UK BOINC Team Forum

As I only have a humble P4 1500 a Seti WU was consistently taking me between 6 3/4 and 7 hours and I was consistently claiming just over 30 credits but receiving less on average due to the way they award it. As a result I have tended to concentrate on CPDN due to the linear relationship of crunching time to RAC, although I still do some SETI crunching for old times sake.

Recently I installed Crunch3r but didn't try it out on any WU's as my CPDN sulphur model had almost reached the end of the first phase and I wanted to see the temperature graph as I appear to have a slight global cooling.

As a result I had 3 Seti WU's where the other three users allocated to the WU had already finished it and therefore the credit had already been determined and awarded. The 3 credits awarded were

19.22 19.38 23.25

This would not have been good news for me considering my consistent claim of around 30. However when I unleashed Crunch3r on the 3 WU's I was amazed to find that they all now took about 2 1/4 hours, only 1/3 as long as before! I got a corresponding drop of claimed credit down to about 10, however my RAC will be the same as before even though the WU's have only taken a third of the time. A check of the WU's involved showed that in only 1 case did anyone claim less than me for any of the WU's.

Therefore I put forward the following hypothesis:

If consistently claiming less than the average credit awarded for each WU it is possible to maximise the actual credit received for each WU by ensuring that the WU is finished in 4th position.

Can anyone else verify or disprove this hypothesis.

Well as SETI works on a quorum of 3 then your 4th result doesn't really play a part of the process of credit as only 3 WU's are needed to process credit.

Your RAC will increase quite quickly if you run SETI all the time on a optimised APP.

Now you say you have downloaded Crunch3r but what the SETI App or BOINC

Robert Boon wrote:Therefore I put forward the following hypothesis:
If consistently claiming less than the average credit awarded for each WU it is possible to maximise the actual credit received for each WU by ensuring that the WU is finished in 4th position. Can anyone else verify or disprove this hypothesis.

The fact is that no-one can "guarantee" that you will maximise the credit amount awarded, whether the WU is returned 1st, 4th, 10th or 195th.

If you return the 4th WU and the 3 other crunchers returned higher scores, then you "win", as your "optimised" SETI client, crunches in less time and hence is generally awarded less credit.

If you return the 4th WU and the 3 other crunchers returned LOWER scores, then you lose out a bit - but it's not significant, as you haven't "wasted" much time to achieve a low "score" - so you can discount it (mentally) and move on.


Of course, if you return a WU quickly, then your own credit award goes into the quorum and depending on whether other crunchers (who've been issued with the same WU) have faster/slower PC's and perhaps use an optimised client, this will dictate what sort of credits the quorum will generate.

Chances are that in most cases (and I haven't got a figure for this) but it is "likely" that most people do NOT use an optimsied SETI client - so they will always spend more time on a WU and hence their time based score will produce a higher credit amount claim than your optimised client.

So, the chances are, in most cases that those using an optimised client will always get more credit for a quickly returned WU, as the quorum will mainly comprise people with non-opti clients generating higher credits.



In the end though, I prefer to discount the issue of whether I'm getting better or worse credit scores - I'm more concerned at returning a higher number of valid WU's in a given time, as that way, we can search more WU's for that elusive signal...!


regards,

Tim

Robert Boon wrote:If consistently claiming less than the average credit awarded for each WU it is possible to maximise the actual credit received for each WU by ensuring that the WU is finished in 4th position.

Can anyone else verify or disprove this hypothesis.

Yep, thats exactly what I do, I run with a 5 day cache which almost guarantees i'm the 4th result and it tends to get a higher granted than claimed.

You could always get the best of both worlds by also using Truxs Calibrating Boinc Client which adjusts the claimed credit based on actual work done rather than cobblestones.
You end up claiming the same as a non optimised application.

And as for your RAC, just watch it climb.

I wouldn't recommend a optmised BOINC client they play havoc with other projects results the apps are ok but the clients are problomatic

UBT - Halifax--lad wrote:I wouldn't recommend a optmised BOINC client they play havoc with other projects results the apps are ok but the clients are problomatic

I thought Truxs Calibrating client got round those issues, doesn't it?

But yes, any other opt boinc client will screw around with other projects claimed credit, I'm currently only claiming 1/2 the norm for Einstein for example.

good point

The optimised BOINC clients inflate the credits which in a way is cheating, I won't touch any of them with a bargepole.

Mainly because I test the new BOINC clients

UBT - Halifax--lad wrote:The optimised BOINC clients inflate the credits which in a way is cheating, I won't touch any of them with a bargepole.

thats a bold statement
In general, they try to take into account the techniques of the optimised science applications.
The standard boinc client doesn't know about the speed ups in the opt apps and claims credit on what it thinks happened.
The optimised clients take into account the extra work of the opt apps, thats all.

If the standard boinc client would award say 20 cedits for a WU, why shouldn't an optimised boinc client award the same

its a bold statement but it is true, it has been looked into several times and the statement is quite correct an optimised BOINC client does inflate results.

People are best off sticking with the offical app as it changes that often as it is

UBT - Halifax--lad wrote:its a bold statement but it is true, it has been looked into several times and the statement is quite correct an optimised BOINC client does inflate results.

People are best off sticking with the offical app as it changes that often as it is

haha, I meant the cheating bit
no doubt about it, they do inflate claimed credit