1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 <2018> 2019 2020 2021 2022 2023 2024 | Index | 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 <2018> 2019 2020 2021 2022 2023 2024 |
<== Date ==> | <== Thread ==> |
---|
Subject: | Re: Idea for new Display Manager |
From: | Pete Jemian <[email protected]> |
To: | Mark Rivers <[email protected]>, "'[email protected]'" <[email protected]> |
Cc: | "[email protected]" <[email protected]> |
Date: | Thu, 1 Mar 2018 12:36:21 -0600 |
To illustrate why I believe adding an additional intermediate component will not help resolve the OP, I offer this hypothetical: I have considerable information and GUI screens in Odesta's Double Helix. [1,2] While still alive today, it suffers EOL issues as do many EPICS DM tools. Are there converters available to get my screens to YADM? No. I'll have to write them. How about migrating the database to a newer SQL database? Same issue.
The point is, I have not solved the DM issue by adding in a SQL backend but have created a new DB issue for future maintenance. And the new SQL backend already appears to have performance issues for a use case (as Mark Rivers points out) that is growing quite popular.
Pete [1] https://www.revolvy.com/topic/Helix%20(database) [2] https://en.wikipedia.org/wiki/Helix_(database) On 3/1/2018 12:20 PM, Mark Rivers wrote:
That's faster than most network cards already (@ 1 Gbps) so it seems to me one would want specialized equipment / setup.So what? I am running the display manager on the same computer as the USB-3 camera. I just use a COTS PC with a USB-3 port, no specialized equipment, and no need to go over the network at all. With SQL data needs to flow: - Camera -> EPICS IOC - IOC->SQL (what software?) - SQL->display manager This is clearly not as efficient as today: Camera->EPICS IOC EPICS IOC->display manager A quick Google search shows lots of performance problems with using SQL for images. Do you know what bandwidth you can actually get for writing and then reading images from SQL? If I want to use SQL and avoid network overhead I need to run SQL on every machine that is running the camera, including Windows (many cameras/detectors are Windows only). This would significantly increase the complexity of deployment. Mark I think Samsung has HDsthat go at 500+ MB/s solid state, but you'd probably want everything to go to a RAM drive, it's as fast as RAM. It see that as an equipment issue, not a method issue.-----Original Message----- From: [email protected] [mailto:[email protected]] Sent: Thursday, March 01, 2018 12:12 PM To: Mark Rivers <[email protected]> Cc: '[email protected]' <[email protected]>; Pete Jemian <[email protected]>; tech- [email protected] Subject: RE: Idea for new Display Manager That's faster than most network cards already (@ 1 Gbps) so it seems to me one would want specialized equipment / setup. I think Samsung has HDs that go at 500+ MB/s solid state, but you'd probably want everything to go to a RAM drive, it's as fast as RAM. It see that as an equipment issue, not a method issue.
We want to display 2 mega-pixel RGB images at 30 Hz = 180 MB/s. Is SQL going to handle that efficiently? Mark-----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of [email protected] Sent: Thursday, March 01, 2018 11:46 AM To: Pete Jemian <[email protected]> Cc: [email protected] Subject: Re: Idea for new Display Manager Well anyone interested in building a display manager right now has to know about Epics and use Epics libraries - which is somewhat specialized, and also how to program - also specialized. If there was a SQL database intermediary, then they only have to know how to interact with a SQL database (part of a typical software curriculum). It segregates the data/control from how said data is displayed or interacted with. If the system used in a display manager changes, becomes outdated (like Motif), no matter - the SQL database standards are the same. As you say, "One can already use any language, any system, to control and display data..." which is true, today - though I have trouble compiling edm for Ubuntu 64. With this method though, a rather intensive effort admittedly, a display manager would be compatible with languages of tomorrow too, ones not even invented yet. As long as they are SQL database compatible, which they will be, or they won't take off as a language. Cheers James Richard Larsson
Now I'm confused. On 3/1/2018 10:46 AM, [email protected] wrote:All the displays managers for Epics *look* far behind what isavailablefrom commercial systems, say Schneider or Rockwell. Which are themselves far behind what programmers are doing today with, a modern operating system or video games (also control interfaces). And that's whatwe'retalking about right, at least in part, how it looks, or how wedisplaydata? Otherwise we could just use camonitor on a terminal withcaput.I think the key to having a decent display manager, one that adaptstonew technologies, better/different formats (say HTML5/PHP, Qt, text a la Lynx, or some fancy Virtual Reality / VTK / QT3d initiative) is to haveEpicseasily push/pull from a database (say MySQL, PostgreSQL). Once thatisdone one could use any language, any system, to control and displaydataas long as said language or method can access a typical database. As a side benefit to using a properly setup database one would automatically get an archive, and the benefits of database synchronization/backup, and seamless data compression. (Also if someone is excited and jumping in to do this my vote is for submitting it the Package Management Debian/Ubuntu repository. I'm a control systems engineer, and want to spend my time controllingsystems,not compiling very particular versions of modules, libraries,plugins,and endless environment variables just to begin my work.) My two cents & cheers! James Richard LarssonWhat is the value in adding an additional intermediate component, such as a SQL database server, to solve the inadequacies of the various existing display managers? That's a new piece of work for this discussion. Each IOC already provides the data servers with a common protocol. One can already use any language, any system, to controlanddisplay data as long as said language or method can access an EPICSIOC.Pete -- ---------------------------------------------------------- Pete R. Jemian, Ph.D. <[email protected]> Beam line Controls and Data Acquisition (BC, aka BCDA) Advanced Photon Source, Argonne National Laboratory Argonne, IL 60439 630 - 252 - 3189 ----------------------------------------------------------- Education is the one thing for which people are willing to pay yet not receive. -----------------------------------------------------------
-- ---------------------------------------------------------- Pete R. Jemian, Ph.D. <[email protected]> Beam line Controls and Data Acquisition (BC, aka BCDA) Advanced Photon Source, Argonne National Laboratory Argonne, IL 60439 630 - 252 - 3189 ----------------------------------------------------------- Education is the one thing for which people are willing to pay yet not receive. -----------------------------------------------------------