WU data doesn't match expectations. Explain the discrepancies, please.

Message boards : Number crunching : WU data doesn't match expectations. Explain the discrepancies, please.
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
Profile Ageless
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 14848
Credit: 4,052,312
RAC: 999
Netherlands
Message 1984921 - Posted: 13 Mar 2019, 12:17:52 UTC - in response to Message 1984914.  

It's called browser cache. Your browser caches the last state until the cache is refreshed. Or you can force a cache refresh with CTRL+F5 (Firefox) or whatever possible equivalent for other browsers.

People read things that aren't there and are much too easily offended these days, and always find they have the right to rip the helper a new one. Their loss as they're the one asking for help, which if they react enough like so, no one will offer again.
Jord

According to Giorgo of the Ancient Astronaut Theorists I do not help with tech questions via private message. He's right: please use the forums for that.
ID: 1984921 · Report as offensive
Profile Siran d'Vel'nahr
Volunteer tester
Avatar

Send message
Joined: 23 May 99
Posts: 6909
Credit: 19,529,586
RAC: 62,150
United States
Message 1984923 - Posted: 13 Mar 2019, 12:43:22 UTC - in response to Message 1984921.  
Last modified: 13 Mar 2019, 12:47:01 UTC

It's called browser cache. Your browser caches the last state until the cache is refreshed. Or you can force a cache refresh with CTRL+F5 (Firefox) or whatever possible equivalent for other browsers.

People read things that aren't there and are much too easily offended these days, and always find they have the right to rip the helper a new one. Their loss as they're the one asking for help, which if they react enough like so, no one will offer again.

Hi Jord,

I was just thinking while on my account page. If the computers are hidden for other users, are they hidden for us too? There are things that we can see on our account page that others cannot such as the IP address. I believe we should still see our own computers even if hidden from others.

Have a great day! :)

Siran

[edit]
I set my computers hidden and went to the main page of the website. I hit ctrl-F5 and the main page reloaded new content. I went back into my account page and I can still see my computers.
[/edit]
CAPT Siran d'Vel'nahr XO - L L & P _\\//
USS Vre'kasht NCC-33187

Windows 10 OS? "What a piece of junk!" - L. Skywalker

Microsoft wants us to "Imagine life without walls"...
I say, "If there are no walls, who needs Windows?"
ID: 1984923 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 17567
Credit: 395,703,553
RAC: 122,815
United Kingdom
Message 1984925 - Posted: 13 Mar 2019, 13:02:50 UTC

You can always see your own computers.
We can only see them if they are not hidden.
When asking for help it can be very useful to have your computers not hidden as then others trying to help have some limited information. They cannot access your computers. If you want to see what others can see then feel free to have a look at mine.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1984925 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 7845
Credit: 474,506,897
RAC: 706,054
Panama
Message 1984927 - Posted: 13 Mar 2019, 13:15:22 UTC
Last modified: 13 Mar 2019, 13:23:16 UTC

Ask for help with the host hide is like call 911 and don't allow them see the patient. Is impossible to give any help in most of the cases.

I still believe the term hide/unhide is what causes all this misunderstandings.

People have fear of open a windows to the outside world, and believes when you unhide a host you do exactly that.
With most of the programs that could be true but that is not true with Boinc, when you unhide the hosts you only allow the others to see besides how many hosts you run, how they are configurated and mainly the stderr file of the crunched WU. All info stored from the server side only not your client side. And this stderr file is who give the clues to how to fix most of the problems. No information about IP or where the host is located is shared.

Maybe if they change the terms hide/unhide to something more adequate who better explain what that means could avoid that.
ID: 1984927 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 11468
Credit: 167,099,003
RAC: 99,692
Australia
Message 1985033 - Posted: 14 Mar 2019, 3:52:36 UTC - in response to Message 1984927.  

Maybe if they change the terms hide/unhide to something more adequate who better explain what that means could avoid that.

Still won't help with paranoia.
Never let the facts interfere with beliefs; these days even more so than ever before.
Grant
Darwin NT
ID: 1985033 · Report as offensive
marmot
Avatar

Send message
Joined: 15 May 99
Posts: 144
Credit: 1,190,504
RAC: 322
United States
Message 1985243 - Posted: 15 Mar 2019, 7:39:37 UTC - in response to Message 1984914.  


My apologies for trying to help and insulting you.
That's obviously not a real apology.

It was not my intent to insult.
Think before posting.
I made it clear that 2 attempts were made to unhide the machines and suggesting it's because a 19 year user is an idiot, is not acceptable help desk behavior.
You threw BOINC cmdline options at me when I was asking about SETI Lunatics cmdline options.
Don't offer help when you are ignorant on the subject.

I did not see that you joined 8 days before me and are more proficient than I.
Never stated I was more proficient at all things than you. Just that I'm not a complete newb and deserve as much respect as you.

I will not attempt to help again.
That's your choice, and to be respected.

Have a great day! :)

After your earlier comments, I'll translate that as "die a horrible death, marmot".

Hope your future help desk scenarios go better. I really mean that.
ID: 1985243 · Report as offensive
marmot
Avatar

Send message
Joined: 15 May 99
Posts: 144
Credit: 1,190,504
RAC: 322
United States
Message 1985244 - Posted: 15 Mar 2019, 7:44:27 UTC - in response to Message 1984921.  
Last modified: 15 Mar 2019, 8:21:35 UTC


People read things that aren't there and are much too easily offended these days, and always find they have the right to rip the helper a new one. Their loss as they're the one asking for help, which if they react enough like so, no one will offer again.


The help given by Siren, was not helpful in either circumstance, it was off topic and unnecessary.
I was polite yet firm in my response.
I am not responsible for the bruised ego and they could have gone and said nothing more about it.

More than half of this thread is dealing with Siren's issues than actually working on the technical issue that started the thread.

Asking for help doesn't mean we are to be derided, demeaned or insulted in the process. I worked helpdesk for many years and know how frustrating it can be to both helper and client.
ID: 1985244 · Report as offensive
marmot
Avatar

Send message
Joined: 15 May 99
Posts: 144
Credit: 1,190,504
RAC: 322
United States
Message 1985245 - Posted: 15 Mar 2019, 7:54:01 UTC - in response to Message 1984927.  

Ask for help with the host hide is like call 911 and don't allow them see the patient. Is impossible to give any help in most of the cases.

I still believe the term hide/unhide is what causes all this misunderstandings.

People have fear of open a windows to the outside world, and believes when you unhide a host you do exactly that.
With most of the programs that could be true but that is not true with Boinc, when you unhide the hosts you only allow the others to see besides how many hosts you run, how they are configurated and mainly the stderr file of the crunched WU. All info stored from the server side only not your client side. And this stderr file is who give the clues to how to fix most of the problems. No information about IP or where the host is located is shared.

Maybe if they change the terms hide/unhide to something more adequate who better explain what that means could avoid that.


My computers are hidden because my configurations are for me and not to be shared.
There's no concerns that the machines could be hacked by unhiding them.
I did unhide the machines before starting the thread and will rehide them eventually.
Not sure why privacy is regarded as a deficit.
ID: 1985245 · Report as offensive
marmot
Avatar

Send message
Joined: 15 May 99
Posts: 144
Credit: 1,190,504
RAC: 322
United States
Message 1985249 - Posted: 15 Mar 2019, 8:43:14 UTC - in response to Message 1984732.  

Yes, the Lunatics SETI app is multi-threaded. This is the Process Explorer view of the NVidia build, but all versions are built from the same code base.



Most of the threads are called from the driver/language DLL, but there are three threads called directly from the Lunatics executable.

Two of these are background threads managed by the BOINC library API, for tasks like timing and message-passing which are required for an application to run under the BOINC environment. As you see here, they normally occupy a tiny fraction of a single CPU core. The 23.75% figure (one full core) is specific to the NVidia OpenCL programming environment, and represents the spin-wait synchronisation loop required by NVidia's implementation of OpenCL - it won't apply on your AMD cards.


Well, yes, technically, that is multi-threading!

I was a actually referring to parallelism in a computing problem solved with multi-threading the data set. Not just on the GPU level but also on the CPU level. SRBase llr.exe uses a -tx command line option to split the data set among 2 to 8 (it can try more but doesn't seem effective from my data) threads within the visible process. YAFU multithreads the data calculations by starting 2 to 32 separate OS processes.

I've seen the Lunatics m8 wrapper have 3 copies of amdocl.dll (AMD Accelerated Parallel Processing OpenCL library) copies drawing down CPU slices in a decreasing percentages, in order to break up the data set at the CPU level. However, they never use more than about 1/2 of a CPU core and could not account for my observed data issues where run time is shorter than CPU time.


So I would say that where you see CPU time exceeding GPU time, there is indeed something unexpected happening on your computer, that is not designed into the Lunatics apps.


Yes, you are right.
After some searching I found that other people had issues with the Crimson 17.11.4 Catalyst drivers crashing on similar motherboards.
DDUed the drivers and installed 19.3 Catalyst and this issue seems to be gone although when I try Mike's suggested cmdline setting of

-sbs 786 -spike_fft_thresh 2048 -tune 1 64 1 4

is still not working reliably with at least 50% of the time the prime amdocl.dll thread stalls at 0.06% CPU usage and no GPU usage at all.
Have to abort those WU's.

It's running default setting fine for the last 4 hours.
This issue seems to be closed with the discovery of the faulty driver.
Optimizing it is a separate issue.

Really appreciate the thought you put into your helpful responses to me, mister Haselgrove.

Thankyou Vyper, Tom M also, and thankyou Siren, you tried.
ID: 1985249 · Report as offensive
Profile Ageless
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 14848
Credit: 4,052,312
RAC: 999
Netherlands
Message 1985250 - Posted: 15 Mar 2019, 8:44:01 UTC - in response to Message 1984725.  

A GPU is a co-processor. The idea is for the CPU to hand off the computation load to its partner, and then be free to do something else. BOINC doesn't measure or report the run-time for the GPU separately, but it is likely to be close to the run time for the task as a whole.


That is completely off topic and information we all know already and has nothing to do with the errors seen. The run time is simply the ending time minus the start time. The time the WU started until it finishes, it's the upper limit of GPU seconds and by definition must be greater than the CPU time unless the WU is multi-threaded. It also must be 4 times longer (or more) than the CPU time since the Lunatics WU is locking the CPU into 25% of a CPU core.
You didn't read Richard's answer, complaining instead about it being off topic when it wasn't. There have also been no errors.

Let me try to phrase it in simple words:
BOINC does not measure how long a task takes on a GPU. It only measures CPU time, no matter if a task is run on a CPU, GPU, different co-processor or any other calculation resource.

Until there is a volunteer developer who spends time on changing that in the client, and on the server, this isn't going to change any time soon.
Jord

According to Giorgo of the Ancient Astronaut Theorists I do not help with tech questions via private message. He's right: please use the forums for that.
ID: 1985250 · Report as offensive
marmot
Avatar

Send message
Joined: 15 May 99
Posts: 144
Credit: 1,190,504
RAC: 322
United States
Message 1985251 - Posted: 15 Mar 2019, 9:00:40 UTC - in response to Message 1985250.  
Last modified: 15 Mar 2019, 9:01:36 UTC

A GPU is a co-processor. The idea is for the CPU to hand off the computation load to its partner, and then be free to do something else. BOINC doesn't measure or report the run-time for the GPU separately, but it is likely to be close to the run time for the task as a whole.


That is completely off topic and information we all know already and has nothing to do with the errors seen. The run time is simply the ending time minus the start time. The time the WU started until it finishes, it's the upper limit of GPU seconds and by definition must be greater than the CPU time unless the WU is multi-threaded. It also must be 4 times longer (or more) than the CPU time since the Lunatics WU is locking the CPU into 25% of a CPU core.
You didn't read Richard's answer, complaining instead about it being off topic when it wasn't.
There have also been no errors.


I did read the answer, responded and yes there were. many reporting errors with run times far shorter than CPU time.
The Seti@Home has already expunged them from the database.

I have them in a spreadsheet.


Let me try to phrase it in simple words:
BOINC does not measure how long a task takes on a GPU.
It only measures CPU time, no matter if a task is run on a CPU, GPU, different co-processor or any other calculation resource.


Because there is no measure taken of the amount of time a set of data is in each of the Computation Units of a GPU, when they start working on a data slice and when they complete. However, run time is still the boundaries of the amount of time the data set spent in the GPU. You don't need to simplify, I was working on a doctorate in computer science.

Richard was very helpful in getting this issue resolved.
ID: 1985251 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 13020
Credit: 141,999,115
RAC: 164,396
United Kingdom
Message 1985252 - Posted: 15 Mar 2019, 9:03:39 UTC - in response to Message 1985249.  

Really appreciate the thought you put into your helpful responses to me, mister Haselgrove.
Marmot and I didn't get off to the best of starts, because neither of us fully understood the depth of prior experience of the other. But I'm glad it worked out well in the end, once we got into the meat of the problem.
ID: 1985252 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 13020
Credit: 141,999,115
RAC: 164,396
United Kingdom
Message 1985254 - Posted: 15 Mar 2019, 9:16:55 UTC - in response to Message 1985249.  

This is the Process Explorer view of the NVidia build, but all versions are built from the same code base.
For future reference, I used Process Explorer because I run Windows 7, and the Windows 7 Task Manager doesn't show sufficient detail.

Those running Windows 10 have access to a much more sophisticated Task Manager, but its output does need to be interpreted with care. The W10 TM does display GPU loading, but the default view concentrates on metrics which are not applicable to distributed computing. You have to drill down into other fields to find where the compute load is hiding.
ID: 1985254 · Report as offensive
marmot
Avatar

Send message
Joined: 15 May 99
Posts: 144
Credit: 1,190,504
RAC: 322
United States
Message 1985260 - Posted: 15 Mar 2019, 9:35:40 UTC - in response to Message 1985254.  

This is the Process Explorer view of the NVidia build, but all versions are built from the same code base.
For future reference, I used Process Explorer because I run Windows 7, and the Windows 7 Task Manager doesn't show sufficient detail.


On the Portable Apps platform there is a nice task manager called Process Hacker. I'm liking it better than Process Explorer.
Among other nice information column headings, advanced process probing features, it can maintain a process priority database and set a process to the priority you assigned when it detects the process in RAM.
ID: 1985260 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 13020
Credit: 141,999,115
RAC: 164,396
United Kingdom
Message 1985261 - Posted: 15 Mar 2019, 9:38:31 UTC - in response to Message 1985260.  

I use Process Lasso for the same purpose. But there's a slight difference between viewing and controlling process behaviour- which could be combined in a single tool, of course.
ID: 1985261 · Report as offensive
Profile Siran d'Vel'nahr
Volunteer tester
Avatar

Send message
Joined: 23 May 99
Posts: 6909
Credit: 19,529,586
RAC: 62,150
United States
Message 1985270 - Posted: 15 Mar 2019, 10:20:49 UTC
Last modified: 15 Mar 2019, 10:21:45 UTC

Greetings,

@Marmot:

First off, you make way too many assumptions, with respect to what I type in a post.

Second, take another look and my username. Please spell it correctly whenever you feel the urge to talk about me.

Have a great day! :) <- Do NOT assume that that means anything other than what it says.

Siran
CAPT Siran d'Vel'nahr XO - L L & P _\\//
USS Vre'kasht NCC-33187

Windows 10 OS? "What a piece of junk!" - L. Skywalker

Microsoft wants us to "Imagine life without walls"...
I say, "If there are no walls, who needs Windows?"
ID: 1985270 · Report as offensive
marmot
Avatar

Send message
Joined: 15 May 99
Posts: 144
Credit: 1,190,504
RAC: 322
United States
Message 1986077 - Posted: 20 Mar 2019, 6:51:54 UTC - in response to Message 1985270.  


Second, take another look and my username. Please spell it correctly whenever you feel the urge to talk about me.


Misspelling your name of Siran could be seen as insulting.
It was unintentional and I do apologize for the misspelling.
I have a fondness for the word "siren" and the mythical creatures; not the alarms.
ID: 1986077 · Report as offensive
Previous · 1 · 2

Message boards : Number crunching : WU data doesn't match expectations. Explain the discrepancies, please.


 
©2019 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.