Posts by Rob.B
log in
1) Message boards : Number crunching : Server Status (Message 8431)
Posted 2770 days ago by Rob.B
I too live out in the sticks in the UK, and our ISP are poor compared to yours, but I have a choice. I doubt your only option is to spent that much cash! As for being a competative thread, it probably is. But why go on and on about op' app's when no one can connect.

Matter of priority I'm afraid!

Sort your ISP.

Rob.B
2) Message boards : Number crunching : Server Status (Message 8395)
Posted 2772 days ago by Rob.B
Nice to see that this thread is active! (Not) What use are optimised app's if the servers are not accessible?

Rob.
3) Message boards : Number crunching : Server Status (Message 8379)
Posted 2773 days ago by Rob.B
Nameserver trace for boinc.thesonntags.com:

* Looking for who is responsible for root zone and followed a.root-servers.net.
* Looking for who is responsible for com and followed f.gtld-servers.net.
* Looking for who is responsible for thesonntags.com and followed ns2.zoneedit.com.

Nameservers for boinc.thesonntags.com:

* ns17.zoneedit.com returned (SERVFAIL)
* ns2.zoneedit.com returned (SERVFAIL)


This is what I am seeing, looks like zoneedit's nameservers are non too happy!!
4) Message boards : Number crunching : Server Status (Message 8377)
Posted 2773 days ago by Rob.B
I think ithis is more ISP related. You are getting DNS errors fom my opsition in the UK, and you bandwidth throughput is falling. Perhaps your ISP is slowly applying a throttle to your account. Check your TOS (Terms Of Service) and get the salses lackey on the phone. He is more likely to sort it out if he sees commission at risk. Tech/Helpdesk do/say whats in the script in my experience. The sales guy (if he is any good) will be able to go around the beaurocracy to the heart of the issue.
5) Message boards : Number crunching : BOINC 6.10.6 is available (Message 1804)
Posted 3038 days ago by Rob.B
Sorry for the rant, but I figure here is a decent place as actual USERS populate this newsgroup.


You won't get any grief from me, as I truly believe the project lacks technical learership (sorry Dr Anderson), but I'll give 6.10.6 a a realistic try as it seems the best of a bad bunch.

Rob.B
6) Message boards : Number crunching : BOINC 6.10.6 is available (Message 1799)
Posted 3038 days ago by Rob.B
So far so good, no real insanity but it has started eight jobs (all with the same time stamp - Rosetta) and seems to be a little confused as to which of these has priority. Other than that nothing to bad report.

Projects running
Collatz CPU CUDA
SETI CPU CUDA
GPUGRID CUDA
Einstien CPU
Climate Prediction CPU
Maleria Control CPU
Leiden Classical CPU
Milky Way CPU
Rosetta CPU
U(micro)fluids CPU
ABC CPU
Prime Grid CPU

No particular project order, just as I remembered them, I hope that this diverse load of a dozen projects should be a tough enough test. I will run this config' until Monday Sept 28th, my RAC is stuffed anyway so what the hell!

Rob.
7) Message boards : Number crunching : BOINC 6.10.6 is available (Message 1751)
Posted 3038 days ago by Rob.B
I installed the v6.10.6 Client on 15 Box's without needing a clean install & every one of them is running fine. So you may have had other issues with the BOINC Installation you already had that forced a clean install.


I tend to agree, mind you with all the WU's from a dozen different projects Boinc 6.20X will be out before my RAC recovers!!

Rob.
8) Message boards : Number crunching : BOINC 6.10.6 is available (Message 1748)
Posted 3038 days ago by Rob.B
Well I have activated every project I have ever run, usually I run two or three at most,in order to try and drive it insane, so far it seems OK. Very odd, given performace the previous few of clients to be released. The only thing I can think is that sonething or other (possibly Long Term Debt) has been screwed up by one of the previous versions and the clean install has (as you would expect) cleared it down.

I will keep an eye open for odd behaviour.

Rob.
9) Message boards : Number crunching : BOINC 6.10.6 is available (Message 1702)
Posted 3038 days ago by Rob.B
OK I now have this running something like, to get to this state I effectively ran down all my cashe's, uninstalled Boinc then did a clean install. I then put back my op' app's and app_info.xml and sanity (or near enough) is restored.

6.10.6 is now downloading and balancing jobs between 3 CUDA projects, and the CPU cash is now active, balancing jobs between 10 projects, swapping in and out as per the set perameters.

Pity it needed a clean install though!

Rob.
10) Message boards : Number crunching : Error while computing collatz v2.01 (cuda23) (Message 1620)
Posted 3039 days ago by Rob.B
I seem to be having around 30% failing at around 1 or 2 seconds in at the moment, I have 2 cards a 9800GTX+ and a 9400GT. The failures seem distributed fairly evenly across both cards as a percentage (the 9800 fails around 4 times more WS's as the 9400, but it is around 4 times as quick).

No SLI, no config' changes. This seems to have started over night. (12 to 14 hours ago). A reboot has achieved nothing with respect to this issue. (Vista Ultimate 32 bit)

Rob.
11) Message boards : Number crunching : BOINC 6.10.6 is available (Message 1554)
Posted 3040 days ago by Rob.B
Obervations on 6.10.6 so far; the GPU sheduler is not all that happy if you run more thane one GPU project, seeming only to download wu's for one project and not the others. On the CPU side,it seems not to be allowing any form of cashing, just downloading a new WU when a CPU becomes free, sort of on-demand scheduling. On the whole I rate it as pretty poor; but at least it is downloading CPU work.

I'll probably revert to 6.6.31 though.

Rob.
12) Message boards : Number crunching : Official 6.10.5 is still bent (Message 1522)
Posted 3040 days ago by Rob.B
So 2 or 3 days ago they released 6.10.5 and yesterday they released 6.10.6?! What the heck is going over there at Berkeley!


Panic?
13) Message boards : Number crunching : BOINC 6.10.x (Message 1510)
Posted 3040 days ago by Rob.B
Trying 6.10.6, looking fine so far, and (the first of the 6.10.x's not to do this) non of my Phenom's cores are sitting idle!!!!

Rob.B
14) Message boards : Number crunching : Errors with the new CUDA app (Message 1326)
Posted 3042 days ago by Rob.B
Yes I'm on 6.10.5 and its rubbish as have been all of the 6.10's so far, once again I have all 4 CPU cores idle (non collatz cuda is fine) with this version - see the 6.10.x thread for comments on this. Why reset the project? I don't see why I should loose my history graph, I have not had to do this for other projects.

Anyway I'm going back to 6.6.31 yet again, at least that downloads CPU based WU's.
15) Message boards : Number crunching : Errors with the new CUDA app (Message 1322)
Posted 3042 days ago by Rob.B
Hi.

I have switched to GPUGRID for a while until I can sort out why all the Collatz 2 WU's fail within a second or two. CUDA driver 190.36 9400GT and 9800GTX 512Mb on both. Win 32 (Vista Ultimate).


-- Addendum
Just moved to 190.64 driver and tried again, trashed 26 wu's in even less seconds. GPUGRID (cuda23) runs ok.
16) Message boards : Number crunching : BOINC 6.10.x (Message 1138)
Posted 3043 days ago by Rob.B
What it comes down to is that it appears that no one in the development team has the bottle to say "the scheduler is rubbish (I would like to use a four letter word beginning with 'c' but I won't) and that they should let it be, as bad as it is, re-do the spec' then recode from scratch!

Rob.B!
17) Message boards : Number crunching : BOINC 6.10.x (Message 1104)
Posted 3044 days ago by Rob.B
That can only mean that the design is flawed


As a full time professional scheduler in the Print industry, were I to even contemplate buying scheduling software as bad as this I would ne on the street (sidewalk/unemployed). Stuff all the plattitudes about giving choice, this scheduling algorithm is pathetic!

Rob.B
18) Message boards : Number crunching : BOINC 6.10.x (Message 1081)
Posted 3045 days ago by Rob.B
Dropped back to 6.6.24, this is not good. 6.10.4/3 will not download any CPU wu's. 6.6.31 has started all 120 Collatz WU's then suspended them all whilst it cycles through them! Bad? You bet; especially with no checkpointing!!

Thinking of giving up Boinc as this scheduling rubbish (I'm being polite) is beyond a joke!

Phenom x 4 core cpu
9400 gpu x 1
9800 gtx+ x 1 gpu
4 gb RAM
Vista (32 bit) ultimate

And it still can't schedule!!!!!!!!!!!!

Rob
19) Message boards : Number crunching : BOINC 6.10.x (Message 1063)
Posted 3045 days ago by Rob.B
Spoke too soon! This version seems quite happy not to download any CPU work and leave all 4 cores idle. As with 6.10.3 the GPU scheduling seems much improved. Back to 6.6.31 for me (again)!
20) Message boards : Number crunching : BOINC 6.10.x (Message 1056)
Posted 3046 days ago by Rob.B
Running 6.10.4, well best of the 6.10.x's so far! No unexpected file deletions on install, and nearly 24 hr's of running with no "wierdness" that I have noticed. Only comment is; as I have a quad with two GPU, it has detected this and reduced my cashe form one-hundred-and-twenty (120) to sixty (60) ws's. Apart from this no complaints ..................... Yet!

Rob.B


Next 20

Main page · Your account · Message boards


Copyright © 2018 Jon Sonntag; All rights reserved.