Valuing our work has always been a bit sticky for these forums.
* Since we provide value even if we find nothing, we can't count observations to credit our efforts.
* Counting flops seems fairer, except it doesn't reward efficiency. That is, if my version of the science app does work in fewer flops than someone elses, I should get the same credit as he does. But that doesn't seem to be a problem when folks work a single project with a single science app. In seti, with several science apps available for the same work, then (I think) the less efficient app gets fewer credits. Yet,with various fudge factors introduced for seti which I don't remember how they work, perhaps this is no longer true. Here work is counting ops and not measuring time or rate.
* But Time should have value as well. That is, how quickly we turn around a wu that has X floating point ops should also be important. Perhaps this concept is ignored due to the catatonic state of this project. But I think it would be an oversight.
* I'm sure there are other perspectives I've overlooked.

Nevertheless, it is my opinion that boinc/seti has fiddled with the credit mechanism for various reasons over the years. So I was looking for some reason to believe the credit process was giving equivalent results now as before.

Perhaps if we ask nicely Seti would run a test set of wu's that were analyzed in the early days of boinc (6 years ago?) and see if people get the same credit (at least statistically); and the same results!

Counting flops seems fairer, except it doesn't reward efficiency. That is, if my version of the science app does work in fewer flops than someone elses, I should get the same credit as he does. But that doesn't seem to be a problem when folks work a single project with a single science app. In seti, with several science apps available for the same work, then (I think) the less efficient app gets fewer credits. Yet,with various fudge factors introduced for seti which I don't remember how they work, perhaps this is no longer true. Here work is counting ops and not measuring time or rate.

Before the "pending credit" feature was turned off, it seemed to me that whenever my ancient Mac G4s were paired with a newer/faster machine, whether running stock or Optimized apps, cpu or gpu, and I finished first, I usually got less (and never more) credit than was claimed for my machine. That leads me to believe that the credit formula IS weighted in favor of the lower flops/higher speed/newer machine. That makes sense, and I have no problem with it.

---

Donald
Infernal Optimist / Submariner, retired

snipped....
Perhaps if we ask nicely Seti would run a test set of wu's that were analyzed in the early days of boinc (6 years ago?) and see if people get the same credit (at least statistically); and the same results!

That wouldn't work as the app has changed at least twice since then, as shown in Message 1093269 by Raistmer.
And you couldn't do the reverse either and use the then Seti V4.18 single beam app on present day multi-beam data.

snipped....
Perhaps if we ask nicely Seti would run a test set of wu's that were analyzed in the early days of boinc (6 years ago?) and see if people get the same credit (at least statistically); and the same results!

That wouldn't work as the app has changed at least twice since then, as shown in Message 1093269 by Raistmer.
And you couldn't do the reverse either and use the then Seti V4.18 single beam app on present day multi-beam data.

Isn't the difference the way in which the raw data is split? Use today's apps but yesterday's raw data, that is.

snipped....
Perhaps if we ask nicely Seti would run a test set of wu's that were analyzed in the early days of boinc (6 years ago?) and see if people get the same credit (at least statistically); and the same results!

That wouldn't work as the app has changed at least twice since then, as shown in Message 1093269 by Raistmer.
And you couldn't do the reverse either and use the then Seti V4.18 single beam app on present day multi-beam data.

Isn't the difference the way in which the raw data is split? Use today's apps but yesterday's raw data, that is.

Not really, the actual data splitting hasn't changed though the preprocessing to replace sections which are polluted with radar signals is new. The WU header which tells applications what to do has changed, and the applications interpret those header parameters differently too.

The project has reference_work_unit.sah in the repository (recorded Mar. 3 2000) in which the data section has not changed at all and only necessary changes have been made in the header. So it has the old beam width of 0.083, chirp limits of 20 and 50 like other tasks from before S@H Enhanced, and the chirp resolution is 0.333 rather than the most recent 0.1665 which doubled the amount of work. If that WU were sent to S@H Enhanced applications here it would take about 1.5x a VHAR shorty's time to crunch so be given ~50 credits on average. With the old pre-Enhanced application it might have gotten about 25 credits and run in about half the time, though my recollection of how an AR 0.775 WU turned out then is foggy.

                                                                 Joe

Look how worried I am about the credits. So long as the numbers go up, I know I am donating process time that is actually being used. Beyond that I don't mind what the numbers are. If this is some kind of competition, I am just an also ran.

I have spare capacity. I donate it to Seti@home. They don't pay me, and I don't want paying either. They give me credits by way of recognition. If I were after credits, I could more than double them just by not hobbling the machine. I could run a 64bit OS instead of the one I do. Neither project gets much donated time after I get home and release the shackles.

I have my machine hobbled to 3/8 speed during the day while it is watching the security setup. I have Boinc use 1/2 the CPUs, and have CPDN get equal CPU time, because Seti sometimes doesn't have the CPU work for it even in its hobbled state.

As well as reducing power usage, I need to reduce the capacity the machine has because otherwise the image analysis sees the reflection of the beating of a butterflies wings in china as a change in image and starts image capture. As it is, with Seti using the GPU and Seti/CPDN half the hobbled CPU, it takes an absent cloud revealing the sun before the system screams "invader". (English weather helps keep it dim normally.)

With CPDN's problems recently and the increased available work from Seti, the two have reversed their credit rating for me.
Len

---

I think I am. Therefore I am. I think.

the real reason Nez stopped running Seti,
he found out the free toaster was only a two slot model
instead of a four slot model.

---

the real reason Nez stopped running Seti,
he found out the free toaster was only a two slot model
instead of a four slot model.

ROFL, I think more appropriately he went to collect his toaster and they realized his raffle ticket wasn't a valid one, and was run out the door.

---

Traveling through space at ~67,000mph!

back crunching seti :) when i first started there was a few server problems so i drifted elsewhere (my computer kept cooling down (is that a good thing?) got my little coloured squares from prime grid (no toaster !! :( ) looking foreward to my fried egg in the morning cooked on my cruncher.cheers chris skilldude and johnny thanks

back crunching seti :) when i first started there was a few server problems so i drifted elsewhere (my computer kept cooling down (is that a good thing?) got my little coloured squares from prime grid (no toaster !! :( ) looking foreward to my fried egg in the morning cooked on my cruncher.cheers chris skilldude and johnny thanks

Welcome back!

The kitties dont't care about toasters....
But in the winter, they do luv the heat the crunchers throw off.

Meow meow.

---

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

thanks mark, i hope everything turns out ok for you and your loved ones


john3760

thanks mark, i hope everything turns out ok for you and your loved ones


john3760

Thanks, friend.

---

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