Boinc 6.10.36 has today become the Recommended version: BOINC Downloads
Boinc version history is available here: BOINC version history

Claggy

Just taken that off my machine and gone back to 6.10.32
not sure what was going on, or if it is just my machine, but 1/2 hour cuda tasks started taking an extra 3-4 minutes with 6.10.36
Has there been any major change in the way that boinc handles cuda tasks?

Has there been any major change in the way that boinc handles cuda tasks?

BOINC doesn't do any of the calculations, so if they're taking longer, it's either different CUDA tasks, or the science application (Seti's).

---

I'm not seeing this on my cuda machine (which I upgraded to 6.10.36 a couple of hours ago).

-Dave

---

Could just be the type of Cuda tasks I've been processing then......
26Ja....tasks were taking 28-30 minutes to complete, then from the same batch started taking about 33 minutes to complete after I upgraded.
Was just wondering if anything could have changed as this is the only thing thats on my machine.
I'll have to try to narrow it down to something else though
Thanks for your help

Pulled down this release and installed just this morning. Really like the ability to idle the GPU separate from the CPU.

All is good otherwise.

Regards,

---

i loaded the new bonic, return to see how everything is going 6 hours later to find 80+ "half cooked" work units! wtf.

---

...Really like the ability to idle the GPU separate from the CPU...

would have been beter if you could idle the cpu seperate from the gpu.

---

Is anyone focussing on "Communicating with Boinc Client. Please wait..."

Your heading towards a major problem ...

L.

---

...Really like the ability to idle the GPU separate from the CPU...

would have been beter if you could idle the cpu seperate from the gpu.

It was discussed, but some of the CPU is always required to feed the GPU, do the communication and house cleaning chores. In some cases a lot of the CPU is needed e.g. SETI's ATI/AP hybrid app. were the CPU does much of the crunching and the GPU is used only to speed up some of calculations.

---

Boinc V7.2.42
Win7 i5 3.33G 4GB, GTX470

Has there been any major change in the way that boinc handles cuda tasks?

BOINC doesn't do any of the calculations, so if they're taking longer, it's either different CUDA tasks, or the science application (Seti's).

But BOINC does GPU queries. This may slowdown science app computations, worth to keep this in mind.

I'm getting a lot of "half cooked" work units on all my rigs, this is silly. cuda work units stopping at 99.99%!!!!

also it is doing it to work units which are not the oldest. please fix this.

by half cooked, i mean not completed.

---

I'm getting a lot of "half cooked" work units on all my rigs, this is silly. cuda work units stopping at 99.99%!!!!

also it is doing it to work units which are not the oldest. please fix this.

by half cooked, i mean not completed.

I guess you're having tasks running "high priority", but after some time it changes to some other tasks running "high priority"?

If so, a change that will appear in v6.10.37 will hopefully fix this behaviour, but this version isn't available yet.

---

"I make so many mistakes. But then just think of all the mistakes I don't make, although I might."

I'm getting a lot of "half cooked" work units on all my rigs, this is silly. cuda work units stopping at 99.99%!!!!

also it is doing it to work units which are not the oldest. please fix this.

by half cooked, i mean not completed.

Get lots of them also and had to suspend a unit last night as it kept wanting to run on GPU0 even though it already had a task. This hasn't been fixed yet and really slows everything....

---

Official Abuser of Boinc Buttons...
And no good credit hound!

I guess you're having tasks running "high priority", but after some time it changes to some other tasks running "high priority"?

If so, a change that will appear in v6.10.37 will hopefully fix this behaviour, but this version isn't available yet.

they are not being run high priority, they just say "running".

Get lots of them also and had to suspend a unit last night as it kept wanting to run on GPU0 even though it already had a task. This hasn't been fixed yet and really slows everything....

i get that too, i didn't know if that was the cuda driver or if it was Bonic.

---

I'm getting a lot of "half cooked" work units on all my rigs, this is silly. cuda work units stopping at 99.99%!!!!

also it is doing it to work units which are not the oldest. please fix this.

by half cooked, i mean not completed.

I have one rig with 4 GPUs, and it got 70 "waiting to run" GPU WUs. I suspended all other GPU WUs in order to work through the backlog. None of the suspended ones started where they left off, not even those at 100%! All started at 0%! With CPU WUs, suspended WUs will continue relative to the percentage complete at the time of suspension. With GPU WUs this never happens, and suspended WUs are in effect being chrunched twice!

After my cleanup, the same thing happens - the suspended list keeps growing - fast! It seems that when two WUs are completed about the same time, all four GPUs are cleaned for new ones - even if there are WUs at 100%!

All WUs are running normal priority.

Previously I saw this behaviour during downloads, but this is on a system where there have not been downloads for several hours.

I don't know which version to revert to as I don't know which version this behavior started.

In any case this should be addressed quickly!

Morten

---

Morten Ross

Is anyone focussing on "Communicating with Boinc Client. Please wait..."

Your heading towards a major problem ...

L.

This will only happen when you have a fair amount of WU's

---

TThrottle Control your temperatures. BoincTasks The best way to view BOINC. Anza Borrego Desert hiking.

Can someone solve this problem: Upgraded to 6.10.36 on my MacBook Pro and can't upload. Keep getting "temporarily failed upload of....:can't resolve host name". Thanks

---

Now I see it's happening to my other Mac. You guys having a server prob?

---

Now I see it's happening to my other Mac. You guys having a server prob?

Not here:

08/03/2010 16:14:32|SETI@home|Computation for task 11mr07ad.27819.10706.16.10.175_2 finished
08/03/2010 16:14:34|SETI@home|[file_xfer] Started upload of file 11mr07ad.27819.10706.16.10.175_2_0
08/03/2010 16:14:39|SETI@home|[file_xfer] Finished upload of file 11mr07ad.27819.10706.16.10.175_2_0
08/03/2010 16:14:39|SETI@home|[file_xfer] Throughput 19673 bytes/sec

"Can't resolve host name" is usually a local DNS error, in your server (if you have one), router or ISP. The upload server is 'setiboincdata.ssl.berkeley.edu' - what happens if you try to ping or tracert to that address?