Posts by HAL9000


log in
1) Message boards : Number crunching : Windows 7 configuration issue that affects GPU crunching - help needed (Message 1720357)
Posted 2 hours ago by Profile HAL9000
Well, some small progress:

I implemented seems all (but comprehensive AV scan) suggestions but nothing helped.

Next I changed UAC settings to "Never notify".
And issue disappeared (!). Now BOINC detects GPU and runs w/o crash on startup. FAR doesn't complain about config files, Chrome allows to download files and so on.

I think it's definitely demonstrates that smth wrong with user security part of OS, not with NTFS file system security.

Cause I would prefer to be able to use UAC now the question is: how to repair UAC ?

Ah yes. I always disable UAC. It is mostly worthless & annoying. Also there are settings for UAC that I disable in group policy. Because they are still in effect even though UAC is disabled.
2) Message boards : Number crunching : Windows 7 configuration issue that affects GPU crunching - help needed (Message 1719706)
Posted 2 days ago by Profile HAL9000
Did you try "Take control" ?
You can change permission about drives and folders.

Is that the same as setting the ownership, which they have tried, or are you talking about a different function? I'm not familiar with "Take control". Unless maybe that is in one of the wizards?
3) Message boards : Number crunching : Panic Mode On (100) Server Problems? (Message 1719698)
Posted 2 days ago by Profile HAL9000
Anyone else notice slowed download speeds since the switch to the campus system? Maybe it's just the WOW event activity but it looks to be around 50% slower recently.

I haven't noticed anything, but most of my machines list their Average download rate from 1300 KB/sec to 1800 KB/sec.
I would have to lookup how BOINC calculates that averages to see if I could reset it. Then see if the average is much less.
4) Message boards : Number crunching : Windows 7 configuration issue that affects GPU crunching - help needed (Message 1719609)
Posted 2 days ago by Profile HAL9000
1) Yes, it looks like in that PNG picture.
2) If it would be only manifestation I would fine with it. Unfortunately, already installed apps really can't write in different drive C: folders... That's the real issue.
3)Will check with command line when be at that PC, thanks.

Even if I set attrib +r temp for my C:\temp folder I can still create new folders & add files. So issue seems to be more related to security rather than attribute. If you create a new folder on C: does it have the same issue?
5) Message boards : Number crunching : Windows 7 configuration issue that affects GPU crunching - help needed (Message 1719606)
Posted 2 days ago by Profile HAL9000
You can try "NTFS Access" by Zeus Software (Freeware, 26 KB)
http://www.zeus-software.com/downloads/ntfsaccess

Do you by chance have installed any "security package" (Antivirus) that may do this lock-of-permissions?


Thanks, but didn't help. I used this app on root of drive C:. It said all done, but then again, for example, BOINCdata dir was with read-only box marked.
It added sequirity record with my user name and "Full access" option, but on properties tab still "read only".... It's very strange but so.
And as I said before, I can clear this box (so user have rights to do so), also it asks to clear read-only on inner folders too, and it starts to clear... but when I open properties tab again - same checked box as before.
So, need to understand what area in Windows configuration hierarchy could override this.

And no, no any 3rd party AV soft there.

When you mentioned you were seeing read-only permission before I thought you meant in the security section.
If you are seeing the general read-only box as partially checked. Like this: http://i.imgur.com/nkI25ao.png. That is normal. You can see a folders true attributes from the command line. http://i.imgur.com/7gZsHzZ.png
6) Message boards : Number crunching : Futher restrictions on when BOINC runs tasks... (Message 1719587)
Posted 2 days ago by Profile HAL9000
For some reason they don't distribute boinccmd with the GUI version of BOINC for the Mac. So you can download the (Unix command-line version) for your version of BOINC, extract boinccmd, & then place it in the /Library/Application Support/BOINC Data/ folder.
Depending on your system you may also need to fix the file permissions. Permissions can be fixed by running the BOINC installer again or with the use of Mac_SA_Secure.sh located in the BOINC command-line download package.
7) Message boards : Number crunching : Windows 7 configuration issue that affects GPU crunching - help needed (Message 1719180)
Posted 3 days ago by Profile HAL9000
There is a commandline tool for setting file & folder permissions you could try called icacls.
I use it on one of our servers to set permissions for newly created folders with specific names that needs special rights that are not inheritable.

I did once try to use it to set Everyone to have Full Access, like in XP, & Windows 7 was not very happy with my attempts to do so.


Yes, I used that tool for exporting sequirity settings from another host and importing into wounded one - but it did not help.

I think at this point I would save this OS to a disk image & try a new install. If the new install will not activate then restore OS & continue to bang head on desk.
8) Message boards : Number crunching : Futher restrictions on when BOINC runs tasks... (Message 1719176)
Posted 3 days ago by Profile HAL9000
The BOINC command too will allow you to control BOINC from a commandline, or via scripts. If there is no OS event that would signify when the battery is fully charged. Perhaps a script to monitor the battery level could be used. Then it could run the correct command to suspend or resume BOINC.
9) Message boards : Number crunching : Windows 7 configuration issue that affects GPU crunching - help needed (Message 1719131)
Posted 3 days ago by Profile HAL9000
There is a commandline tool for setting file & folder permissions you could try called icacls.
I use it on one of our servers to set permissions for newly created folders with specific names that needs special rights that are not inheritable.

I did once try to use it to set Everyone to have Full Access, like in XP, & Windows 7 was not very happy with my attempts to do so.
10) Message boards : Number crunching : Windows 7 configuration issue that affects GPU crunching - help needed (Message 1719053)
Posted 3 days ago by Profile HAL9000
IIRC its only possible to convert Fat32 to NTFS but not vice versa.

There are 3rd part tools to convert NTFS partitions to FAT. Normally that is done for data rather than boot partitions. Like the one they listed.

Windows 7 may not like booting to a FAT32 partition that is larger than 32GB. If possible shrinking the partition could be a work around.
11) Message boards : Number crunching : Windows 10 - Yea or Nay? (Message 1719048)
Posted 3 days ago by Profile HAL9000
That`s why i never do automatic updates.

Data collection are coming on Win 7 and 8 too.

http://winaero.com/blog/telemetry-and-data-collection-are-coming-to-windows-7-and-windows-8-too/

And to that I say. No thank you.
http://i.imgur.com/TH5OjMS.png
12) Message boards : Number crunching : Panic Mode On (100) Server Problems? (Message 1719040)
Posted 3 days ago by Profile HAL9000
For those who don't bother to read Matt's post in Tech News on 21 August:

Message 1716662 - Posted: 21 Aug 2015, 22:47:12 UTC

Those panicking about a coming storm due to lack of data... The well is pretty dry but Jeff and I just uncovered a stash of tapes from 2011 that require some re-analysis, so that's why you'll see a bunch showing up in splitter queue over the weekend (hopefully before the the results-to-send queue drops to zero).

In the meantime, we are still recording data at AO (not fast enough to keep our crunchers supplied), but.... this situation has really pushed us to devote more resources to finally finishing the GBT splitter, which will avail to us another reserve supply of data in case we hit another dry spell.

The network switch on Tuesday seems to have gone fairly well. We are now sending all our bits over the campus net just like the very old days .

- Matt


So, the tapes being split just now have already been split once, but are being culled for more usable data. More effort is being put into the splitter for the GBT data (coo, what a surprise - its "tapes" are not the same format as those from AO...) Having two data sources should really improve the situation somewhat.

Next challenge. We must burn through data faster than two sources can provide!
13) Message boards : Number crunching : Panic Mode On (100) Server Problems? (Message 1717791)
Posted 6 days ago by Profile HAL9000
and if the WU results exist already, they're not done all over again).

Don't they keep track? Or are they just trowing in any tape that happens to be sitting around?

It needed to be split again for MB. While it's there, the AP splitters take a peek, just to be on the safe side. There was one over the weekend which generated AP tasks from about the first six channels, but not the rest. I'm assuming there was an error the first time round which meant some got missed - so the re-check was worth while.


And I assume that AP peek is going on now to 17ja11ac...

It's either peek, or poke, nobody really knows :-)

Hopefully it isn't puke.
14) Message boards : Technical News : More Data (Aug 21 2015) (Message 1717790)
Posted 6 days ago by Profile HAL9000
Good news on the switchover. Maybe we will get a fancy new bandwidth graph sometime this quarter as well.

When we run low on data to split & send out I just think of this little guy.
15) Message boards : Number crunching : apps (Message 1717465)
Posted 7 days ago by Profile HAL9000
ok thanks but it's all a little greek to me. so i guess that meens no stock mb app for linux nvidia . thanks for the info Richard . i've only just started on linux so learning at over 60 can be slow thanks again have a good one!

It seems you have it figured out now. However here is the applications list with some descriptions for the MB apps & AP apps if you find it helpful.
16) Message boards : Number crunching : 12 V overcharge warning message. (Message 1716968)
Posted 8 days ago by Profile HAL9000
A problem here a short while ago.

Got this message a short while ago from AI Suite II, which is the monitoring software for the ASUS motherboard, including temperature.

Used to the warning messages about the temperature, this time I rather got a warning message about a 0.576 V overcharge on the 12 V on the motherboard.

For now I suspended the genefer_524288 CUDA task at 95.872 % completed.

Edit: Checking in with the monitoring software, it now is back at 11.904 V.

Should I pay more attention to this problem?

Any advice would be appreciated.

Thanks!

The ATX specification states the tolerance for the 12v line is +/- 5%. So anything from 11.40v to 12.60v is within spec. In practice it seems like many PSU manufactures run closer to 1-2% on the 12v line.
How much of that variance your GPU can handle is a different question. I think it was the 550ti GPUs we found had a very low tolerance to 12v fluctuation.

Also consider that some monitoring software can go off sometimes. HWMonitor from the people that make CPUz sometimes tells me that my CPU fan is running 750,000 RPM.
17) Message boards : Number crunching : Cricket Graphs? (Message 1716588)
Posted 9 days ago by Profile HAL9000
Yes, they mentioned the attempt, but I wasn't sure if it was actually successful.

Tracert isn't going anywhere near HE @ Palo Alto any more. Instead, we're taking the 'cenic' route direct to Berkeley.

I haven't seen HE in my trace routes since the hardware change at the colo.
I have been getting something like this since spring.
13 73 ms 74 ms 74 ms dc-oak-agg4--svl-agg4-100ge.cenic.net [137.164.46.144]
14 74 ms 73 ms 74 ms dc-oak-agg4--svl-agg4-100ge.cenic.net [137.164.46.144]
15 75 ms 74 ms 74 ms ucb--oak-agg4-10g.cenic.net [137.164.50.31]
16 75 ms 74 ms 75 ms et3-48.inr-311-ewdc.berkeley.edu [128.32.0.101]

Depends on which IP you traced. The web server was always on campus lines, the upload/download servers were not. They are now on campus.

Last I checked the addresses are setiboincdata.ssl.berkeley.edu for upload & boinc2.ssl.berkeley.edu for download. I did save the trace as boinc2.ssl.berkeley.edu.txt when we were trying to figure out where the new location of the bandwidth graphs might turn up. However I did truncate the output so it starts at the 5th line. So there is a chance I mislabeled it.
18) Message boards : Number crunching : Cricket Graphs? (Message 1716484)
Posted 9 days ago by Profile HAL9000
Yes, they mentioned the attempt, but I wasn't sure if it was actually successful.

Tracert isn't going anywhere near HE @ Palo Alto any more. Instead, we're taking the 'cenic' route direct to Berkeley.

I haven't seen HE in my trace routes since the hardware change at the colo.
I have been getting something like this since spring.
13 73 ms 74 ms 74 ms dc-oak-agg4--svl-agg4-100ge.cenic.net [137.164.46.144]
14 74 ms 73 ms 74 ms dc-oak-agg4--svl-agg4-100ge.cenic.net [137.164.46.144]
15 75 ms 74 ms 74 ms ucb--oak-agg4-10g.cenic.net [137.164.50.31]
16 75 ms 74 ms 75 ms et3-48.inr-311-ewdc.berkeley.edu [128.32.0.101]
19) Message boards : Number crunching : Cricket Graphs? (Message 1716454)
Posted 9 days ago by Profile HAL9000
Well, now that we have had the great router reshuffle, is there any chance of getting to see our traffic on the crickets, or any other service?

I was wondering that same thing. One thing that I was pretty sure I interpreted correctly is.. we're not using Hurricane Electronics anymore, right? We're using the campus' ISP now?

Could be. I believe there was mention they had a plan to switch at some point in order to save money. I imagine if it has been switched we may not know until we get a "what has been happening" update post.

Seems some missed what Matt said.

Some big news is that campus is, for the first time in more than a decade, allowing SETI@home traffic back on the campus network infrastructure, thus obviating our need to pay for our own ISP. We are attempting this switchover tomorrow morning. Thus there will be more chaos and outages and DNS cache issues but this has been in the works for literal years so we're not stopping now. I apologize if this seems sudden but we weren't sure if this was actually going to happen until this past weekend.


So yes SETI@home is now on the Campus network, that was what the outage was for.

Yes, they mentioned the attempt, but I wasn't sure if it was actually successful.
20) Message boards : Number crunching : Dropped from Top 5% to 25% overnight. (Message 1716435)
Posted 9 days ago by Profile HAL9000
Looking at one of the charts from a stat site. Your RAC looks have been falling for a while now. Probably as we ran out of AP work again & are only doing MB work right now.

Perhaps some GPU application tuning could help you increase your output.


Next 20

Copyright © 2015 University of California