RAC will come back when Raistmer & company finish optimizing the new parts of the MB app, AND get used to the new build tools they were forced to adopt for legal reasons.

Patience.

and

All hail the volunteer coders.

No you are wrong... I belive you miss the Eric post:

Every credit method we've ever used, with the exception of FLOP counting punished optimization, especially optimization of the stock apps.

and clearely any optimization will not make reach the balance AP vs MB again...

---

Again, return to the real world, humans are highly competitive, is part of the human nature you can´t take that from us, credits is the only way we have to measure our producton, we humans need an incentive (could be even meaningless credits) specialy on a long term project like SETI.

...

Yes, credit could have no real value for science and we could be here to try to find an E.T. but we never could forget, first we are humans, we need some kind of reward for what we doing, even a "zero value credit"!, that´s the way our brains works.

I'm in the real world. Last I checked I'm still human. I'm in the minority crowd who is not competitive what-so-ever. When I see overly competitive people trying to compete with me, I let them have whatever objective they want just to get them off my case.

I guess my brain works differently than most. I would be perfectly OK if the project got rid of credits altogether. Sure, a lot of people would be upset and possibly leave.

But my incentive is simply to find an answer to the question: "Are we alone?" I don't need credits to remember the science.

All volunteers are not scientific, William, and without the volunteers and their taste for credits and competition, no SETI@home project...
Don't forget that please...
Welcome to the real world ;)

I don't think your assertion is accurate. I'm sure many who put the science first would remain. You can't state definitively that without credits SETI@home wouldn't exist until we actually test the hypothesis. I'd be willing to try the test, but I doubt any one of the top crunchers would be happy about it.

How many volunteers are complaining about credits 20 or 30 ?
Go figure.

How may realy read the forums? 20 or 30?

Just one is enoght, if we loose his contribution could be the one who is destiny to find the signal...

Pure nonsense. "Destiny" is not a concept found in the natural sciences. To suggest that someone specific is meant to find a signal is ludicrous to suggest. I mean, what if it isn't even someone interested in SETI@home? Will that mean we never find the signal?

No. A vocal minority is not necessarily something that needs to be addressed.

No you are wrong... I belive you miss the Eric post:

Every credit method we've ever used, with the exception of FLOP counting punished optimization, especially optimization of the stock apps.

and clearely any optimization will not make reach the balance AP vs MB again...

Eric wrote that in reply to me. It's an obscure point that doesn't have much bearing on the current hysteria.

Creditnew decides how much credit an app gets based on the elapsed run times the stock CPU version of the app. Basically it's elapsed time x FLOPS rating of the device. All stock cpu versions of all apps including AP and MB should be producing the same number of credits/sec. If anyone has actual numbers to show otherwise that would be interesting.

All versions including gpu and opt get that amount of credit. The gpu and opt versions do it faster and so get more credit/sec.

AP uses CreditNew and nobody is complaining about that app. The reason being that the opt versions are so much faster than stock cpu that everyone piles up credits at a rate that makes them happy.

The current MB opt app is not crazy factors faster than stock cpu, because of reasons. Thus the great panic. When the guys come out with something better, peace and serenity will once again descend upon the land.

+11

---

Grade : Prestige Golf

I would remind folks not to attempt to hijack this thread, or to flame each other. There are a number of posts here that getting very close to both, so take note.

---

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

I totaly agree with rob, we are here just to ask for balance the MB vs AP "credit ratio" nothing related to more credit or anything else.

---

Here's a little detail about why it's impossible to have Astropulse return credit at the same rate as SETI@home. I pulled all of the results currently in the database and figured out for each app version the rate at which credit is being granted (i.e. sum(granted_credit)/sum(elapsed_time)).

First the CPU versions since it's easy to do cross comparison. Here is the most recent version of Astropulse for each platform, against the most recent version of SETI@home.

app             platform              version  plan_class             credit/sec
--------------------------------------------------------------------------------
astropulse_v6   i686-pc-linux-gnu       601                              0.0034
setiathome_v7   i686-pc-linux-gnu       701                              0.0047
                                                                           -28%

astropulse_v6   x86_64-pc-linux-gnu     603                              0.0058
setiathome_v7   x86_64-pc-linux-gnu     701                              0.0068
                                                                           -15% 

astropulse_v6   windows_intelx86        601                              0.0046
setiathome_v7   windows_intelx86        700                              0.0058
                                                                           -21%

astropulse_v6   i686-apple-darwin       602                              0.0059
setiathome_v7   i686-apple-darwin       700                              0.0059
                                                                            tie

astropulse_v6   powerpc-apple-darwin    602                              0.0020
setiathome_v7   powerpc-apple-darwin    700                              0.0014
                                                                           +43%
--------------------------------------------------------------------------------

So the windows and linux version grant more credit per second to SETI@home than they do for Astropulse. The only one that grant significantly more to astropulse is the powerpc version. I'm not sure why that would be.

Now the GPU versions. It's more difficult to directly compare because the Astropulse app is OpenCL on NVIDIA rather than the 5 flavors of CUDA. So I'll just group by manufacturer and OS and then sort by decreasing credit/sec.

app             platform              version  plan_class             credit/sec
--------------------------------------------------------------------------------
astropulse_v6   windows_intelx86        606     opencl_ati_100           0.0687
astropulse_v6   windows_intelx86        606     ati_opencl_100           0.0683
setiathome_v7   windows_intelx86        703     opencl_ati5_cat132       0.0477
setiathome_v7   windows_intelx86        703     opencl_ati_cat132        0.0399
setiathome_v7   windows_intelx86        703     opencl_ati5_sah          0.0244
setiathome_v7   windows_intelx86        703     opencl_ati_sah           0.0231
astropulse_v6   windows_intelx86        608     cal_ati                  0.0108

astropulse_v6   x86_64-pc-linux-gnu     607     opencl_ati_100           0.0835
setiathome_v7   x86_64-pc-linux-gnu     704     opencl_ati_linux         0.0405
setiathome_v7   x86_64-pc-linux-gnu     704     opencl_ati_linux_cat132  0.0312
setiathome_v7   x86_64-pc-linux-gnu     704     opencl_ati5_linux        0.0273
setiathome_v7   x86_64-pc-linux-gnu     704     opencl_ati5_linux_cat132 0.0200

setiathome_v7   windows_intelx86        700     cuda50                   0.0791
setiathome_v7   windows_intelx86        700     cuda42                   0.0637
astropulse_v6   windows_intelx86        604     opencl_nvidia_100        0.0551
astropulse_v6   windows_intelx86        604     cuda_opencl_100          0.0413
setiathome_v7   windows_intelx86        700     cuda32                   0.0335
setiathome_v7   windows_intelx86        700     cuda23                   0.0305
setiathome_v7   windows_intelx86        700     cuda22                   0.0084

astropulse_v6   x86_64-pc-linux-gnu     607     opencl_nvidia_100        0.0908
astropulse_v6   x86_64-pc-linux-gnu     607     cuda_opencl_100          0.0478
--------------------------------------------------------------------------------

The conclusions I draw from this is that ATI card are much better at Astropulse than they are at SETI@home. The primary reason is probably that the GPU routines in SETI@home were written in CUDA and ported to OpenCL. In Astropulse the GPU routines were written directly in OpenCL. And I think that's where the perceived discrepancy between Astropulse and SETI@home credit rates comes from.

---

@SETIEric@qoto.org (Mastodon)

All stock cpu versions of all apps including AP and MB should be producing the same number of credits/sec. If anyone has actual numbers to show otherwise that would be interesting.

All versions including gpu and opt get that amount of credit. The gpu and opt versions do it faster and so get more credit/sec.

AP uses CreditNew and nobody is complaining about that app. The reason being that the opt versions are so much faster than stock cpu that everyone piles up credits at a rate that makes them happy.

The current MB opt app is not crazy factors faster than stock cpu, because of reasons. Thus the great panic. When the guys come out with something better, peace and serenity will once again descend upon the land.

So now it's at least two of us, who understand why we (users of optimized apps) get more credit for astropulse tasks...

---

The conclusions I draw from this is that ATI card are much better at Astropulse than they are at SETI@home. The primary reason is probably that the GPU routines in SETI@home were written in CUDA and ported to OpenCL. In Astropulse the GPU routines were written directly in OpenCL. And I think that's where the perceived discrepancy between Astropulse and SETI@home credit rates comes from.

Thanks Eric.

This is was i thought.

---

With each crime and every kindness we birth our future.

Here's a little detail about why it's impossible to have Astropulse return credit at the same rate as SETI@home.

A question remains why with V6 that works? a simple coincidence?

If that not what exactly creditnew is expecting to do? How it could create a balance interprojects if it even can´t create the balance within a single project?
Another paradox?

Anyway thanks for your return, explanations and time. That´s is highly apreciated.

---

+12

---

A question remains why with V6 that works? a simple coincidence?

I wish I could tell you. There aren't enough V6 results in the database to get accurate statistics. And any that are there are from slow machines, or machines that spend most of their time turned off.

---

@SETIEric@qoto.org (Mastodon)

Edit- Thomas you have a good point, Maybe its time to just end opt apps.

Just delete your app_info &/or app_config files and let us know how you go in 2 months time. ;-)

Cheers.

You dont see me crying over credit do you? Im saying with all the crying about it Dr.A just might say no more opt apps. From now on we do stock. How would you like that? For myself I wouldnt care what my rac did. But some of you would.

---

[/quote]

Old James

Eric, is it possible that the differences in credit rates for MB and AP cpu are due to different hardware populations? IE people with higher-end hardware choose not to run AP?

Unless I completely misunderstand CN, for equal hardware, the most efficient versions (min app_version.pfc_avg) should always factor pfc by 1 and therefore make credit at the same rate.

A question remains why with V6 that works? a simple coincidence?

I wish I could tell you. There aren't enough V6 results in the database to get accurate statistics. And any that are there are from slow machines, or machines that spend most of their time turned off.

Thanks for the honesty, i realy apreciate your return, hope someday you could answer that question.

---

A question remains why with V6 that works? a simple coincidence?

I wish I could tell you. There aren't enough V6 results in the database to get accurate statistics. And any that are there are from slow machines, or machines that spend most of their time turned off.

Do the GPU numbers factor in the ability to run multiple instances of stock per device using app_config.xml now? or do we naturally assume most users are using the default of 1 ?

---

"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 would remind folks not to attempt to hijack this thread, or to flame each other. There are a number of posts here that getting very close to both, so take note.

Interesting. I've seen neither. I see people offering their views on a given situation. If people are able to give their "+1" in here in agreement with a specific viewpoint, are dissenting views not allowed anymore?

I am assuming 1 per GPU. I assume that accounts for the majority of our users and result.

---

@SETIEric@qoto.org (Mastodon)