For those of us who like to live on the leading edge, BOINC 6.6.23 is out. The change log can be found here

Of particular interest for GPUgrid is that there are a couple of GPU related changes.

---

BOINC blog

Those mentioned changes *MAY*, I repeat, ***MAY*** have corrected the multi-GPU issue where tasks were taking up to 4 times longer to run. I have been running it for a couple of days now and I have not seen that problem come back.

You still have to tick the local preference for running the GPU, or better still, select preferences here at GPU Grid, hit update, and then update on the client so you pull down the updated preferences. If you did this for 6.6.21 or 22 you should not have to do it again.

There may still be work fetch issues, I have seen one posted on the mailing lists though I have not yet been hit by one (yet) ... other than the standard ones long known (work queue "flexing" for one) ...

It seems to have done the trick with the GPU tasks that seem to run forever. In my book that would make it an improvement on 6.6.20 (the current release version).

Nasty little thing that checkbox. Glad I had local prefs on all my machines. They all started up with it checked as I went from 6.6.20 (default=ticked) to 6.6.23 and it kept the setting.

---

BOINC blog

Apparently a new version on its way with additional debugging messages. They can't work out from the ones it has why it swaps work out in the middle of running it to do something else.

This is something we've been complaining about for a while now. Still I suppose recognising you have a problem is the 1st step...

---

BOINC blog

Apparently a new version on its way with additional debugging messages. They can't work out from the ones it has why it swaps work out in the middle of running it to do something else.

This is something we've been complaining about for a while now. Still I suppose recognising you have a problem is the 1st step...

Bigger problems than that with 6.6.23 ... on my system GPU Grid was building negative debt uncontrollably as the only GPU project on the system. There is a guy on Rosetta that had the opposite problem, he could not get Rosetta work with a 50/50 split between Rosetta and GPU Grid.

They want to tinker with the rules rather than take a step back and examine if we have the right approach in the first place.

They want to tinker with the rules rather than take a step back and examine if we have the right approach in the first place.

This approach always seems to be more convenient. Until a few days or weeks later you realize that you really should have gone for the other option..

MrS

---

Scanning for our furry friends since Jan 2002

They want to tinker with the rules rather than take a step back and examine if we have the right approach in the first place.

This approach always seems to be more convenient. Until a few days or weeks later you realize that you really should have gone for the other option..

MrS

Wasn't it Churchill that said Americans first try everything else before opting for the right solution?

Wasn't it Churchill that said Americans first try everything else before opting for the right solution?

Does it make more sense to find the right solution and then try all the wrong ones? :-)

Wasn't it Churchill that said Americans first try everything else before opting for the right solution?

I don't think this is limited to Americans. Looks like pretty much the standard routine every human would quickly adopt ;)

MrS

---

Scanning for our furry friends since Jan 2002

Wasn't it Churchill that said Americans first try everything else before opting for the right solution?

Does it make more sense to find the right solution and then try all the wrong ones? :-)

Sadly that is actually the normative case. The right solution is right there and is ignored for "political" or "social" reasons.

Just as an example, we tried prohibition with alcohol and proved that that tactic merely increased crime and lost tax revenue. So, what have we done now? Repeated the experiment with drugs.

Just because drugs became legal I am no more likely to go down to the drug store to buy some hash than I was yesterday. And making something illegal does not make it unavailable.

The fundamental problem is that most people would rather cling to their assumptions than to fact facts. Learning is that something you believe in ain't so is a real downer.