Message boards :
Number crunching :
Anyone else only getting half their usual work buffer since the feeder error?
Message board moderation
Author | Message |
---|---|
Send message Joined: 23 Aug 11 Posts: 7 Credit: 1,217,109,541 RAC: 0 |
As the title says before the feeder problem I was getting a total of 80 work units cached in buffer at a time now it's just 40 for my ATI cards. Same PC's same settings as before. An argument could easily be made for double the cache amount not halved, especially with the original milkyway WU's that I use that take about 20 - 25 secs each to run. The way it was before they'd run out for some reason or another three or four times a week or more anyway. Now with half the usual cache amount I have about a little over 14.5 minutes supply worth which will cause a ridiculous amount of short supply problems for me. Anyone else seeing this? |
Send message Joined: 18 Jul 09 Posts: 300 Credit: 303,562,776 RAC: 0 |
I have never had more than 40 |
Send message Joined: 23 Aug 11 Posts: 7 Credit: 1,217,109,541 RAC: 0 |
Mmm odd i've been getting 80 for as long as I can remember on all my PC's. My only guess is that I should have been getting 40 as well all along but because I have an Nvidia & an ATI card in each PC it was seen as from boinc point of view that I had 2 GPU's & thus 80 Wu's it works like that on other projects too. During the outage I did upgrade to the latest stable boinc version so maybe my good luck ran out? I'll try downgrading boinc on one of my PC's to see if my good fortune returns. Even 80 is still one heck of a short cache. I mean I realize they need a quick turn around but under 15mins worth per GPU? Doesn't anyone else think thats a tad light especially for original milkyway WU's & higher end GPU's? Thanks for your reply swiftmallard. |
Send message Joined: 2 Jul 14 Posts: 15 Credit: 20,991,384 RAC: 3 |
That's pretty darn fast, even for a good GPU! Are you sure the tasks are validating and not just erroring out? I'm getting off-topic... After the feeder comes back from an outage, it's usually overloaded. It has to rebuild the work cache for the scheduler as requests come in. Yesterday, the entire MilkyWay project was running out of work at about 400 tasks per application. Today it's back at about 37,000 for the N-Body Simulation. I don't have a graphics card, but maybe the application that runs on GPUs is just running out of work, and doesn't want to give all the work to the same person. (tasks have to validate with more than 1 person) |
©2024 Astroinformatics Group