VJ blog update: Bigadv, SMP & you.

Right thing to do, right way to do it. So far, it seems this news has been as well-received as can be hoped for on most of the teams I spy on.

To continue my thought from Nathan_P's earlier thread... In terms of the benchmark - can the client-calculated Performance Fraction work for this? This would account for both per-frame performance and time spent running the client since not all machines fold 100% 24/7. Since the client uses UTC, it seems like the PF would be a PITA to manipulate and thus pretty accurate metric on balance.
 
Kendrak said:
The way we are trying to move is to have a mini bench built into the client. It would give you a score on the speed of the system (not just the thread count) and then assign only WU that the system can handle. This will let high priority WU go to systems with a high score.

The DAB is trying to move us away from thread count as a measure because it is only loosely correlated with system speed.

This will take time to put in place, but it is where we want the system to go.
Click to expand...

I'm sure this plan will be a steaming pile of FAIL, just like v7. Any benchmark will probably be hot on the list of things to implement in v8 or v9. They already have ways of benching the client, have had them for 6 years that I know of, and yet they have never bothered to figure out how to use the info. What makes you think that they will miraculously figure it out now. My suggestion works as a band-aid until they do. Though I doubt they ever will. They are scientist who suck at programming and continue to refuse offers by programers who suck at science, but could fix their issues in hours.
 
Bigadv = beta. While disappointed my 970s won't get the points it did, I went into this with my eyes wide open.
 
I went into this with my eyes wide open
Click to expand...
Many of our fellow Folders have not been with the project very long. They don't realize in the over 10 years F@H has been running, there have been many substantive changes. Every time one of those change waves hit, those without a long term perspective believe it to be not only the first wave, but a tidal wave. My respect to all who fold, long or short term, but this enterprise really does require some perspective to keep sane. :cool:
 
Linden said:
Many of our fellow Folders have not been with the project very long. They don't realize in the over 10 years F@H has been running, there have been many substantive changes. Every time one of those change waves hit, those without a long term perspective believe it to be not only the first wave, but a tidal wave. My respect to all who fold, long or short term, but this enterprise really does require some perspective to keep sane. :cool:
Click to expand...

Well said. Those freaking out are losing it because their crack is a slightly different tint while delivering the same end goal...
 
This doesn't bother me at all. I'm happy to fold whatever Pande Group feels my hardware is best suited for. My days of freaking out about PPD changes are long over.
 
Thanks Kendrak, an answer is all it takes to shut me up. I have no idea how gnar the math is to make big GPU happen...sounds like pretty gnar.
 
this thread is about Bigadv, SMP, etc

lets keep it that way

also any other reference to EVGA will be deleted immediately
 
In terms of what the goal of the bigadv program is, I think this makes sense. It's meant for the most powerful machines (i.e. multi-socket for the most part), and should reasonably change as HW continues to improve. Also it's always said that it's subject to change because it is still experimental. As long as there is some sort of balance made eventually, I don't mind. Hell, I don't care that much anyways, I just run whatever I get unless it horribly under-utilizes my hardware (e.g. if I forget to set -smp on an L5640, I'm turning it on).
 
Kendrak, do you think you could ask Vijay for the statistical bigadv WU return data that lead them to revise the requirements?

No biggie if it's too much to ask, I'd find them interesting...
 
tear said:
Kendrak, do you think you could ask Vijay for the statistical bigadv WU return data that lead them to revise the requirements?

No biggie if it's too much to ask, I'd find them interesting...
Click to expand...

Like average return time?

I also think "everyone" doing -bigadv and fewer people doing SMP has something to do with it was well.
 
It just means they get less cpu time off me. Since with bigave I would never stop it because it would mean losing too many points, now I turn it on and off when I want some more heat (might as well do some good instead of just burning fuel). This is the second time they have moved the goal post within a year. I brought a 2600k @ 4.8ghz specifically for the project, now it's only doing ~20k ppd, whereas it used to do the best part of 50).:(

Since they can pick and choose who completes the units, luckily I guess I'm in the minority.
 
I've been chewing Bigadv Wus on 4 core/4 thread intel chips for quite some time now, I started whith Core2Quads with the help of grandpa. then I moved to current champ on bang for buck i5 2500k, how I did this? well many guides out there on how to do it, wont go into details.

so if I was able to "add" 4 extra threads to an OC i5 and finish Bigadv, I don't see how hard can it be to "add" 4 extra threads in a 980x

I can see how Intels 6 core processors will be now budget bigadv folders for many
 
budgetfolder said:
I've been chewing Bigadv Wus on 4 core/4 thread intel chips for quite some time now, I started whith Core2Quads with the help of grandpa. then I moved to current champ on bang for buck i5 2500k, how I did this? well many guides out there on how to do it, wont go into details.

so if I was able to "add" 4 extra threads to an OC i5 and finish Bigadv, I don't see how hard can it be to "add" 4 extra threads in a 980x

I can see how Intels 6 core processors will be now budget bigadv folders for many
Click to expand...
What you are describing is against project rules and not welcome on this forum. Additionally, this cheating of the system results in WU's not meeting deadlines, not being available for truly capable hardware, and generally just defeats the whole point of quick return. Spoofing came up as one of the specific reasons driving the change in bigadv requirements and you are doing everyone a disservice by engaging in it.
 
R-Type said:
What you are describing is against project rules and not welcome on this forum.
Click to expand...

its not like I am hacking the client.. :rolleyes: and science is being done and finished on time..

I'll Just get me a cheap 6 core intel chip and OC it and will be bringing them Bigadv Wus while some of you do just SMP..;)
 
R-Type said:
What you are describing is against project rules and not welcome on this forum. Additionally, this cheating of the system results in WU's not meeting deadlines, not being available for truly capable hardware, and generally just defeats the whole point of quick return. Spoofing came up as one of the specific reasons driving the change in bigadv requirements and you are doing everyone a disservice by engaging in it.
Click to expand...

+1. This is an ethical team.
 
You're welcome to write it, but I would expect it to be deleted within the day.
Regardless of whether it works, it's against Stanfords policy for the project, and as such this team won't condone it.
 
budgetfolder said:
its Ok with me, I'll get more points than you doing SMP while on the same hardware, while doing science, believe me this intel 6 processors are very powerful

tell you what, I'll write a "How To" on how to do Bigadv on intel 6 core(amd can't do it) and I'll post it her and you can all thank me latter..;)
Click to expand...


don't push it new guy...everyone here know's exactly how to do it but they don't. not everyone thinks like you

most of our "how to" guides are screened by the mods, i am guessing your new "how to" guides & related posts like this will get u banned in no time.

anyways welcome to the forum!!
 
Last edited:
budgetfolder said:
its Ok with me, I'll get more points than you doing SMP while on the same hardware, while doing science, believe me this intel 6 processors are very powerful

tell you what, I'll write a "How To" on how to do Bigadv on intel 6 core(amd can't do it) and I'll post it her and you can all thank me latter..;)
Click to expand...

We are fully aware of how powerful a 6c/12t intel CPU is, we have got more than a few of them in use here;)

However PG have stated that from 16/1/12 they will no longer be able to fold -bigadv and as a team we follow the rules.

Besides this is the [H]ard [H]orde, cheating is for wimps. We like to do things the old fashioned way. When in doubt add more CPU's - that's [H]ard.:cool::D
 
What Nathan said lol

and speaking of How To guides, I'm guessing ours are kind of old, eh? Might be something I look into when I get back to the states and fold again, I'll have down time and I'll be coming in from a 13 month absence so I should have a pretty good "noob" point of view lol
 
Last edited:
Geez.

Economy 101. The project created point-based rewards hence can't blame the donors who
maximize points production.
Perhaps points aren't currency of the realm for some but one can't take away else's right
to be driven solely by points.

The "science", as they call it, and points are both equal motivations. Don't judge. Thank you.

If the project wished the science to be the only motivation, point-based rewards would not
have been conceived. Yet, here we are.


Cheers,
tear
 
Kendrak said:
Like average return time?

I also think "everyone" doing -bigadv and fewer people doing SMP has something to do with it was well.
Click to expand...

Like, for instance, time-to-return/number of returns histograms. Average doesn't quite
reflect the nature of the environment...
 
The bigadv change, whilst painful for many, seems to have been handled better than the last one. A bit of consultation, consistency with previous statements/direction and a notice period makes a world of difference to me.

PG may not ever acknowledge mistakes but at least they seem to be learning from them. There's still a lot that could be done better but nobody's perfect. If the last change was handled this way and if it wasn't for the free for all that is ff.org, I'd probably still be folding.
 
phoenicis said:
The bigadv change, whilst painful for many, seems to have been handled better than the last one. A bit of consultation, consistency with previous statements/direction and a notice period makes a world of difference to me.

PG may not ever acknowledge mistakes but at least they seem to be learning from them. There's still a lot that could be done better but nobody's perfect. If the last change was handled this way and if it wasn't for the free for all that is ff.org, I'd probably still be folding.
Click to expand...

You can always fold here and never step foot into FF. Many of us never leave the [H] for our folding info.

The DAB is working on getting things back to a fair structure. This is a first step.
 
Thanks Kendrak, I do still think about it at times but I'm having way too much fun with the Commandos right now. DCing is again a relaxing pursuit :)

You seem to be doing a sterling job representing [H]'s interests on the DAB :cool: It would be nice if we could all view the posts so that PG's responses to questions asked could be read directly. Just my 2c worth but it'd go a long way to removing the transparency gripe. On the flip side I guess it may limit the DAB member's questions if there was an audience.
 
phoenicis said:
It would be nice if we could all view the posts so that PG's responses to questions asked could be read directly. Just my 2c worth but it'd go a long way to removing the transparency gripe. On the flip side I guess it may limit the DAB member's questions if there was an audience.
Click to expand...

This conversation doesn't just happen out here.
 
I just picked up a 6900 on my FX 8120 box running Ubuntu. That's the first bigadv this CPU has ever seen. It's all grown up now :) For the next 3 months :(
 
Kendrak said:
This conversation doesn't just happen out here.
Click to expand...

I am glad that, unlike beta team, this is information that is supposed to be shared not kept secret. Sure it isn't perfect but it is WAY better than it was. Took some pain to get there but progress is good.
 
You must log in or register to reply here.
Back
Top