No credit issued for most of November

Wandering
Wandering
Joined: 10 Jun 08
Posts: 2
Credit: 22002568
RAC: 0
Topic 220076

I have more than five pages of returned tasks waiting for validation. My average points per day has tanked from well over 100,000 to just over 60,000. More than -40%. When do you plan to issued this credit? I have suspended the project for now, as there is little point in wasting this time. Is there a problem? What's going on?

San-Fernando-Valley
San-Fernando-Valley
Joined: 16 Mar 16
Posts: 260
Credit: 6915701637
RAC: 20325672

Please help me understand

Please help me understand your point of concern:

What do you mean by "... waisting this time"?

 

Betreger
Betreger
Joined: 25 Feb 05
Posts: 987
Credit: 1421895891
RAC: 777327

A quick non comprehensive

A quick non comprehensive glance at your computer shows you are processing gravity waves, first prize, which only pay 1000 credits, rather than pulsars, which I call second prize and pay 3465 credits. Additionally the pulsars run much faster. 

Yep when gravity wave became available to GPUs my RAC dropped from apx. 250,000 down to a bit over 100,000.

Bobw
Bobw
Joined: 28 Jan 09
Posts: 5
Credit: 163275838
RAC: 0

Yes my RAC  hsd gone to ruin.

Yes my RAC  hsd gone to ruin.

wujj123456
wujj123456
Joined: 16 Sep 08
Posts: 18
Credit: 1491805294
RAC: 2340297

3465 credits every few

3465 credits every few minutes is actually pretty ridiculous if you compare the rewards with other GPU projects... The 1000 credits per WU feels more inline with other projects, though I think I started seeing gravitational wave 2.03 workloads taking 40 min instead of a bit over 10 min on my 1080Ti starting just today/yesterday.

Betreger
Betreger
Joined: 25 Feb 05
Posts: 987
Credit: 1421895891
RAC: 777327

Gary gave a lucid

Gary gave a lucid explanation, which I am unable to find so I will summerize as best I can.

The BRP tasks are awarded the correct credit when run on CPUs,  3465.

The same task can also be run on a GPU and is awarded the same credit. 

I don't see a problem with that.

The same thing hold for Gravity waves. 

The reason they pay so relatively poorly on GPUs is that they are not as well suited to parallel processing. 

If I've got it wrong please feel to correct. 

Wandering
Wandering
Joined: 10 Jun 08
Posts: 2
Credit: 22002568
RAC: 0

I am not so much concerned

I am not so much concerned with how much they pay, or how fast they run. The problem is as stated. Five pages of stuff waiting to validate. The wasted time was the time I spent running those tasks when I could have run something that mattered to both me and the project. Unvalidated projects are a waste of everyone's time. It is evidenced by the much lower average daily score. Unless I get credit for these tasks, I am wasting my time, and spending money for power that benefits no one. Why the huge backlog of unvalidated tasks? They are not invalid, and they are not in error.

So, for San Fernando Valley, the point is five pages of unvalidated tasks. Why?

Betreger
Betreger
Joined: 25 Feb 05
Posts: 987
Credit: 1421895891
RAC: 777327

My guess they are waiting for

My guess they are waiting for wingmen to report. 77 of them are GWs which are waiting for wingmen. Unfortunately on a bunch of them your first wingman timed out and the work was sent to a third. Methinks you're not wasting processing time. 

Betreger
Betreger
Joined: 25 Feb 05
Posts: 987
Credit: 1421895891
RAC: 777327

My guess they are waiting for

My guess they are waiting for wingmen to report. 77 of them are GWs which are waiting for wingmen. Unfortunately on a bunch of them your first wingman timed out and the work was sent to a third. Methinks you're not wasting processing time. 

wujj123456
wujj123456
Joined: 16 Sep 08
Posts: 18
Credit: 1491805294
RAC: 2340297

Betreger wrote:Gary gave a

Betreger wrote:

Gary gave a lucid explanation, which I am unable to find so I will summerize as best I can.

The BRP tasks are awarded the correct credit when run on CPUs,  3465.

The same task can also be run on a GPU and is awarded the same credit. 

I don't see a problem with that.

The same thing hold for Gravity waves. 

The reason they pay so relatively poorly on GPUs is that they are not as well suited to parallel processing. 

If I've got it wrong please feel to correct. 

Oh, I am not saying the 3465 is wrong. I just meant the exceptional efficiency BRP tasks have on GPU is more of an exception rather than a rule from the few GPU projects I've participated in. So I am not upset with Gravitational Wave's reduction in credits and hopefully most other people wouldn't feel too mad either.

wujj123456
wujj123456
Joined: 16 Sep 08
Posts: 18
Credit: 1491805294
RAC: 2340297

Betreger wrote:My guess they

Betreger wrote:
My guess they are waiting for wingmen to report. 77 of them are GWs which are waiting for wingmen. Unfortunately on a bunch of them your first wingman timed out and the work was sent to a third. Methinks you're not wasting processing time. 

Exactly. I don't think I've ever seen my work not getting rewarded, unless the result is invalid of course. It's just BOINC waiting for second result to cross check and the credits will eventually come when the second result checks out. GPU projects are relatively fast already. There are some long deadline CPU projects I run which takes days to finish and weeks to months before credits arrive. Some projects wait for more than two identical results too. The credits always eventually arrive, so I've stopped worrying about this and just check once a while to ensure my hardware isn't constantly generating invalid results. Invalid results would be a waste of process time and power, but pending validation is not a problem.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.