Thanks, HAL, that was a great explanation, and it makes sense to me now. Wonder what Jason has found to hopefully make it more stable and such? I hope he can work his magic on it and it will fix it so everyone will be reasonably happy with how credit is granted again.

For the time being, having accumulated a lot of research and experimentation there over a number of years, am focussed on the application side of things (at least preparing infrastructure for the next generation).

Probably once we have something a bit more flexible on the science app side (some supporting tools, semi-automated configuration, easier cross platform builds etc), I can probably turn back to where I was up to with 'that stuff' last year.

General gist is likely going to be a generic client side estimate fix (Initial and onwards), with the main purpose of illustrating how the main concepts are 'fine', how estimate localisation drives everything (scheduling and control of tasks and projects), how it is normally done in engineering, with some means of comparison against current implementation and CreditNew's own stated intent.

As the same estimate localisations are mirrored server side (with minor variation) per host-app-version-planclass on validation, a tuned client estimate mechanism forms a reasonable template for synthesis of the server side, and its mass audience statistics (used to drive credit).

The problems with fixing the quite noticeable, classic, and fixable instabilities are several.

First, a possible misnomer: as soon as anyone mentions credit, RAC, or CreditNew, there tends to be an outcry of 'Don't care we're here for the science!' (or similar) which tends to cloud that it's about estimates and control (which are clearly problematic). Maybe a better name for it, if a mouthful, would be 'the estimate tracking portions of the scheduler', clearly more important sounding , and credit just a byproduct-symptom.

second, as I'm finding with the state of my own MB Cuda App, the coding styles don't really lend themselves well to flexibility/maintenance/configurability that well. Various reasons ranging from reasonable age of the designs+code through to technologies perhaps taking turns beyond the original scope (pushing design limits).

So in the scheme of things plenty of work to do (lol), and those come as second fiddle to getting the app optimisations in place in a cross-platform and configurable way. At least as I can spend a little time with Cuda8 and Petri's optimisations, some of the more modern & flexible methods should start to work their way into some of the more stale infrastructure areas. In the end the same accumulated techniques will begin to roll into other areas (like some tools and custom client code), and hopefully begin to make the bigger picture a bit less tangled.

---

"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.

Appreciate all your work Jason!!

DA isn't interested to let Jason, or anyone else fix it. If DA says it isn't broken, then nothing will be fixed.

We just have to live with ever decreasing credit, or just stop participating in the project.

Nothing as beautifully complicated as CreditNew, can ever be wrong. Only a genius could come up with something so perfect:

Well not completely true. There are fixes applied to Boinc from time to time. Anyone can fork the codebase on Github, and customise the code (which projects do), and there are contributors patching bugs and vulnerabilities on a regular basis.

It's true about the 'perceived' complexity of CreditNew though, but that complexity is an illusion. The only thing really complicating CreditNew [aside from some basic omissions/instabilities] is it has little to nothing to do with Credit.

---

"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.

Appreciate all your work Jason!!

+1000!

---

Bloody hell Jason is there anything your not doing? :) Crack open a tinny mate it's the weekend :)

As to DA his problem is dismissing anything presented to him by people who are not coders. And even if you are a coder it's likely he'll dismiss it.

<sarcasm>Which is probably why there's been a massive influx of people wanting to continue to develop boinc now its gone to a new governance modal, which is the same as it was before but with more words.</sarcasm>

---

Member of the People Encouraging Niceness In Society club.

Bloody hell Jason is there anything your not doing? :) Crack open a tinny mate it's the weekend :)

As to DA his problem is dismissing anything presented to him by people who are not coders. And even if you are a coder it's likely he'll dismiss it.

<sarcasm>Which is probably why there's been a massive influx of people wanting to continue to develop boinc now its gone to a new governance modal, which is the same as it was before but with more words.</sarcasm>

As someone that has been a software test engineer for the past 15 years. I can tell you it is up to the person claiming something is broken to prove it is broken. Saying "I think this is broken. Fix it!" doesn't really work on any developer. If you can sufficiently explain an issue it is likely to be resolved.
For example. I had an issue running BOINC on some clients where I was using virtual box to run actual VMs. I explained my issue and new we have the option to tell BOINC to ignore virtual box.

Due to the expected long term averaging of CreditNew it isn't very practical to do a black box test trying to pinpoint issues.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

As to DA his problem is dismissing anything presented to him by people who are not coders. And even if you are a coder it's likely he'll dismiss it.

I'm not a coder, but a lot of what I brought forward to David is included in BOINC. Perhaps it's the way you -or others- asked it that he ignored you. Friendly and requesting brings you more than accusing and demanding.

And in all these cases, patience is a virtue as well. I remember I got an email back from him 6 months after I sent one in, just to show that some things may take time.

---

Ah well, it's quite clear now. With the introduction of GBT work, my RAC is falling like a stone on MB work. Since AP is just a "once per week" thing nowadays, my RAC doesn't have time to recover.

Thank you endlessly much CreditScrew.

+1 but Spose we should of expected it to happen ;-)

---

Life is what you make of it :-)

When i'm good i'm very good , but when i'm bad i'm shi#eloads better ;-) In't I " buttercups " p.m.s.l at authoritie !!;-)

Since AP is just a "once per week" thing nowadays

I reckon you meant to say "one per week". ;-)

Sleepy

My main beef about CreditScrew started way back when V8 was introduced. Prior than that my machine was in the top 50 computers, now it seems that I'm stuck at 200+. Some of this can be blamed on the lack of a steady diet of AP work units/tasks, ah the good olde days, but most of it is because of the steady reduction in credit. I don't know where I'll end up now that we have started work on Greenbank. I don't know how long credit will get normalized with the new data.

---

I don't buy computers, I build them!!

IMO that is not true because we all have suffered the same credit depreciation.

---

IMO that is not true because we all have suffered the same credit depreciation.

Not quite true, or at least not true in general, in my opinion.

If you are comparing to people that also before the recent and less recent changes were mostly working on APs, then the problem is the same and all should suffer the same. No big changes here.

But I have noticed, right through the way they were not suffering so much, that many people were crunching mostly or only MB WUs. For them not really much has changed each step APs have got more difficult to get.

To recap: people who have been crunching mostly MBs all along have not suffered that much.

And among them there are even many heavy crunchers, who, as such, you would otherwise think they would try to get as many APs as I am (was) doing. And they were not, flying smoothly over MBs.

Just my 2 cents.

Sleepy

Mind you, the whitepaper on CreditNew is ok. it's just the implementation that is lacking ;)

I never understood why people care about some arbitrary computer numbers anyway. Must be another case of 'mine is bigger than yours'.
What I care about is WHEN you do something like that you should do it mathematically sound and not end up with a bunch of spaghettis that would do well in a chaos engine.

And yes, Jord, David will listen if you present stuff the right way.

---

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

The big problem for me (as a developer) is that these 'arbitrary' numbers become a locus of effort, when most (not all) agree the effort should be on the science, While certainly the science is my motivation and inspiration (in many ways other than proving or disproving ETI existence), these distractions do present themselves on a competitive pedestal.

In its purest interpretation, CreditNew works well and presents a more accessible challenge to the non-scientifically inclined. Where it falls short IMO, is in the rejection of well established engineering and scientific principles in its implementation, in favour of a naive renaming of stuff [That already have proper names....].

---

"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.

I never understood why people care about some arbitrary computer numbers anyway. Must be another case of 'mine is bigger than yours'.

Yes.
And that is why it is of importance.
Most people like some sort of recognition for the things that they do. Credit is the way that people are rewarded for their participation in this project. For those that feel the need, it allows them to compare themselves with others. For most it just lets them know how much work they have done.

For those reasons, it is important that it is a "as good as possible" indicator of work done. Unfortunately, as it stands, Credit New doesn't do that.

---

Grant
Darwin NT

Exactly. Having the biggest penis is perceived as important, so we must strive to do so. [Edit: FOR SCIENCE!!!!]

---

"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.

Exactly. Having the biggest penis is perceived as important, so we must strive to do so. [Edit: FOR SCIENCE!!!!]

Wait... is that not what the P in FLOPs represents?

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

DA....Dr. Anderson.
Is a very intelligent scientist.
This is HIS project.
I usually refer to Eric, as he turns the nuts and bolts that keep the project on track.
But, it is Dr. Anderson's brainchild.

I think I have been amiss for not admitting this for so long.
Sorry, DA.

---

"Time is simply the mechanism that keeps everything from happening all at once."

Exactly. Having the biggest penis is perceived as important, so we must strive to do so. [Edit: FOR SCIENCE!!!!]

And there comes Bitcoin Utopia.... Well, exclude the "FOR SCIENCE"...

Exactly. Having the biggest penis is perceived as important, so we must strive to do so. [Edit: FOR SCIENCE!!!!]

And there comes Bitcoin Utopia.... Well, exclude the "FOR SCIENCE"...

Is bitcoin still a thing ? I thought they ran out of blockchain space ?

---

"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.