Just checked pending tasks on my main cruncher (7785479) and noticed this.


SETI@home v8 v8.00 (opencl_ati5_nocal)windows_intelx86


And I have a full cache of them.

Happy days are back....

---

Got 13 GPU tasks on my Windows 10 PC with Radeon graphics.
Tullio

---

Got 19 cached. 7 pending. Average run time looks in the upper 700's for me.

---

That RTS buffer is falling off pretty quick though.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

That RTS buffer is falling off pretty quick though.

Not as fast as I thought it might, though. It's only dropped by 50% in ~4 hours. In fact, in the last half hour it's actually increased by 1K. I do periodically get the "No tasks are available" message, though, when the feeder is apparently drained.

---

I do periodically get the "No tasks are available" message, though, when the feeder is apparently drained.

There are a lot of VLARs about, so you'll get the "No tasks available" message if you're requesting GPU work and VLARs are all that's available.

---

Grant
Darwin NT

Ok so no GPU WU for weeks, Now that I have a bunch of V8's for my GPU I get 13 V7s also for my GPU. Where were they for the last 2 weeks..lol

Bob

---

Sometimes I wonder, what happened to all the people I gave directions to?
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
SETI@home classic workunits 4,321
SETI@home classic CPU time 22,169 hours

My main rig, that was set back to stock apps recently after a hard drive failure, is fully loaded now with MB's and I'm just waiting 2 Einstein tasks to finish on a GPU before going all out on them in an attempt to see if it'll sort out that "New Credit" crap before changing back to running optimised apps again (wishful thinking on my part there).

I'll leave my backup rig as it is until I'm sure that the new optimised apps are working well on most other peoples' rigs before I change that over.

Cheers.

With what I've crunched so far, I'm seeing a considerably shorter times with my HD7870 and R9-380.

With V7 tasks the 7870 was taking 900-1000 second range and the R9-380 was in the 550-700 second range.

the V8's are being done in the 750-850 (7870) and 400-550 (R9-380).

Unfortunately less time = less credit. So while I crunch more tasks, the overall credit per hour doesn't change.

Give me my Astropulse back please.

---

Unfortunately less time = less credit.

The way Credit New works is very mysterious.
What should happen is you get credit for doing a certain amount of work. So the more work you do per hour, the more credit you will get. Doing each WU faster allows you to do more WUs per hour. Hence more credit.

---

Grant
Darwin NT

It's just CreditScrew

---

That RTS buffer is falling off pretty quick though.

Not as fast as I thought it might, though. It's only dropped by 50% in ~4 hours. In fact, in the last half hour it's actually increased by 1K. I do periodically get the "No tasks are available" message, though, when the feeder is apparently drained.

I think what's helping is the fact that it is a new application, so initial estimates are sky high, so most people, even multi-GPUs, can only get a few WUs at a time until those estimates stabilize.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

That RTS buffer is falling off pretty quick though.

Not as fast as I thought it might, though. It's only dropped by 50% in ~4 hours. In fact, in the last half hour it's actually increased by 1K. I do periodically get the "No tasks are available" message, though, when the feeder is apparently drained.

I think what's helping is the fact that it is a new application, so initial estimates are sky high, so most people, even multi-GPUs, can only get a few WUs at a time until those estimates stabilize.

I also think it's very hard to tell what the splitters are actually doing when v7 and v8 figures are still being averaged together......

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

DOn't forget that new BOINCs go on extended backup times if they can't get tasks. they max out at 24h - so it will be a while before all the unattended machines are back crunching.

---

A person who won't read has no advantage over one who can't read. (Mark Twain)

Exeter, (GTX-760 system, Win XP Pro x64), just completed two v8 GPU tasks. 18:01, and 18:00 minutes each on Lunatics 0.44, CUDA50. :-) (System running two WUs at a time.)

Prometheus, (GTX-750 TI SC system, Win 7 Pro x64), seems to be completing two at a time in 24:30 min, to 30 min each on Lunatics 0.44, CUDA50. (System running two WUs at a time.)

MOST EXCELLENT!!!!! :-)


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

I'm not at home right now, but I am loaded with v8 GPU WUs. I forgot to reset my fan speed this morning, they should be o.k. at default. A full boat of APs would only push them to 79 or 80C anyway.

On the other hand, I had some bad v7 WUs go through my system. Don't know what happened...unusual....

Anyway, looking forward to getting home to check things out. It looks like the snow is going to send us home early.

Happy crunching folks!

---

Now that the GPU version of V8 is out. How can you get it to run multiple instances on GPUs?

---

The best way is to run BOINC v7.0.40 or later, and add an Application configuration (app_config.xml) file. There are plenty of examples posted, both here and at Beta: we'll try and gather some decent ones into a coherent form when we've recovered from the last three weeks.

Now that the GPU version of V8 is out. How can you get it to run multiple instances on GPUs?

If you are running NVIDIA GPU:

------------ "app_config.xml" ------------

<app_config>
<app>
<name>astropulse_v7</name>
<max_concurrent>2</max_concurrent>
<gpu_versions>
<gpu_usage>.5</gpu_usage>
<cpu_usage>.5</cpu_usage>
</gpu_versions>
</app>
<app>
<name>setiathome_v8</name>
<max_concurrent>2</max_concurrent>
<gpu_versions>
<gpu_usage>0.50</gpu_usage>
<cpu_usage>0.04</cpu_usage>
</gpu_versions>
</app>
<app>
<name>setiathome_v7</name>
<max_concurrent>2</max_concurrent>
<gpu_versions>
<gpu_usage>0.50</gpu_usage>
<cpu_usage>0.04</cpu_usage>
</gpu_versions>
</app>
</app_config>

------------ End "app_config.xml" ------------

Copy the above into Notepad, Save As ---> All Files ---> ANSI, save file in BOINC ---> Projects ---> setiathome.berkeley.edu Folder. Click "Yes"/"OK" to prompt(s). Then have BOINC Reread Config Files; or, Suspend SETI@Home, Exit BOINC, Restart BOINC, Resume SETI@Home.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

That's a great start, thank you. Note that by setting two tasks per GPU, and a maximum of two tasks of each type running at the same time, this example is really designed for people running SETI on a single GPU only, and not on their CPU. People using more than one GPU, or crunching on their CPUs as well, may wish to increase the values shown for <max_concurrent>.