Which is the one of the v7 that you like to run?

And why go dry to switch?

I thought that would be if one was going from v7 to v6.

Claggy mentioned this is one of the BOINC v7 threads.

If you don't have flops values in your app_info, you'll want to run your GPU cache down first before upgrading,
otherwise existing GPU Wu's will be on the verge of going Maximum Time Exceeded, (because of a increased internal flops value for GPU Wu's)
New GPU Wu's will get revised <rsc_fpops_est> and <rsc_fpops_bound> values.

Or Reset the project once you're upgraded (only needed to be done at projects where you run an app_info that doesn't have a flops entry).

I think v7 also changes the client_state.xml so that if you want to go back to v6 you need to be sure to back up everything before hand.

After v7.0.40 app_config.xml was added to give easier control to running multiple instead of tasks on hardware. I'm planning to switch all of my 6.12.33 machines over once there is a newer recommended version, or I might just use the 7.0.42 I have been playing with on one machine. Since there have not been a lot of messages about newer versions of BOINC having major issues I would guess it is going well.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

This is the reason to at least run down your GPU work

http://setiathome.berkeley.edu/forum_thread.php?id=71160

It's only GPU work that is effected, so you only need to run your Cache down of GPU work.

Claggy

This is the reason to at least run down your GPU work

http://setiathome.berkeley.edu/forum_thread.php?id=71160

Nice, but I don't supply a flops value, haven't for more than 6 months to 1 year in fact.

Which means you WILL need to run down you GPU cache

---

So since I have no flops value which I've read wasn't a good idea, is now recommended practice?

No.
You just need to run down your GPU work.
Once that is done, upgrade BOINC & there is still no need for the FLOPS value.


You will need to change your cache settings on the website/locally though.
The values become reversed.

Maintain enough tasks to keep busy for at least y days.
and up to an additional x days
For me, it's y= 0.2 & x=4. Changing to V7 they would become y=4 & x=0.2

---

Grant
Darwin NT

Well I can't see a way to run down the gpu work by themselves, all I can do is run both down to where I run out of gpu work and report all the gpu work after uploading them.

There may be a better way, but just setting No New Tasks will mean you'll run out of GPU work long before you run out of CPU work.
Once all the GPU work is gone, press update to report it all, upgrade BOINC, then allow new tasks again.
That's how i'd do it.

---

Grant
Darwin NT

Has anyone got downloads that aren't downloading?

Hmmm. Now they're gone.
Had about 5 on one machine that just sat there for about 5min, a couple on the other machine that were there for a minute or 2, then they all decided to download.

---

Grant
Darwin NT

This is the reason to at least run down your GPU work

http://setiathome.berkeley.edu/forum_thread.php?id=71160

Nice, but I don't supply a flops value, haven't for more than 6 months to 1 year in fact.

Which means you WILL need to run down you GPU cache

Well I can't see a way to run down the gpu work by themselves, all I can do is run both down to where I run out of gpu work and report all the gpu work after uploading them.

You can also deselect 'Use Nvidia GPU', so that you can't ask for Nvidia work, once you're completed your Nvidia work, upgrade and reselect 'Use Nvidia GPU'.

Claggy

Not all of the MB splitters are running, and with the present shorty storm the Ready-to-send buffer is dropping rapidly; the splitters aren't keeping up. Could be short of work in a few hours if the shorties continue in their present numbers.

---

Grant
Darwin NT

Yuup, it's nearly quarter to 5 pm in Berkeleyland. getting ready to run out of workunits for the weekend.

---