My point is that before SETI MB7, there was really no difference with the credit per hour whether you did AP's or MB's, now people are doing some stupid things out there because you can get better credit per hour doing AP's only (but a lot of rigs out there are trashing AP's because they don't understand the CPU/GPU relationship).

AP's here only go to my CPU's when I get them, but I'm happy just doing MB's because there's more of them.

So if this discrepancy was addressed then this discussion wouldn't be needed, reducing awarded AP credit would work.

Cheers.

But if your going to be competitors, wouldn't it be better
to have some ground rules to even things up?

Why on Earth would I want to do that? The idea is to crunch as much as any given individual is able.

The idea isn't to make anyone feel superior or inferior, it's to crunch as much as possible.

I don't harbor ill-will against the guys I "compete with" because they can afford GTX 690s and my last purchase was a pair of refurbished 470s. No it wasn't. I found a HD 7770 on sale. In fact, I kid them about how much money they spend and they joke about losing their houses and missing car payments and it being all so-and-so's-fault for "making" them go buy a new video card. I run cheap, sometimes very cheap, CPUs. Other people have these neat-o 6 core 5GHz overclocked Intel jobs. I don't feel the least bit jealous. I do what I can do, and truthfully, here pretty soon I won't be able to do as much.

None of those guys have ever put me down because I can't hang with them. It has never happened. In fact, I not-so-secretly think it's fun to "make" someone go get a nicer card so they can pass me. We JOKE about it. Nobody does anything they don't want to do. It's for fun.

If I started comparing what I can do (financially or space-wise) with what others can do, I'd make myself miserable. One day I'd like to try my hand at a water-cooled rig. That's expensive, but maybe one day.

In fact, the truly mega-class crunchers are all the nicest people I've ever had any dealings with at SETI@Home. They are quick to try to help others increase their RAC and optimize their machines. They are among the most profuse with compliments for the software developers.

They also tend to be the most "on top of things" when their rigs throw errors.

Heck, several of them are among the biggest contributors to the purchase of the servers, HDDs, switches, cables, and the cases that the HDDs travel-in to get back and forth from Arecibo. They've funded workstations for the lab. We all enjoy the fruits of that effort, don't we? I've never heard one of them, ever, say, "I paid for that," or "SETI owes me." Not once.

Maybe I'm missing something. If I am, I don't think it is anything important.

But I'm certainly open to the idea of a little fun with "cheapest cruncher per 'credit'" or something like that. Since so many of us live in countries where the cost of the power is more than the cost of the equipment, "the most miserly" power-wise would be sort-of fun (and maybe useful), but then that means that the contestants would have to all agree on a way to measure power. That could get complicated. Even the cost-thing is complicated. Suppose you find a machine someone's throwing away and not everyone can?

I don't know. It's just not that important. If it gets complicated and you can start arguing about the rules it isn't fun anymore.

There will always be people richer than I am, smarter than I am, better looking than I am, more committed than I am, and those who are less of all of that... that's fine.

Ah HA!! I think I see. Look, just because I compete with someone doesn't make me think I HAVE to win. If it felt that way I'd never play a game of any kind. I'm happy to congratulate someone else for winning.

Is that the difference in our philosophies?

I say "Go, man, go!" I don't say, "That's not fair!"

But here's a sad reality: A bunch of the guys who regularly made my contribution look insignificant are gone. I'm not sure why. They didn't say. They just went.

EDIT: I've just re-thought it. Yes, it would make perfect sense for there to be "classes" for a competition. How would you set one up that isn't power or cost, though? Could be fun.

My point is that before SETI MB7, there was really no difference with the credit per hour whether you did AP's or MB's, now people are doing some stupid things out there because you can get better credit per hour doing AP's only (but a lot of rigs out there are trashing AP's because they don't understand the CPU/GPU relationship).

AP's here only go to my CPU's when I get them, but I'm happy just doing MB's because there's more of them.

So if this discrepancy was addressed then this discussion wouldn't be needed, reducing awarded AP credit would work.

Cheers.

Wiggo, is that right? I'll accept your word, but I was doing APs a long time ago and gave it up because my AMD CPUs were awful at it. I actually like the AP project.

So, I was one happy camper when the OpenCL version was released and I felt I could go back to doing them. Didn't that happen about the same times as the v7 tasks being distributed? It's all a blur, mostly blurred by all the fighting about credits.

Cheers back.

@tbert

"aplause" + "aplause" + "aplause"... to your Message 1404649

---

@tbert

"aplause" + "aplause" + "aplause"... to your Message 1404649

Easy-there, buddy. Agreeing with me might make you a hated man. My position does not seem to be very popular.

@tbert

"aplause" + "aplause" + "aplause"... to your Message 1404649

Easy-there, buddy. Agreeing with me might make you a hated man. My position does not seem to be very popular.

I´m all ready a hated man exactly for the same... but who cares if what you say is the real true?

+ "aplauses"... to your real true msg.

---

Wiggo wrote:

but if MB's gained the same credit per hour that AP's do

And very likely it does, on stock (I will check few wingmen running stock when I have time to do that). Since the optimized astropulse app is now better optimized than MB (compared to stock), people running optimized application will get higher credit on astropulse for the same amount of time. If you give higher credit for MB, people running stock will get less credit per time on astropulse. So you would just move the "issue" from people running optimized to those running stock and after all those running stock should be those that don't need to care what their computer is crunching.

---

I´m all ready a hated man

Oh, that's right. I just forgot to hate you for a minute. I'll start again if it makes things easier.

Wiggo wrote:

but if MB's gained the same credit per hour that AP's do

And very likely it does, on stock (I will check few wingmen running stock when I have time to do that). Since the optimized astropulse app is now better optimized than MB (compared to stock), people running optimized application will get higher credit on astropulse for the same amount of time. If you give higher credit for MB, people running stock will get less credit per time on astropulse. So you would just move the "issue" from people running optimized to those running stock and after all those running stock should be those that don't need to care what their computer is crunching.

Just consider my nVidia 8800 GT running x41zc, Cuda 2.30 in Windows 8. The normal AstroPulse times using AP6 r1843 are around 18000 secs. Let's use the x41zc MBs that are taking around 2400 secs and gaining around 100 credits. 18000/2400 = 7.5 x 100 = 750 credits for MB tasks. AstroPulse tasks gain around 700 for the same amount of time. You actually gain more credits running MBs than APs on the 8800 GT.

Of course, mileages DO vary...

Sorry 2x post to much pain killers.

---

[quote]Wiggo wrote:

but if MB's gained the same credit per hour that AP's do

And very likely it does, on stock (I will check few wingmen running stock when I have time to do that). Since the optimized astropulse app is now better optimized than MB (compared to stock), people running optimized application will get higher credit on astropulse for the same amount of time. If you give higher credit for MB, people running stock will get less credit per time on astropulse. So you would just move the "issue" from people running optimized to those running stock and after all those running stock should be those that don't need to care what their computer is crunching.

Just consider my nVidia 8800 GT running x41zc, Cuda 2.30 in Windows 8. The normal AstroPulse times using AP6 r1843 are around 18000 secs. Let's use the x41zc MBs that are taking around 2400 secs and gaining around 100 credits. 18000/2400 = 7.5 x 100 = 750 credits for MB tasks. AstroPulse tasks gain around 700 for the same amount of time. You actually gain more credits running MBs than APs on the 8800 GT.

Of course, mileages DO vary...

As you say YMMV on one of my 2x690 for example... ussing cuda50 of course...

1 MB - running 2WU at a time (on each sub GPU - the 690 has two) - crunch in 600 sec and paids about 60 credits ... about 0.1 credit per second x 2 = 0.2 credit per sub-GPU and uses almost no CPU so your host is free to be used in other task without almost no interference, and no user complains about speed.

1 AP - running 2 WU at a time crunch in 3400 sec and receives 700 credits so it receive about 0.2 credit per sec x 2 =0.4 credit per sub-GPU. But to do that it uses all the 8 cores of the CPU, so your host must not be used to other tasks, not good for bussines and a lot of user complains about speed.

To avoid the user complains about speed i allow run 1AP + 1 MB at the same time, so i actualy receive 0.2 form the AP + 0.1 form the MB or an equivalent of 0.3 credit per second.

So AP paid almost 2 x bigger credit at least on the 690 even if it uses the same resources as the MB in the past the numbers are almost the same and you don´t need to worry about users complains because MB uses almost no CPU time.

The problem is even bigger with the I5 2xGPU´s hosts because it only have 4 cores to feed the GPU´s, so when you run AP on then you practicaly loose the host for another job, something realy bad for bussines and a lot of users complains, that not happening with MB.

My 2 Cents of the day...

---

Too bad the 690s are so expensive most people will never use one. Since the 690 user base is so small, one would wonder if they are even considered in the equations. My guess would be the credit system is focused on a much lower performance level. Sorta reminds me of the Spock scene in 'Trek #2...

Too bad the 690s are so expensive most people will never use one. Since the 690 user base is so small, one would wonder if they are even considered in the equations. My guess would be the credit system is focused on a much lower performance level. Sorta reminds me of the Spock scene in 'Trek #2...

But the same happens with the 670 or the 770 and most of the newer cards, something i could agree faster your GPU bigger your troubles and the need of the many are more important than the need of an individual if what you means about Spock scene. On other hand a single 690 could do more work than a lot of 8800 with a lot less power needed and lot less enviroment impact. So at some time some how a balance must be reached.

---

Too bad the 690s are so expensive most people will never use one. Since the 690 user base is so small, one would wonder if they are even considered in the equations. My guess would be the credit system is focused on a much lower performance level. Sorta reminds me of the Spock scene in 'Trek #2...

Of course, that's exactly backwards but we know what he meant and we won't debate his wisdom at these proceedings. (Somebody help me. I remember the burial at space scene.)

The sad part is that if we were to do a "most crunching per watt" contest, everyone with a 690 or two or three would win. Great cards. Wish I had a few.

I'd be just as happy for you to find it as I would for me to find it (Ok, so immediately I'd think it was cool that one of my computers found it and get all immature and stuff, but it isn't like I'd be able to "take credit" for the discovery).

It is a good thing you feel that way; a little bird told me that the most promising WUs from the most promising areas of the sky go to project insiders. Perhaps they will thank all of the "little people" when they accept the Nobel Prize.


When I first started it was all about ET, now it is more about real world benchmarking so a mark should be a mark should be a mark.

---

Perhaps they will thank all of the "little people" when they accept the Nobel Prize.

You know, I'm not even holding my breath for that. If we, any of us, ever gets a mention it will go something like this:

"...and we would like to thank the millions of SETI@Home volunteers around the globe who donated the spare cycles of their computers to help in the search."

Nevermind that there aren't millions, or that some people donate thousands of dollars per year in electricity quite beyond "spare cycles", or that some volunteers have also given much (like donating servers) and made great sacrifices of time and money in the belief that a discovery would change humanity.

But that isn't a complaint, it's just the way things are. When the Red Cross saves a life they don't trace the blood donation back and single-out the donor. We are all part of something much larger than any of us and if we have visions of future personal glory, then we are deluded.

When I first started it was all about ET, now it is more about real world benchmarking so a mark should be a mark should be a mark.

Who could possibly argue with that? (although someone will)

I've long suspected that the preliminary screening of the data prior to being distributed had an effect on its distribution, but I thought that might just be paranoia. Well, I thought that until we learned the Kepler data was being looked-at in-house before being turned-over to us. But again; that's fine. I'm not thinking in terms of personal glory. I would like to feel like I'm not wasting my time, though.

I probably shouldn't admit this, but I suspect that if a signal is found we will never know. The best clue we will have is when the project declares failure and shuts-down. At that point I will suspect we found something months or years earlier. I think that's terrible, but our "overlords" the world over seem to think themselves ordained to "guide humanity." You can see it in almost every act of every government, religion, and philosophy, whether it's tribal or "democratic" or despotic. Some megalomaniac or committee of megalomaniacs thinks they know better than everyone else.

My gripe (if I have one, at this point) with the credit system is that I would like for it to more-or-less accurately reflect something. I don't even know what that "something" might be but, something. That's not simple, or easy, and may not be doable. I've come to accept that. Computer capacities these days are almost all estimated and it doesn't seem there is agreement on how to count FLOPS with the advent of HPC, clusters, and the like.

I've decided not to worry about it anymore.

The whole preference for AP and the higher credit is a self-limiting problem. We don't get new AP until we've crunched-through a lot of MB. The solution is defined by the problem, it is already implemented, and all of the MB and all of the AP gets crunched and everyone has the same chance proportionally at the higher 'credits' of the AP as everyone else. Problem solved. It never really was a problem unless you run multiple projects. Then the problem becomes one of perception. "Gee, I did the same 'work' for _____ as I did for SETI@Home, but the pie chart shows I only did 20% as much work for SETI@Home as for the other project."

I've got all sorts of things to worry-about in my life and worrying about that can't be a high priority. In fact, it's worse than a waste of my limited time on Earth because it is unpleasant and completely out of my control.

A lot of fellow-crunchers have left us. I had no idea that my contribution had risen to be "one of the top" and that's sad because my contribution did not used to be "one of the top" and my contribution hasn't changed. A lot of people who did a lot of this work must have left.

If I have to choose something to worry-about concerning the credit system, that would be it, but they may have left because of the weather, too. Resolved: I do what I can and worry less about all the rest of it.

Great bunch of people here, though. Even when it gets childish and testy it's a good group on-the-whole. That's a good thing to focus-on.

There are great people here. And they share their knowledge with no hesitation.

Im guilty of getting on my high horse sometimes and not seeing the other guys point. So To any poster who I have offended by my remarks I appologize.

So until I can get that Seti@Home toaster I still think credits are meaningless:)

---

[/quote]

Old James

Don't know why people say that they are looking for ET. I am searching for Spock. He has a lot better technology! LOL

So until I can get that Seti@Home toaster I still think credits are meaningless:)

Well put :-)

(Where's my toaster???)

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Don't know why people say that they are looking for ET. I am searching for Spock. He has a lot better technology! LOL

I'm old-school; I write in cursive and use semicolons. I also forgot what I was going to type; never mind.

---