Host hitting 400 quota every day, now what?

Message boards : Number crunching : Host hitting 400 quota every day, now what?
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Profile Bob Mahoney Design
Avatar

Send message
Joined: 4 Apr 04
Posts: 178
Credit: 9,205,632
RAC: 0
United States
Message 806746 - Posted: 10 Sep 2008, 19:07:00 UTC

I've recently completed construction of a new computer. Daily, it is receiving the following warning from SETI in red text:

Message from server: No work sent
Message from server: (reached daily quota of 400 results)

I know the 400 max is there to limit hosts that are erring out on multiple WU's. Fine.

In this case, the computer in question is simply getting the work done quickly. This computer is dedicated to SETI analysis, so please don't suggest spreading it across multiple projects.

I've seen computers at the top of the list with enough processors to eat this one alive. How are they getting enough work with the 400 limit?

Forgive me if I've missed an obvious solution.

What to do?
ID: 806746 · Report as offensive
Fred W
Volunteer tester

Send message
Joined: 13 Jun 99
Posts: 2524
Credit: 11,954,210
RAC: 0
United Kingdom
Message 806748 - Posted: 10 Sep 2008, 19:14:52 UTC - in response to Message 806746.  

I've recently completed construction of a new computer. Daily, it is receiving the following warning from SETI in red text:

Message from server: No work sent
Message from server: (reached daily quota of 400 results)

I know the 400 max is there to limit hosts that are erring out on multiple WU's. Fine.

In this case, the computer in question is simply getting the work done quickly. This computer is dedicated to SETI analysis, so please don't suggest spreading it across multiple projects.

I've seen computers at the top of the list with enough processors to eat this one alive. How are they getting enough work with the 400 limit?

Forgive me if I've missed an obvious solution.

What to do?

Not a solution, I'm afraid but ...

The 400 limit comprises 100 / day / processor. So those at the top of the list will get a bigger quota.

If you snag some Astropulse WU's, that will slow your machine down a bit ;)

F.
ID: 806748 · Report as offensive
Alinator
Volunteer tester

Send message
Joined: 19 Apr 05
Posts: 4178
Credit: 4,647,982
RAC: 0
United States
Message 806749 - Posted: 10 Sep 2008, 19:16:43 UTC
Last modified: 10 Sep 2008, 19:18:48 UTC

You don't need to do anything. None of your hosts are in danger of running out of work currently, and most likely the TDCF on the new host is adjusting down as track record builds up for it. You are just running out of daily quota as it tries to fill up the cache to your preferences settings.

Eventually the cache will fill when the TDCF stabilizes and the DQ warning will stop.

HTH,

Alinator
ID: 806749 · Report as offensive
john deneer
Volunteer tester
Avatar

Send message
Joined: 16 Nov 06
Posts: 331
Credit: 20,996,606
RAC: 0
Netherlands
Message 806755 - Posted: 10 Sep 2008, 19:47:41 UTC - in response to Message 806746.  
Last modified: 10 Sep 2008, 19:57:20 UTC

I've recently completed construction of a new computer. Daily, it is receiving the following warning from SETI in red text:

Message from server: No work sent
Message from server: (reached daily quota of 400 results)

I know the 400 max is there to limit hosts that are erring out on multiple WU's. Fine.

...

What to do?

Hi bobmah,

You don't have to do anything. It's a 4-core machine, since your quotum is 400 units. Now suppose that you will be crunching 250 units per day on that machine (of course heavily depending upon the type of wu's and the speed of the machine). If you have set a cache of e.g. 5 days, it will begin downloading wu's like a maniac. Once it has downloaded 400 wu's you'll get that message of reaching the quota. It will download another 400 on the second day, and you will get your message after 1 or 2 download sessions on the second day too. And on the third .... and it will continue to do so until it has its cache nicely filled, after which it will settle down and only ask for new wu's once it gets somewhat below a 5-day.

In short: this is perfectly normal behaviour for a fast machine and a cache with the size of a couple of days

Regards,
John.
ID: 806755 · Report as offensive
Profile Sutaru Tsureku
Volunteer tester

Send message
Joined: 6 Apr 07
Posts: 7105
Credit: 147,663,825
RAC: 5
Germany
Message 806756 - Posted: 10 Sep 2008, 19:52:24 UTC


@ bobmah45

Nothing to worry about..

~ 24 hours after the message you rig will get again up to 400 WUs.. ;-D

ID: 806756 · Report as offensive
1mp0£173
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 8423
Credit: 356,897
RAC: 0
United States
Message 806757 - Posted: 10 Sep 2008, 19:54:41 UTC - in response to Message 806746.  

I've recently completed construction of a new computer. Daily, it is receiving the following warning from SETI in red text:

Message from server: No work sent
Message from server: (reached daily quota of 400 results)

I know the 400 max is there to limit hosts that are erring out on multiple WU's. Fine.

In this case, the computer in question is simply getting the work done quickly. This computer is dedicated to SETI analysis, so please don't suggest spreading it across multiple projects.

I've seen computers at the top of the list with enough processors to eat this one alive. How are they getting enough work with the 400 limit?

Forgive me if I've missed an obvious solution.

What to do?

My question is: are you completing 400 work units/day?

If not, you're likely still building up a queue, it just takes more than a single day quota.
ID: 806757 · Report as offensive
Profile JDWhale
Volunteer tester
Avatar

Send message
Joined: 6 Apr 99
Posts: 921
Credit: 21,935,817
RAC: 3
United States
Message 806761 - Posted: 10 Sep 2008, 20:17:03 UTC - in response to Message 806746.  
Last modified: 10 Sep 2008, 20:25:29 UTC


I've seen computers at the top of the list with enough processors to eat this one alive. How are they getting enough work with the 400 limit?

What to do?


Doing the math on a few of the WUs you've returned...
You are crunching VHAR WUs in ~9 minutes. If you were only to crunch these you could complete 600+ WUs/day.

Figuring that you will receive a mix of Angle Ranges from the project and the speed that you are crunching the MidAR WUs (~0.44 AR), ~32 minutes each, it is conceivable that you will crunch more than the 400WU/day download limit for your CPU on some days. This is mainly due to the proliferation of the shorties (VHARs) currently being distributed. Expect your cache to build slowly if at all.

I figure your new host might make the top 10 hosts with RAC > 8000 if the project can keep feeding you enough work.

Nice box you've built.

Regards,
JDWhale
ID: 806761 · Report as offensive
Profile Bob Mahoney Design
Avatar

Send message
Joined: 4 Apr 04
Posts: 178
Credit: 9,205,632
RAC: 0
United States
Message 806775 - Posted: 10 Sep 2008, 21:33:26 UTC

OK, most of the machines at the top are running 8 cores. They'd have a DQ of 800. No problem for them. (Except for that guy with the phase-change QX9650 at 4400MHz. Hoo hah!)

I received lots of those desirable 16.xx WU's, mixed with a ton of the 43.xx. The host is completing them in approximately 550sec and 1920sec, respectively.

Hypothetical: If all the WU's were 16.xx, the total tasks for the day would be:

(4core x 86400sec) / 550sec = 628 completed tasks

If they were all 43.xx:

(4core x 86400sec) / 1920sec = 180 completed tasks

It looks like a I had a few lucky days with >400 results/day. The WU mix is constantly changing on SETI. I'll give it a few days.

Hey, not that I'm complaining about getting the small WU's!

Bob
ID: 806775 · Report as offensive
Profile Bob Mahoney Design
Avatar

Send message
Joined: 4 Apr 04
Posts: 178
Credit: 9,205,632
RAC: 0
United States
Message 806780 - Posted: 10 Sep 2008, 21:55:56 UTC - in response to Message 806761.  



Doing the math on a few of the WUs you've returned...
You are crunching VHAR WUs in ~9 minutes. If you were only to crunch these you could complete 600+ WUs/day.

Figuring that you will receive a mix of Angle Ranges from the project and the speed that you are crunching the MidAR WUs (~0.44 AR), ~32 minutes each, it is conceivable that you will crunch more than the 400WU/day download limit for your CPU on some days. This is mainly due to the proliferation of the shorties (VHARs) currently being distributed. Expect your cache to build slowly if at all.

I figure your new host might make the top 10 hosts with RAC > 8000 if the project can keep feeding you enough work.

Nice box you've built.

Regards,
JDWhale


JDWhale,

Judging from the similarity of the last two posts (mine and yours), we agree on the analysis.

My Mom and Dad died in July. I've dedicated this QX9770 host to the memory of Dad. He enjoyed my projects.

Mom is just gonna need to wait until those 25nm chips come out . . .

And, yes, I do now realize a QX9650 would likely perform just as well. Sigh.

Bob
ID: 806780 · Report as offensive
Profile JDWhale
Volunteer tester
Avatar

Send message
Joined: 6 Apr 99
Posts: 921
Credit: 21,935,817
RAC: 3
United States
Message 806787 - Posted: 10 Sep 2008, 22:28:57 UTC - in response to Message 806780.  

JDWhale,

Judging from the similarity of the last two posts (mine and yours), we agree on the analysis.

My Mom and Dad died in July. I've dedicated this QX9770 host to the memory of Dad. He enjoyed my projects.

Mom is just gonna need to wait until those 25nm chips come out . . .

And, yes, I do now realize a QX9650 would likely perform just as well. Sigh.

Bob

Sorry to hear of your recent losses :-(

As you probably know, performance depends on a few factors... Your "Dads" box has higher BOINC benchmarks than Marks Frozen Penny(QX9650), but he recently switched over to DDR3 RAM and he is running at a higher freqency. Additionally, he's not getting the benefit from crunching all the VHAR WUs that you are. Hard to get an apples-to-apples comparison...

One thing remains certain, when it comes to bang-for-the-buck it still seems single quad-core boxes rule... Too much overhead and contention for bandwith when running more than 4 cores, IMO.
[quote]if (SETI client gets faster) {
    you'll be waiting for work
} ;-)
[/quote]Regards,
JDWhale
ID: 806787 · Report as offensive
Profile Bob Mahoney Design
Avatar

Send message
Joined: 4 Apr 04
Posts: 178
Credit: 9,205,632
RAC: 0
United States
Message 807011 - Posted: 11 Sep 2008, 14:50:50 UTC - in response to Message 806787.  


...
As you probably know, performance depends on a few factors... Your "Dads" box has higher BOINC benchmarks than Marks Frozen Penny(QX9650), but he recently switched over to DDR3 RAM and he is running at a higher freqency. Additionally, he's not getting the benefit from crunching all the VHAR WUs that you are. Hard to get an apples-to-apples comparison...
...
JDWhale


I did an apples-to-apples comparo between "Dad's" water-cooled QX9770 and Msattler's phase-change QX9650 Frozen Penny last week. On similar WU's he was 8% faster. Today, on a per-WU basis, he is 10% faster. Very impressive, since this indicates he's optimized FSB and memory to match the 4400-4420MHz processor frequency, not yet getting into an area of diminishing returns per MHz.

So if he would fully dedicate that crazy system 24/7 he should occasionally bump into the 400/day task limit.

Let these hosts be seen as an evolutionary step for SETI with the arrival of 100 task/day processor cores.

To be fair, there probably are faster 'per-core' systems running bursts of SETI here and there. IMHO they need to make it into the top 40 computers before being considered a fast AND consistent host.

Side note: My host hit the 400 task limit again this morning. It seems to be hitting earlier each day. I still like those short WU's. We'll see how the WU balance unfolds.

Bob
ID: 807011 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 807020 - Posted: 11 Sep 2008, 15:41:49 UTC
Last modified: 11 Sep 2008, 15:42:21 UTC

The Frozen Penny has had no problem maintaining it's cache, as long as there are WUs available on the servers.....

It will download 20 to 60, go into EDF mode and crunch up the little shorties, leaving the longer deadline WUs in the cache. Repeat cycle a couple of times, and when it is satisfied that the cache is full, it goes back to normal processing at the top of the cache again.

I haven't gone into my results to count the number of WUs per day it is actually returning, but the info is there if somebody wants to tally it up.

Nice cruncher there, bobmah45!!!

Oh, and yes....the Frozen Penny has been running so stable lately I had to do something to tweak it, LOL. Just last night I bumped the FSB 2mhz from 440 to 442...taking the cpu from 4.4ghz to 4.42ghz and the ram speed up to 1768mhz DDR3...we'll see if it stays stable today while I am away at work. I'll have the kitties keep an eye on it for me.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 807020 · Report as offensive
john deneer
Volunteer tester
Avatar

Send message
Joined: 16 Nov 06
Posts: 331
Credit: 20,996,606
RAC: 0
Netherlands
Message 807025 - Posted: 11 Sep 2008, 16:09:38 UTC - in response to Message 807011.  

.. Let these hosts be seen as an evolutionary step for SETI with the arrival of 100 task/day processor cores.

Bob

Hello Bob,

Reacting on Mark's post (he wondered how many wu's he is actually returning per day) I made a quick scan of the results of your 9770. Between Sept 10 15:00 and September 11, 15:00 (times are UTC) you returned approx. 270 wu's, approx. 100 of which were still pending. Of the 270 wu's some 110 were shorties, of which 28 were still pending.

So although your rig is doing great (better than mine, anyway) you are still a long way from 100/day/core. You're running into the limit of 400/day because you are still building your cache and, moreover, there are not a lot of wu's to be had during the last 12 hours or so, and your rig may therefore not get any new wu's eventhough you haven't hit the daily limit.

Regards,
John.
ID: 807025 · Report as offensive
Profile Bob Mahoney Design
Avatar

Send message
Joined: 4 Apr 04
Posts: 178
Credit: 9,205,632
RAC: 0
United States
Message 807055 - Posted: 11 Sep 2008, 18:38:32 UTC - in response to Message 807025.  

... Between Sept 10 15:00 and September 11, 15:00 (times are UTC) you returned approx. 270 wu's, approx. 100 of which were still pending. Of the 270 wu's some 110 were shorties, of which 28 were still pending.

So although your rig is doing great (better than mine, anyway) you are still a long way from 100/day/core. ...

Regards,
John.


I have confirmed your analysis. I slid a 24-hour window back through the last 3 days of the "Time Reported" column of the "Tasks for Computer" screens for the QX9770. Highest I could find was 300 reported results.

It looks like rumors of the untimely demise of the 100/day/core barrier were greatly exagerated. By me. The megalomania might not be in remission after all.

Some technical touch-points:

Let's consider this host to be 270 tasks/day with today's task mix. At 4-cores, that gives 68 tasks/core/day. At that pace, SETI is reporting a TDCF (Task Duration Correction Factor) averaging around .097.

What TDCF would be needed for 100 tasks/day/core?

I contend:

68tasks now / 100tasks goal x .097TDCF = .068TDCF for 100 tasks/day/core

The first person to achieve a TDCF of .068 should achieve 100 tasks/day/core, and I'd estimate a RAC of 12,500 for such a 4-core machine.

Can anyone confirm this theoretical TDCF holy grail? If I'm right, everyone should calculate a goal TDCF around .068, no matter what your current computer speed.

Hints in case anyone is new at this: On this site, click on 'tasks' for your host. Click "next 20" until you see completed tasks. Count the number of screens (20 taks per screen) for a few full days indicated by the dates in the "Time Reported" column. Calculate the tasks/day/core. Now look up your Task Duration Correction Factor - the last line of info on your "Computer Information" screen on the SETI site.

Plug the info into this equation:

(your current tasks per day per core) / 100 x (your TDCF) = x

What is your x? In theory, that is the TDCF needed for 100 tasks/day/core, 12,500 recent average credit.

Perhaps from this experiment we can learn to predict anyone's potential RAC as a simple derivative of their current TDCF?

And I do apologize if this seems insane.

Bob
ID: 807055 · Report as offensive
Profile Bob Mahoney Design
Avatar

Send message
Joined: 4 Apr 04
Posts: 178
Credit: 9,205,632
RAC: 0
United States
Message 807061 - Posted: 11 Sep 2008, 19:05:12 UTC - in response to Message 807020.  

The Frozen Penny has had no problem maintaining it's cache, as long as there are WUs available on the servers.....

It will download 20 to 60, go into EDF mode and crunch up the little shorties, leaving the longer deadline WUs in the cache. Repeat cycle a couple of times, and when it is satisfied that the cache is full, it goes back to normal processing at the top of the cache again.

I haven't gone into my results to count the number of WUs per day it is actually returning, but the info is there if somebody wants to tally it up.

Nice cruncher there, bobmah45!!!

Oh, and yes....the Frozen Penny has been running so stable lately I had to do something to tweak it, LOL. Just last night I bumped the FSB 2mhz from 440 to 442...taking the cpu from 4.4ghz to 4.42ghz and the ram speed up to 1768mhz DDR3...we'll see if it stays stable today while I am away at work. I'll have the kitties keep an eye on it for me.


My clandestine analysis of your machine over the past week indicates you've recently achieved a 1.5% increase in WU speed. I would attribute most of this to your fine tuning of the memory access speeds.

I would predict your currently achievable RAC to be "y" in:

.097 x (your TDCF) x 8500 = y

This equation might be conservative, but is probably accurate for finding a minimum achievable RAC for a steady host with today's WU mix.

Ultimate cruncher there, msattler! And no, I don't want to hear what voltages you are running. I scare easily.

Bob
ID: 807061 · Report as offensive
Profile dnolan
Avatar

Send message
Joined: 30 Aug 01
Posts: 1228
Credit: 47,779,411
RAC: 32
United States
Message 807077 - Posted: 11 Sep 2008, 20:26:38 UTC - in response to Message 807055.  



Plug the info into this equation:

(your current tasks per day per core) / 100 x (your TDCF) = x

What is your x? In theory, that is the TDCF needed for 100 tasks/day/core, 12,500 recent average credit.

Perhaps from this experiment we can learn to predict anyone's potential RAC as a simple derivative of their current TDCF?

And I do apologize if this seems insane.

Bob


Ok, ran one of mine through this equation:
Machine has 185 tasks for last 24 hours, that's 46.25 per core, current TCDF = .103038
46.25 / 100 * .103038 = .04765

Seems a lot lower than what you got, maybe the slower the machine, the lower the TCDF would need to go to compensate? The result makes sense, though, because if I'm less than 1/2 way to 100 tasks/core/day, then the TCDF needs to be cut by more than 1/2.

-Dave
ID: 807077 · Report as offensive
Profile Bob Mahoney Design
Avatar

Send message
Joined: 4 Apr 04
Posts: 178
Credit: 9,205,632
RAC: 0
United States
Message 807139 - Posted: 11 Sep 2008, 23:03:07 UTC - in response to Message 807077.  


Ok, ran one of mine through this equation:
Machine has 185 tasks for last 24 hours, that's 46.25 per core, current TCDF = .103038
46.25 / 100 * .103038 = .04765

Seems a lot lower than what you got, maybe the slower the machine, the lower the TCDF would need to go to compensate? The result makes sense, though, because if I'm less than 1/2 way to 100 tasks/core/day, then the TCDF needs to be cut by more than 1/2.

-Dave


OK, that might be enough to debunk this particular theory - lack of consistency in the TCDF.

From this small sample population, we can guess the goal TCDF for 100 tasks/core/day must be a number below .07, something that seems unobtainable with today's consumer technology. I wonder if it will take us a full year or more to achieve that type of speed with an Intel product? I like AMD, but their crunching performance just isn't there unless you go to lots and lots of processors. The bang-per-buck and bang-per-watt on those boxes is not competitive at this time.

I'm taking this to a new thread here. I'll call it something like Predicting RAC. There must be a way to predict RAC without crunching two months to get there.

Bob
ID: 807139 · Report as offensive
Profile dnolan
Avatar

Send message
Joined: 30 Aug 01
Posts: 1228
Credit: 47,779,411
RAC: 32
United States
Message 807153 - Posted: 11 Sep 2008, 23:37:00 UTC

I'm pretty sure TCDF relies on what the system THINKS your cpu should take to process a WU, if that's correct, then it would have to be based on benchmarks, so the basic idea is that the system THINKS you should take 2 hours to do a particular WU, but you take 20 minutes, or 1/6 of 2 hours, so it adjusts the TCDF to account for this, so you get a TCDF of .1666 or so. But then maybe you run into a WU that is supposed to take 30 minutes, but you do it in 10, or 1/3, then it has to adjust back up. So it seems like it would be a combination of what the system thinks you should be doing for a WU vs. how accurate the system is in general for those WU times (for the angle range), vs. if your system for some reason benches high or low... so I don't think it could be used to predict RAC. But I could be wrong on any of my assumptions...

-Dave
ID: 807153 · Report as offensive
Profile zoom3+1=4
Volunteer tester
Avatar

Send message
Joined: 30 Nov 03
Posts: 65709
Credit: 55,293,173
RAC: 49
United States
Message 807195 - Posted: 12 Sep 2008, 1:55:53 UTC - in response to Message 807020.  

The Frozen Penny has had no problem maintaining it's cache, as long as there are WUs available on the servers.....

It will download 20 to 60, go into EDF mode and crunch up the little shorties, leaving the longer deadline WUs in the cache. Repeat cycle a couple of times, and when it is satisfied that the cache is full, it goes back to normal processing at the top of the cache again.

I haven't gone into my results to count the number of WUs per day it is actually returning, but the info is there if somebody wants to tally it up.

Nice cruncher there, bobmah45!!!

Oh, and yes....the Frozen Penny has been running so stable lately I had to do something to tweak it, LOL. Just last night I bumped the FSB 2mhz from 440 to 442...taking the cpu from 4.4ghz to 4.42ghz and the ram speed up to 1768mhz DDR3...we'll see if it stays stable today while I am away at work. I'll have the kitties keep an eye on it for me.

Gee maybe I should take a chance and bump the fsb of PC2 to 444 from 439(this ram can do 468(DDR2), piece of cake, The cpu just needs water cooling or an IFX-14 to cool the QX6700 cpu), PC1 is already at 468MHz(DDR2), Good thing I have a 2nd QX6700 in PC3, For when I upgrade the motherboard and the ram, I already know what works(Thanks to the Kittyman). But 4.42GHz? I don't think I'll get there, At least not as close as the Frozen Penny is to 5GHz, But then I don't have the warp drive cooler or the money for It, The closest I could get would be a Freezone Elite cpu cooler(around $400+UPS) and I have no idea where that would take a QX6700 cpu or a Q9xx0 cpu, But It would be cooler(pardon the pun).
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 807195 · Report as offensive
John McLeod VII
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 15 Jul 99
Posts: 24806
Credit: 790,712
RAC: 0
United States
Message 807210 - Posted: 12 Sep 2008, 2:44:57 UTC - in response to Message 807139.  


Ok, ran one of mine through this equation:
Machine has 185 tasks for last 24 hours, that's 46.25 per core, current TCDF = .103038
46.25 / 100 * .103038 = .04765

Seems a lot lower than what you got, maybe the slower the machine, the lower the TCDF would need to go to compensate? The result makes sense, though, because if I'm less than 1/2 way to 100 tasks/core/day, then the TCDF needs to be cut by more than 1/2.

-Dave


OK, that might be enough to debunk this particular theory - lack of consistency in the TCDF.

From this small sample population, we can guess the goal TCDF for 100 tasks/core/day must be a number below .07, something that seems unobtainable with today's consumer technology. I wonder if it will take us a full year or more to achieve that type of speed with an Intel product? I like AMD, but their crunching performance just isn't there unless you go to lots and lots of processors. The bang-per-buck and bang-per-watt on those boxes is not competitive at this time.

I'm taking this to a new thread here. I'll call it something like Predicting RAC. There must be a way to predict RAC without crunching two months to get there.

Bob

Bad analysis.

The Duration Correction Factor starts with the benchmarks and the FLOPs estimate to determine how long the task would take if both the benchmarks and the FLOPs estimate were accurate. If both are perfectly accurate, then the DCF is 1.0 (it takes exactly as long on the machine as the benchmarks would leave you to believe). If the computer takes less time than expected, then the DCF would be lower than 1.0. If it takes longer than the estimate, the DCF would be higher than 1.0.

Another factor is that DCF moves fairly slowly (to avoid the effects of a string of -9 tasks forcing a work fetch that is too large for the machine).


BOINC WIKI
ID: 807210 · Report as offensive
1 · 2 · Next

Message boards : Number crunching : Host hitting 400 quota every day, now what?


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.