On 5/29/20 4:37 AM, Mark Rivers via Tech-talk wrote:
>> For the IOC I set EPICS_CA_MAX_ARRAY_BYTES=800000000.
>
> That is nearly 1 GB. My understanding is that you should not use such a large value, because CA will actually allocate arrays of that size any time the actual array size is greater than 16 KB. It's better to set EPICS_CA_MAX_ARRAY_BYTES to be just a bit larger than the largest array you will actually use.
On the IOC side, in the Base 3.16 and 7.0 series EPICS_CA_MAX_ARRAY_BYTES is
ignored by default, and buffers are realloc()'d as needed beyond the 16KB minimum.
(the old behavior can still be restored by setting EPICS_CA_AUTO_ARRAY_BYTES=NO)
I believe that CAJ has similar logic, though it has been awhile since I looked.
In any event, it's probably not a good idea to set EPICS_CA_MAX_ARRAY_BYTES
so large. It's either unnecessary (>=3.16) or wastes a lot of memory (<=3.15).
> Mark
>
>
>
>
> ________________________________
> From: Tech-talk <tech-talk-bounces at aps.anl.gov> on behalf of Jörn Dreyer via Tech-talk <tech-talk at aps.anl.gov>
> Sent: Friday, May 29, 2020 2:00 AM
> To: tech-talk at aps.anl.gov
> Subject: Re: Strange problem with areaDetector driver
>
>
> My CS-Studio is based on the 4.5.1 release. I build it myself to have a site specific version.
>
>
>
> As I mentioned, I have successfully used it with aravisGigE Cameras, mainly from Basler.
>
>
>
> For the IOC I set EPICS_CA_MAX_ARRAY_BYTES=800000000.
>
> I have set a similar Value in CS-Studio But I can't check it at the moment. Yesterday I tried to build Phoebus and this somehow screwed up my CSS instalation.
>
>
>
>
>
> Jörn
>
>
>
> Am Donnerstag, 28. Mai 2020, 15:54:44 CEST schrieb Hu, Yong via Tech-talk:
>
>> That is odd. We are using CS-Studio heavily for areaDetector at NSLS-II and
>
>> have not seen such problems.
>
>
>
>> Which version of CS-Studio are you using? What is the settings of
>
>> EPICS_CA_MAX_ARRAY_BYTES both on your camera IOC side and on your CS-Studio
>
>> side?
>
>
>
>> We are using rules with PVs in IntensityGraph to automatically set Data
>
>> width / height instead of manually configuring those parameters. I can send
>
>> our .opi to you if you want (after your problems are fixed).
>
>
>
>> Best luck!
>
>>
>
>> Yong Hu
>
>> NSLS-II Controls Group
>
>>
>
>>
>
>> On 5/28/20, 4:24 AM, "Tech-talk on behalf of Jörn Dreyer via Tech-talk"
>
>> <tech-talk-bounces at aps.anl.gov on behalf of tech-talk at aps.anl.gov> wrote:
>
>
>
>> Hello
>
>>
>
>> After some debugging, it turns out that the reason for the problem is
>
>> CS-Studio.
>
>>
>
>> I have set up an opi Scene with an IntesityGraph for the Image. If I set
>
>> the
>
> Data width and Data high to zero, the scene chanes from "No data" to
>
>> "disconnected". If I set Data width to 4565 and Data height to 3520, the
>
>> scene can not be loaded and results in java.lang.IllegalArgumentException.
>
>> No further information is available. The scene can then even not get loaded
>
>> into the OPI Editor and needs to be modified in a text editor.
>
>> For other cameras I was able to set the data height and data width
>
>> parameters
>
> of the IntensityGraph to 2048 x 2048. Is there a size limit
>
>> for this Widget? I did not find any hints obn that in the documentation.
>
>>
>
>> Regards,
>
>>
>
>> Jörn Drteyer
>
>>
>
>> Am Donnerstag, 28. Mai 2020, 09:29:58 CEST schrieb Jörn Dreyer via
>
>> Tech-talk:
>
>> > Am Mittwoch, 27. Mai 2020, 16:10:26 CEST schrieb Mark Rivers via
>
>> > Tech-talk:
>
> >
>
>> > > > Closed the OPI screen that shows the image. Now the number of
>
>> > > > threads is
>
>> > > > constant. System load is at normal level.
>
>> > >
>
>> > >
>
>> > > When you saw the threads and high system load did you use top -H to
>
>> > > see
>
>> > > whether the threads with high CPU load were CSS threads or IOC
>
>> > > threads? I
>
>> > > suspect they are CSS threads.
>
>> >
>
>> >
>
>> > No they are all started from the IOC. After 300 threads spawned I
>
>> > decided to
>
> stop the IOC. I used
>
>> >
>
>> > ps -eLF | grep st.cmd | wc
>
>> >
>
>> >
>
>> > > > Will need some time to gat that going. Java on my system seems to
>
>> > > > have
>
>> > > > problem with the code.
>
>> > >
>
>> > >
>
>> > > ImageJ on both Linux and Windows comes with its own version of Java
>
>> > > in the
>
>> > > ImageJ/jre directory, so it should be completely independent of
>
>> > > your
>
>> > > system
>
>> > > version of Java. Get the latest version of the ImageJ zip file and
>
>> > > install
>
>> > > it in your own directory so you don't need to worry about system
>
>> > > permissions, etc.
>
>> >
>
>> >
>
>> > I got it working, but there seems to be a problem that only half of
>
>> > the
>
>> > image is intrerpreted correctly. The ArrayType is UInt16 instead of
>
>> > UInt8.
>
>> > Will need to find out teh reason for that.
>
>> >
>
>> > Regards,
>
>> >
>
>> > Jörn
>
>> >
>
>> >
>
>> > > Mark
>
>> > >
>
>> > >
>
>> > >
>
>> > > -----Original Message-----
>
>> > > From: Tech-talk <tech-talk-bounces at aps.anl.gov> On Behalf Of Jörn
>
>> > > Dreyer
>
>> > > via Tech-talk Sent: Wednesday, May 27, 2020 8:56 AM
>
>> > > To: tech-talk at aps.anl.gov
>
>> > > Subject: Re: Strange problem with areaDetector driver
>
>> > >
>
>> > > Am Mittwoch, 27. Mai 2020, 14:46:09 CEST schrieb Mark Rivers:
>
>> > >
>
>> > > > Hi Jörn,
>
>> > > >
>
>> > > >
>
>> > > > > I ran into a severe problem in the image readout handler.
>
>> > > > >
>
>> > > > > When calling doCallbackGenericPointer() the IOC creates every
>
>> > > > > time a
>
>> > > > > couple of new threads and consumes lot of CPU power ( 300% on
>
>> > > > > a
>
>> > > > > quad core).
>
>> > > >
>
>> > > >
>
>> > > > First a few questions:
>
>> > > >
>
>> > > >
>
>> > > > Is this Linux or Windows?
>
>> > >
>
>> > >
>
>> > > Linux
>
>> > >
>
>> > >
>
>> > > > What frame rate are you using?
>
>> > >
>
>> > >
>
>> > > 0.5 Hz
>
>> > >
>
>> > >
>
>> > > > What plugins are running?
>
>> > >
>
>> > >
>
>> > > Only NDStdArray.
>
>> > >
>
>> > >
>
>> > > > Is the CS-Studio GUI on the same machine as the IOC or a
>
>> > > > different
>
>> > > > machine?
>
>> > >
>
>> > >
>
>> > > On the same machine
>
>> > >
>
>> > >
>
>> > > > The first thing to establish is whether the problem is the
>
>> > > > areaDetector IOC or the CS-Studio client. I suggest you do the
>
>> > > > following:
>
>> > > >
>
>> > > >
>
>> > > > - Disable all plugins except NDPluginStdArrays for testing.
>
>> > >
>
>> > >
>
>> > > Done,
>
>> > >
>
>> > >
>
>> > > > - Exit CS-Studio so there is no display client running. What is
>
>> > > > the
>
>> > > > CPU load under those conditions?
>
>> > >
>
>> > >
>
>> > > Closed the OPI screen that shows the image. Now the number of
>
>> > > threads is
>
>> > > constant. System load is at normal level.
>
>> > >
>
>> > >
>
>> > > > - Try the ImageJ viewer instead of CS-Studio for testing. Does
>
>> > > > that
>
>> > > > also have the problem?
>
>> > >
>
>> > >
>
>> > > Will need some time to gat that going. Java on my system seems to
>
>> > > have
>
>> > > problem with the code.
>
>> > >
>
>> > > Regards,
>
>> > >
>
>> > > Jörn
>
>> > >
>
>> > >
>
>> > > > Mark
>
>> > > >
>
>> > > >
>
>> > > >
>
>> > > >
>
>> > > >
>
>> > > > ________________________________
>
>> > > > From: Tech-talk <tech-talk-bounces at aps.anl.gov> on behalf of Jörn
>
>> > > > Dreyer via Tech-talk <tech-talk at aps.anl.gov> Sent: Wednesday, May
>
>> > > > 27,
>
>> > > > 2020 7:03 AM
>
>> > > > To: tech-talk at aps.anl.gov
>
>> > > > Subject: Strange problem with areaDetector driver
>
>> > > >
>
>> > > > Hello,
>
>> > > >
>
>> > > > currently I have a very strange problem in a code I'm developing
>
>> > > > to
>
>> > > > read out cameras from ZWO. I ran into a severe problem in the
>
>> > > > image
>
>> > > > readout handler. When calling doCallbackGenericPointer() the IOC
>
>> > > > creates
>
>> > > > every time a couple of new threads and consumes lot of CPU power (
>
>> > > > 300%
>
>> > > > on a quad core). So after a very short time the system becomes
>
>> > > > unresponsive and has to be switched off. The image is 3520*4656
>
>> > > > bytes in
>
>> > > > size and is in monochrome mode, so UInt8 as data type.
>
>>
>
>
- References:
- Strange problem with areaDetector driver Jörn Dreyer via Tech-talk
- Re: Strange problem with areaDetector driver Jörn Dreyer via Tech-talk
- Re: Strange problem with areaDetector driver Hu, Yong via Tech-talk
- Re: Strange problem with areaDetector driver Jörn Dreyer via Tech-talk
- Re: Strange problem with areaDetector driver Mark Rivers via Tech-talk
- Navigate by Date:
- Prev:
Re: pco Camera USB and IEEE interface Bradnick, Ben (Tessella, RAL, TEC) via Tech-talk
- Next:
Re: pco Camera USB and IEEE interface Mark Rivers via Tech-talk
- 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
- Navigate by Thread:
- Prev:
Re: Strange problem with areaDetector driver Mark Rivers via Tech-talk
- Next:
display builder web runtime 吴煊 via Tech-talk
- 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
|