Now theres a visual I could do without...

I don't care if it's me, Petri, Raistmer, or someone else entirely. Everyone involved in some small way will have to do the 'jiggly worm EUREKA dash'

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

1 Credit or 1,000,000 RAC. Wat's Da Bigggeee?

I can live with it (though it's kind of a pain) but there are volunteers either leaving the project over it or aborting GUPPI work units by the thousands, neither being good for the long-term. I can empathize as they are/were spending not only thousands on hardware, but for many similar amounts on power to do a certain amount of work. And now suddenly that is cut in half...

As well, as SETI@Home has been tapped to do the processing for BreakThrough Listen, there is now too much work available, but this work is being done at roughly a third the speed it should/could by volunteers' GPUs. If this bottleneck could be reduced or eliminated, we may be able to keep up.

OK, got the obvious out of the way. :^)

I don't care if it's me, Petri, Raistmer, or someone else entirely. Everyone involved in some small way will have to do the 'jiggly worm EUREKA dash'

So it turns out to be a software issue after all. :^D

---

So it turns out to be a software issue after all. :^D

Oh dear, need a lie down, a nap and a cup of tea now. Many years since I saw my doodle referred to as software

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

Hey Jason

Have you recovered from that lie down and hot drink. Hope so.

This might not be the brightest idea I have had for awhile, but another factor that will hurt your RAC is the users who abort their tasks en masse.

This will affect the turnaround time on the workunits crunched and slow down credit assisgned as the aborts will have to go through the system again.

Regards

Hey Jason

Have you recovered from that lie down and hot drink. Hope so.

This might not be the brightest idea I have had for awhile, but another factor that will hurt your RAC is the users who abort their tasks en masse.

This will affect the turnaround time on the workunits crunched and slow down credit assisgned as the aborts will have to go through the system again.

Regards

Yeah, large numbers will affect the patterns, while individuals notsomuch. Mass aborts are a problem for efficiency, but the Boinc mechanism was built to deal with unreliable clients (how well it does that is open to debate).

With so much technological turmoil at once (many avenues), I think extended lie downs and tea breaks are well justified. We can't control what's going on without affecting the result, so watching what's happening (and trying to learn from it) IMO is a good move. [It's not like there are any experts here*, and be skeptical of anyone preaching ideal solutions]

[* note that expertise & skills come from experience, and nobody went through this before afaik]

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

This might not be the brightest idea I have had for awhile, but another factor that will hurt your RAC is the users who abort their tasks en masse.

This will affect the turnaround time on the workunits crunched and slow down credit assisgned as the aborts will have to go through the system again.

The effect of that on credit would be minimal due to RACs extremely slow response to change.
While it does fall faster than it rises, it still falls very, very slowly when the changes are only minor. Even with major changes it takes a long time to "settle" around it's new normal.
So the effect of the aborters on RAC would be negligible IMHO.

---

Grant
Darwin NT

Greetings All

Now you put it that way, it does make sense, learn something new everyday around here lol.

Yep those lie downs are worth it lol...

Regards

Have we hit the bottom?
All my RACs stopped dropping. Some may be moving upwards :)

---

Have we hit the bottom?
All my RACs stopped dropping. Some may be moving upwards :)

We did have a few days of much more Arecibo than GBT work, and now an AP run, so everyone's seems to be on the rise.

---

We did have a few days of much more Arecibo than GBT work, and now an AP run, so everyone's seems to be on the rise.

Mine hadn't started to rise, but it looked like it might have stopped falling.

Decided to do some Beta work, that way when I move back to main my RAC will have fallen enough to climb again.

---

Grant
Darwin NT

This is ridiculous...going from RAC 4500-5000 to now 2700-3000? Same computers, same GPUs...nothing changed, only SoG & sah's! :(

WTF?!

---

non-profit org. Play4Life in Zagreb, Croatia, EU

Well I am running the Open Beta test: SoG for NVidia, Lunatics v0.45 on my main machine

Since I started on the 29th May my RAC has gone from 6,459 to 10,034

It will drop a bit today as I have been running exclusive app most of the day.

---

This is ridiculous...going from RAC 4500-5000 to now 2700-3000? Same computers, same GPUs...nothing changed, only SoG & sah's! :(

WTF?!

Hmmm, another one with Low End nVidia cards.
Have you checked your results lately? Look at this one; http://setiathome.berkeley.edu/results.php?hostid=7785610&offset=60
You have two tasks there where the GPU accomplished Nothing for over 3 Days at a time. It looks as though it is again doing Nothing on another opencl_nvidia_SoG.
Look at your other host; http://setiathome.berkeley.edu/results.php?hostid=7422266&offset=60
Note how the guppis are taking 4.5 hours running the opencl_nvidia_SoG tasks. How many tasks are you running at once? I think the CUDA App can finish them faster, it does on similar cards anyway. Looking at the APR for that card it is showing the SoG App is somehow faster than CUDA. My guess is you where running Two CUDA tasks which lowered your APR and now the Single SoG appears faster to the Server. How many CUDA tasks were you running and how many SoG tasks are you running?
I would suggest the same thing I did before with the other User, Run the Current Lunatics installer on those two machines and install either the CUDA42 or CUDA50 App. Of course it all depends on how many tasks you where running of each App, but 4.5 hours is even long on that card if you were running two SoG tasks.

I have simular results with my GT430, with the latest non beta Lunatics. I only run 1 task. This sucks but I don't care because I feel the GBs data may be the best we have ever had.

---

This is ridiculous...going from RAC 4500-5000 to now 2700-3000? Same computers, same GPUs...nothing changed, only SoG & sah's! :(

WTF?!

Hmmm, another one with Low End nVidia cards.
Have you checked your results lately? Look at this one; http://setiathome.berkeley.edu/results.php?hostid=7785610&offset=60
You have two tasks there where the GPU accomplished Nothing for over 3 Days at a time. It looks as though it is again doing Nothing on another opencl_nvidia_SoG.
Look at your other host; http://setiathome.berkeley.edu/results.php?hostid=7422266&offset=60
Note how the guppis are taking 4.5 hours running the opencl_nvidia_SoG tasks. How many tasks are you running at once? I think the CUDA App can finish them faster, it does on similar cards anyway. Looking at the APR for that card it is showing the SoG App is somehow faster than CUDA. My guess is you where running Two CUDA tasks which lowered your APR and now the Single SoG appears faster to the Server. How many CUDA tasks were you running and how many SoG tasks are you running?
I would suggest the same thing I did before with the other User, Run the Current Lunatics installer on those two machines and install either the CUDA42 or CUDA50 App. Of course it all depends on how many tasks you where running of each App, but 4.5 hours is even long on that card if you were running two SoG tasks.

1st of all, don't want to "baby sit" SETi@home on BOINC...so it's THEIR job to get the program & science done correctly!
2nd, yes it's a low end GT520 card...it has done it's job correctly so far...problems arrive with SoG & sah jobs!
3rd, I've run 2xCUDA jobs on Quadro2000 & GT730 cards...but only single jobs on all other cards...also running single SoG & sah jobs...so it's not that!
4th, NO SoG app is not faster than CUDA job! Managed to see it myself so far...& nobody can tell me differently! Not to mention that SoG & sah uses a whole CPU...& nobody in SETi@home has done anything about it, even though several users requested that app has to be fixed!!!

can we somehow "boycott" the SoG & sah jobs?!

---

non-profit org. Play4Life in Zagreb, Croatia, EU

can we somehow "boycott" the SoG & sah jobs?!

u could use Lunarics Apps and can choose between cuda32,42,50 and SoG apps,

for sure.

But with the newest Lunarics my SoGs (also guppies) are always as long as
cuda 5.0 workunits.

On Lunarics my gtx 670 took up to 18 minutes to complete one workunit.

---

can we somehow "boycott" the SoG & sah jobs?!

If you're using Windows 7-10, run the Lunatics v0.44 app that will modify the content of C:\ProgramData\BOINC\projects\setiathome.berkeley.edu

Not only will you be able to crunch on the GPU exclusively with the CUDA apps, but it will also optimize your CPU apps!
On my Xeon W3550, my CPU assigned WUs crunched almost 50% faster with the SSE3 app that it installed and configured.

In your case, the beauty of using Lunatics v0.44 is that it doesn't even offer the choice for crunching with SoG :-D

Cheers,
Rob :-}

---

can we somehow "boycott" the SoG & sah jobs?!

If you're using Windows 7-10, run the Lunatics v0.44 app that will modify the content of C:\ProgramData\BOINC\projects\setiathome.berkeley.edu

Not only will you be able to crunch on the GPU exclusively with the CUDA apps, but it will also optimize your CPU apps!
On my Xeon W3550, my CPU assigned WUs crunched almost 50% faster with the SSE3 app that it installed and configured.

In your case, the beauty of using Lunatics v0.44 is that it doesn't even offer the choice for crunching with SoG :-D

Cheers,
Rob :-}

Make sure you flush your cache down to empty before doing that. Lunatics v0.44 won't preserve work issued as v8.12 - i.e. NVidia SoG and SAH work issued after 18 May 2016, 1:10:51 UTC. Lunatics v0.45 will preserve that work, and only deploy SoG if you actively choose it. There should be a Beta4 sometime soon, but I'm waiting on positive feedback from this morning's bugfix before releasing.

Well, there's no v0.44 in download area of Lunatics...

---

non-profit org. Play4Life in Zagreb, Croatia, EU

I just clicked on the link & there it was.
Tried again just now & it's still there.

---

Grant
Darwin NT