Sry for spaming around, only want to help finishing the queue :(

I only want to tell i getting tasks again. Seems i hit any limit before..

---

DSKAG Austria Research Team: http://www.research.dskag.at

Since today there is only 1 user left who connected the last 24hours to the short cuda31 queue (Serverstats). Im proud to tell, im this lonely guy ;) So i need at least 3 more (@24h crunching) days to clear this queue up (~4 hours per WU). Only as little estimate when the adminstaff can deactived it, and the problems with the queue selection on some computers should go away then ;)

Sorry to spoil your party but I am getting only cuda 3.1 short runs on one of my system while this pc will accept all sorts jobs.
I think the server page is not corect with this, as yesterday was to see as well, its always 2 left, however my pc did already 4 today and 1 will be ready in the next hours (could be just Saturday in UTC).

---

Greetings from TJ

It counts down on the " in work" unit when i finished something when i remeber right. Perhaps it shows 2 units free to send but all "in work" units are ready to resend and it fills the ready to send queue only until 2 or something. But im happy when someone help me, so we can finish it up faster :) gogogooo

---

DSKAG Austria Research Team: http://www.research.dskag.at

I am a bit confused and I hope it gets resolved soon, I have my lowest power GPU set to only crunch Short Run tasks, but somehow it keeps on being given the Long Run Tasks, even though it has no chance of finishing them even within the 48 hour Bonus time. I hope the settings just didn't set in once I realized Short run 4.2 were available.

---

I am also getting only long runs on my computers that are set to accept only short runs. I have them set to receive long runs only if there are no short runs available, but there seems to be plenty available.

---

XtremeSystems.org - #1 Team in GPUGrid

I am a bit confused and I hope it gets resolved soon, I have my lowest power GPU set to only crunch Short Run tasks, but somehow it keeps on being given the Long Run Tasks, even though it has no chance of finishing them even within the 48 hour Bonus time. I hope the settings just didn't set in once I realized Short run 4.2 were available.

There are three things you must do to prevent getting the long runs. You have to double-check, read carefully and make sure you have done all of these things.

1) uncheck long runs

2) uncheck the box that says "If no work for selected applications is available, accept work from other applications?"

3) click the update button

I am willing to bet you did not do step number 2) or possibly not 3) and I am willing to bet you will post at least 2 more times that it does not work properly before you RTFP and realize you neglected to do 2). The settings work properly for everybody else therefore if they do not work for you then you are not setting them properly.

---

BOINC <<--- credit whores, pedants, alien hunters

I am also getting only long runs on my computers that are set to accept only short runs. I have them set to receive long runs only if there are no short runs available, but there seems to be plenty available.

What makes you think there are plenty short runs available, the server status page? That page lies. Never trust the numbers you see there, never, not at this project or any other project. If you're getting long runs then it's because when your computer requests work there are no short runs. Notice that leaves open the possibility that sometime AFTER your host receives the long runs someone/something puts tasks in the short run queue.

---

BOINC <<--- credit whores, pedants, alien hunters

The server status page doesn't lie, but since it's usually updated every 10 minutes it never shows the current status either (i.e. not giving you the information you're looking for).

If you set up your machines as Dagorath suggested you should have a backup project setup, as you'll likely not get enough work from GPU-Grid to keep your GPU busy.

BTW: GPU-Grid generates WUs based on the results of a previous WU. That's why short return times are important. And that's probably why the amount of "WUs ready to send" always increases by 1 as soon as someone returns a result.

MrS

---

Scanning for our furry friends since Jan 2002

It's normally reasonably reliable (unlike some projects), but it might be telling a few porkies at the minute; some tasks were pulled out manually. Perhaps the server isn't aware of this?
http://www.gpugrid.net/forum_thread.php?id=3285&nowrap=true#28522

I tried to get short tasks at one stage (yesterday or the day before) when it said there were hundreds of tasks, but I was unable to for hours.
http://www.gpugrid.net/forum_thread.php?id=3284
Don't know what the situation is right now though and I don't think there is any point second-guessing what's in the queue - lots of failures, beta testing different task types, 3.1 due to be deprecated... It's likely to keep changing.

Certainly with so many quick failures, task return and re-issue could be rapid.

I tried to get short tasks at one stage (yesterday or the day before) when it said there were hundreds of tasks, but I was unable to for hours...

What is the difference between the two "Short runs (2-3 hours on fastest card)" listed on the server status page?

---

The first one is cuda31 support, "specially" for 2xx cards with drivers <3xx.xx. It will deactivaded soon. The second is cuda42 support for 4xx or better cards with driver >=3xx.xx

You can compute with eatch card in every queue, but i recommend so for the most performanceoutput.

---

DSKAG Austria Research Team: http://www.research.dskag.at

The second is cuda42 support for 4xx or better cards with driver >=3xx.xx

Thanks for the feedback, dskag. That's what I thought. :-)

The problem for me is that I've got two hosts set to request tasks for that app only (in preferences, run test apps? No; only short runs CUDA42 checked; accept other work? No), yet I'm not getting any of those tasks. :-( One host has a 550Ti w/306.97 and the other is a 260 w/310.90.

---

The second is cuda42 support for 4xx or better cards with driver >=3xx.xx

Thanks for the feedback, dskag. That's what I thought. :-)

The problem for me is that I've got two hosts set to request tasks for that app only (in preferences, run test apps? No; only short runs CUDA42 checked; accept other work? No), yet I'm not getting any of those tasks. :-( One host has a 550Ti w/306.97 and the other is a 260 w/310.90.

Same here.. I have both short queues checked yet don't get any tasks. Beta and long queues are disabled

---

Team Belgium

That's a topic already being discussed in the No tasks available for ACEMD short runs cuda 3.2 thread.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

That's a topic already being discussed in the No tasks available for ACEMD short runs cuda 3.2 thread.

Thanks for the feedback... Sorry to have missed that, but there's been so much talk lately about similar things in different threads that it's hard for me to keep track! I will set my prefs to accept both short queues and see what happens.

---

That's a topic already being discussed in the No tasks available for ACEMD short runs cuda 3.2 thread.

Thanks for the feedback... Sorry to have missed that, but there's been so much talk lately about similar things in different threads that it's hard for me to keep track! I will set my prefs to accept both short queues and see what happens.

From reports in this very thread it is evident that sometimes neither of the two short queues has tasks. If your host requests a task when both queues are empty then it won't get one. And remember, even if the server says there are tasks in one of the queues it isn't necessarily so.

---

BOINC <<--- credit whores, pedants, alien hunters

...I will set my prefs to accept both short queues and see what happens.

And I got one...a short run CUDA42, that is. :-)

---

Dagorath,

I understand the settings, and I do have them set how I want them. I was just making a remark as around the time I got those tasks I checked the server status and it appeared that there were plenty of short run tasks ( but yes I realize it is not a perfect exactly up to date page, just a ball park). I do not mind crunching long runs on the card, so I do leave the box checked that says it can grab basically any GPU grid task it finds if my initial request does not come up with a result.

---

Adam,

It's obvious I misunderstood your other message and I'm glad you've got it worked out now. Happy crunching!

---

BOINC <<--- credit whores, pedants, alien hunters