SETI@home News Vol. 111 (23/07/2010)

Man of Honour
Joined
30 Oct 2002
Posts
15,807
Location
Surrey
I have nothing old :( I don't collect things just buy new bin the old :(

Hot chocolate and Jammie D's on the go :D

But will be getting new water cooling gear as at the moment mine is just not cutting it and putting 2 470's in a loop will need either 2 loops or a few rads :)

If my experience is anything to go by the gpu's are not the problem keeping cool. Also better doing one loop and use multiple pumps in serial (latest thinking) as gpu blocks are pretty restrictive.
 
Associate
Joined
29 May 2010
Posts
510
Location
SW Scotland
Mornin' all. Things not looking too bright over at Berkeley. No splitters are running, which = no tasks to send out... Dunno what the plan is (or even if they have one) :(
 
Associate
Joined
29 May 2010
Posts
510
Location
SW Scotland
I'm curious....which indicates a small level of brain activity :confused:

Have had shedloads 650/700 of these cpu tasks seemingly time out on me, i take it the elapsed time limit is in seconds which means round about 50 mins is allowed per task. But being the curious type (and plain bored today) i sat and actually watched them start-progress-time out, they're all erroring out after 5-6 mins.
Could not think of anything i could do except restart boinc, did not want to get too into it and loose the last 3 days work.....any ideas?


Hi Toxic, are these -177 errors? There is a thread over at S@H with a workaround for avoiding these. For the tasks that have -177'd on you, you could manually reset the tasks by editing your client_state.xml file and delete the result files, apply the fix - and re-process. I guess if you have over 600 though, its probably not practical given that today is Friday.......

I've had a strange mix of tasks this time round - quick CPU tasks, and long GPU tasks.
 
Associate
Joined
29 May 2010
Posts
510
Location
SW Scotland
Yay - should give your RAC a nice bump :D
Post back if you have any trouble and Im sure we can get you on the fermi credit superhighway, you will be stomping area51 in no time unless he 'does a grimgoth' and upgrades his rig as well...

......hmmmmm....

Na, Although I do plan to eventually put 2x470's in and run the 2x285's as well (assuming the CUDA dlls will let me do that, and the 285s are still working), it won't be for another 8-12 months. There is no way my setup can cool those things, so I'm starting to think about wc, since It will be a must for me if I'm gonna do the 2x470's. That makes the whole 470 thing veeeeeerrrrry expensive, and I'm a strictly cash only sort of guy. Started to read up on wc, but its so far in the future, its probably not worth doing much more than starting to stash the cash ready for the big day!

It is inevitable that Marine will get me - its just a matter of how quickly, and how fast he goes past!
 
Soldato
Joined
8 Mar 2010
Posts
7,184
Location
TBagTowers
Hi Toxic, are these -177 errors? There is a thread over at S@H with a workaround for avoiding these. For the tasks that have -177'd on you, you could manually reset the tasks by editing your client_state.xml file and delete the result files, apply the fix - and re-process. I guess if you have over 600 though, its probably not practical given that today is Friday.......

I've had a strange mix of tasks this time round - quick CPU tasks, and long GPU tasks.

Morning Area51, not sure what tasks they are tbh the elapsed time limit is 50 minutes but they time out after 4-6 minutes weird. I did not want to go mucking about just yet untill i get my work uploaded, would be my luck to go charging into the client_state.xml and do something wrong.
Soon as all is upped will delve a little deeper, do a much needed case change and be back up and running by this evening.
I thought they were going to start up the splitters and other stuff early this time but hey ho, i wouldn't expect and activity till 2pm ish and the upload/download servers about 3.30pm then iiiiiiiittttsssssss hammer time!!:D
 
Associate
Joined
29 May 2010
Posts
510
Location
SW Scotland
Morning Area51, not sure what tasks they are tbh the elapsed time limit is 50 minutes but they time out after 4-6 minutes weird. I did not want to go mucking about just yet untill i get my work uploaded, would be my luck to go charging into the client_state.xml and do something wrong.
Soon as all is upped will delve a little deeper, do a much needed case change and be back up and running by this evening.
I thought they were going to start up the splitters and other stuff early this time but hey ho, i wouldn't expect and activity till 2pm ish and the upload/download servers about 3.30pm then iiiiiiiittttsssssss hammer time!!:D

Unfortunately, if you upload all the evidence will be gone - until the tasks are listed as a shed load of errors on your tasks page. I'm keeping away from downloads 'till at least Sunday - got plenty in me cache 'till then. It seems crazy that they've not set the ball rolling on the splitting.....

Are you switching across to the HAF case? BTW, nice tidy builds those 800D's allow. Not much obstructing your airflow in them - just not enough or the dang stuff!!!!
 
Soldato
Joined
8 Mar 2010
Posts
7,184
Location
TBagTowers
Dang it, am afraid i will have to risk just uploading the tasks i have and if the problem re occurs then i will look into the client state, no errors today so far thank goodness.
Just opened the hafx case, TBag likes the side plate sized fans!
On first impressions yup i like it will take some pics of the days proceedings and if i haven't had a meltdown by say 7pm will post the shennanigans :)
 
Associate
Joined
21 Dec 2007
Posts
426
Location
Nr Grimsby
Dang it, am afraid i will have to risk just uploading the tasks i have and if the problem re occurs then i will look into the client state, no errors today so far thank goodness.
I'd make a backup of your folder that way you can least play with it later. You wont be able to report them if you get the work around working, as they will already have been, but you'll still be able to try it out. Just make sure you have the names of a good few errored WU's before you report so you know which ones to look for.
 
Soldato
Joined
10 Oct 2005
Posts
4,040
Location
London
seeing some gpu weirdness...
It makes no difference whether I run one, two or 3 wu's per card, but Im seeing my gpu usage drop off after 5 mins crunching everytime I restart boinc.
Im seeing gpu usage on one card of 43% and the other card go as low as 4% :eek:
Usually both cards are over 90% - anyone else seeing varying gpu usage???
Wondering whether it is just the type of wu?
 
Associate
Joined
29 May 2010
Posts
510
Location
SW Scotland
seeing some gpu weirdness...
It makes no difference whether I run one, two or 3 wu's per card, but Im seeing my gpu usage drop off after 5 mins crunching everytime I restart boinc.
Im seeing gpu usage on one card of 43% and the other card go as low as 4% :eek:
Usually both cards are over 90% - anyone else seeing varying gpu usage???
Wondering whether it is just the type of wu?

Mine are behaving as per normal, although I have had some fairly long tasks for them this time round. They're not mega-shorties are they?
 
Soldato
Joined
8 Mar 2010
Posts
7,184
Location
TBagTowers
seeing some gpu weirdness...
It makes no difference whether I run one, two or 3 wu's per card, but Im seeing my gpu usage drop off after 5 mins crunching everytime I restart boinc.
Im seeing gpu usage on one card of 43% and the other card go as low as 4% :eek:
Usually both cards are over 90% - anyone else seeing varying gpu usage???
Wondering whether it is just the type of wu?

Now you mention it regardless of wether i run 2 or 3 wu's at a time my gpu is around 70 percent, i dont know what the heck? At least no errors so far today, now hurry up and upload my pretties before i loose patience and get my screwdriver out :)
 
Associate
Joined
29 May 2010
Posts
510
Location
SW Scotland
Well I gotta say, it looks like the server side peeps did a pretty good job of bringing the project on-line smoothly today. I was concerned by the lack of split tasks, but the splitters appear to have built a buffer pretty quickly, and are maintaining it. I only hope that they increase the Sunday download limits to help reduce the burden on Monday and Tuesday.....time will tell.
 
Soldato
Joined
8 Mar 2010
Posts
7,184
Location
TBagTowers
Still uploading but seems to be going smooth enough, i'd go along with that not getting involved in the scrum tommorrow or sunday, luckily still have a largeish cache which means it won't be click retry click retry untill monday.
Case unboxing was good, untill i started noticing marks all over the insides and even a small chip on the metal work.
In all fairness ocuk were great about it and they dont quality check every single item so this was down to the supplier, quick to respond with an apology and an offer which i thought was good. Must admit i have had afew probs with stuff but oc have allways been quick to step upto the mark and put things right.
Now how much longer are these b......y uploads going to take, i have a pot of coffee a pack of marlboro and a fully charged screwdriver on standby hhhmmmmmm :)
 
Soldato
Joined
10 Oct 2005
Posts
4,040
Location
London
Well I have reported 1649 tasks and Im still uploading, so all is going well..
Pending is upto 531,050.07 and climbing :eek:
Still have gpu weirdness, but will look at that tomorrow..
Good luck with the screwdriver Toxic!
Im off down the pub :p
 
Associate
Joined
29 May 2010
Posts
510
Location
SW Scotland
Well I have reported 1649 tasks and Im still uploading, so all is going well..
Pending is upto 531,050.07 and climbing :eek:
Still have gpu weirdness, but will look at that tomorrow..
Good luck with the screwdriver Toxic!
Im off down the pub :p

Mines at 375k - never been up this high before - lovely view - "I can see the pub from here" :D
 
Back
Top Bottom