide deck

Paul Grun grun at cray.com
Thu Apr 12 12:29:34 PDT 2018


Still looking for a victi…ummm… “volunteer” to present this result tomorrow?
Don’t be shy – looks great on your resume.
-Paul

From: Bernard Metzler [mailto:BMT at zurich.ibm.com]
Sent: Thursday, April 12, 2018 10:20 AM
To: Voigt, Doug <doug.voigt at hpe.com>
Cc: Paul Grun <grun at cray.com>; ofa_remotepm at lists.openfabrics.org
Subject: Re: [Ofa_remotepm] First pass at a presentation slide deck


This is great work. Reads very interesting!

Please let me post some general comments, no changes to the current deck. And please forgive me if those comments are already completely captured by the deck, and account that to my limited understanding and limited patience to thoughtfully read through 20+ slides.

  1.  generally: we probably shall not take persistency of memory as a challenge in this work group. it is a nice add-on, which we may make explicit use of, or ignore. if we ignore it, we still benefit implicitly, since that memory is denser and needs less energy (since it is persistent). any memory can be made persistent. it is just a question of the electrical energy one wants to spend, which is determined by technology used to store the bits. so it is a cost factor then only.
  2.  use case: write once, read once scenarios are something we may look into as well. we may want to understand if persistency is a feature we can benefit from for this type of applications. think about serverless approaches (aws lambda, openwhisk, ...) which may provide great flexibility to classic big data computing (for what it is currently not being used), if intermediate results can be stored in remote PM, or local PM, which becomes remote PM for the next function.

Thanks,
Bernard.


[Inactive hide details for "Voigt, Doug" ---04/12/2018 10:32:18---Fixed minor spelling error on slide 5. Corrected slides 15-18]"Voigt, Doug" ---04/12/2018 10:32:18---Fixed minor spelling error on slide 5. Corrected slides 15-18 to align cross-references with new dec

From: "Voigt, Doug" <doug.voigt at hpe.com<mailto:doug.voigt at hpe.com>>
To: Paul Grun <grun at cray.com<mailto:grun at cray.com>>, "ofa_remotepm at lists.openfabrics.org<mailto:ofa_remotepm at lists.openfabrics.org>" <ofa_remotepm at lists.openfabrics.org<mailto:ofa_remotepm at lists.openfabrics.org>>
Date: 04/12/2018 10:32
Subject: Re: [Ofa_remotepm] First pass at a presentation slide deck
Sent by: "Ofa_remotepm" <ofa_remotepm-bounces at lists.openfabrics.org<mailto:ofa_remotepm-bounces at lists.openfabrics.org>>
________________________________



Fixed minor spelling error on slide 5.
Corrected slides 15-18 to align cross-references with new deck.

Doug

From: Ofa_remotepm [mailto:ofa_remotepm-bounces at lists.openfabrics.org] On Behalf Of Paul Grun
Sent: Thursday, April 12, 2018 2:41 AM
To: ofa_remotepm at lists.openfabrics.org<mailto:ofa_remotepm at lists.openfabrics.org>
Subject: [Ofa_remotepm] First pass at a presentation slide deck

I took the slide deck as it existed as of this evening and:
- Moved it onto the OFA template
- Began structuring the results in a presentation-style format
- We only have 20 minutes on Friday, so we obviously can’t go through everything that was captured in the notes from the meetings.
- I kept the original slides and moved them into backup one by one as I captured the salient points and combined them with others. Hopefully nothing got lost in the translation. Also, I may not have understood the point being made on one of the original notes slides, which is another reason for retaining the originals.
I took a crack at summarizing the key points and observations, and focused on the three or so use cases that got some significant discussion on Monday
- Checkpoint Restart
- Streaming Sensor Data
- Machine Learning
- The Dreamworks Case
I retained Doug’s excellent consolidation.

If this feels like the right way to go, let’s refine and focus these a little further and call it good. We don’t have a shared repository, so there’s no locking mechanism. Not sure how to deal with that except to hope for the best. I suggest if you change the slide deck, re-publish it to the reflector and include your initials in the file name, e.g. OFA_RPM_2018_0411_jp.pptx. That way you’ll at least know if your changes got incorporated in later versions.

-Paul

Ps – Is there a volunteer to present this stuff on Friday?

Advanced Technology Group
Cray, Inc.
Office – (503) 620 – 8757
Mobile – (503) 703 - 5382
[attachment "OFA_RPM_2018_0412.pptx" deleted by Bernard Metzler/Zurich/IBM]
_______________________________________________
Ofa_remotepm mailing list
Ofa_remotepm at lists.openfabrics.org<mailto:Ofa_remotepm at lists.openfabrics.org>
https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.openfabrics.org_mailman_listinfo_ofa-5Fremotepm&d=DwIGaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=2TaYXQ0T-r8ZO1PP1alNwU_QJcRRLfmYTAgd3QCvqSc&m=a0fcKXXWWwe9Mp3yYclJDVY-eqk8AfGMXmQKMSiZ3XU&s=8WWB4QVRaC-XMTNBoZ6YOWi5O5HZIWMijWJOKKOsges&e=


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openfabrics.org/pipermail/ofa_remotepm/attachments/20180412/62f0826c/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.gif
Type: image/gif
Size: 105 bytes
Desc: image001.gif
URL: <http://lists.openfabrics.org/pipermail/ofa_remotepm/attachments/20180412/62f0826c/attachment.gif>


More information about the Ofa_remotepm mailing list