Ok now that the politics has landed on the Milky way project to make them bring their project more in line with projects like SETI. Where is the ATI application on SETI so that I can get a copy to use? I mean come on I want to use it to compare with Milky Way project.

Maybe I am mistaken and maybe SETI does not have one for this project. If not why would you bully other projects to conform to something that you don't have here. Just think about that one..

Heck you can't even keep this project running for more that two days.

SM..

---

Not sure what politics you are talking about, as I only run Seti, but there isn't an ATI app for any type of Seti at the moment. There is only a Cuda app because NVidia programmers/engineers ported the Seti code to it.

regards, Gizbar.

---

A proud GPU User Server Donor!

Not sure what politics you are talking about, as I only run Seti, but there isn't an ATI app for any type of Seti at the moment. There is only a Cuda app because NVidia programmers/engineers ported the Seti code to it.

regards, Gizbar.

There is another BOINC community composed of projects that run under BOINC, and from what I've read (very indirectly) there has always been an ongoing (sometimes heated) discussion over credit.

If a project overpays dramatically, other projects believe that they have to raise their credit to match, or lose crunchers to other projects.

At the same time, raising credit at a project devalues earlier work, so that isn't exactly a good thing.

That said, I'm sure that community is like this one, and sometimes "good ideas" seem like demands or requirements.

I pick projects based on my own personal interest, not on credits.

---

Thanks Ned.

Sorry to Star Man, I wasn't trying to hijack your thread...

regards, Gizbar.

---

A proud GPU User Server Donor!

Ned you guessed right. The credit police complained that Milkyway was overpaying(awarding too much credit.) So they reduced the amount being granted. Not sure if the reduction was really called for. Some ATI cards were returning completed work units in just a few seconds. The same work done by a CPU took several hours. Both got the same number of credits. The result is that ATI cards were getting huge awards. But they were also doing a lot of work. Not sure I really understand whats wrong with giving large credit awards if the work was done.

Ned you guessed right. The credit police complained that Milkyway was overpaying(awarding too much credit.) So they reduced the amount being granted. Not sure if the reduction was really called for. Some ATI cards were returning completed work units in just a few seconds. The same work done by a CPU took several hours. Both got the same number of credits. The result is that ATI cards were getting huge awards. But they were also doing a lot of work. Not sure I really understand whats wrong with giving large credit awards if the work was done.

Thank you Carlos,

You have hit the nail right on the head.

If you buy the equipment to go faster, you should get the credit for your labors.

SM

---

Ned you guessed right. The credit police complained that Milkyway was overpaying(awarding too much credit.) So they reduced the amount being granted. Not sure if the reduction was really called for. Some ATI cards were returning completed work units in just a few seconds. The same work done by a CPU took several hours. Both got the same number of credits. The result is that ATI cards were getting huge awards. But they were also doing a lot of work. Not sure I really understand whats wrong with giving large credit awards if the work was done.

Thank you Carlos,

You have hit the nail right on the head.

If you buy the equipment to go faster, you should get the credit for your labors.

SM

... and since one of the constants in the BOINC world is "equal pay for equal work" it doesn't make sense that there would be complaints based only on the ATI scores.

But if we're talking about ATI being more than 3000 times faster than the same work done by CPU, that just doesn't make sense.

At the same time, this doesn't seem to be the right forum for the argument.

The people here do not represent SETI@Home, we're participants. I don't know if Star Man represents Milkyway or if he's just a participant there.

---

Just FYI, copied from the MilkyWay homepage:

Credit Reduction

September 11, 2009

Milkyway's credit awarded has been a bit too high for awhile now, and we're bringing it down to be in line more with other projects. We've been using a credit multiplier of 7.5 per double precision FLOP and other projects are using around 5.4 per double precision FLOP. We're reducing our multipler to this, so credit should be down around 20-30%. I know everyone hates credit reductions, but it was really needed. We also have some performance improvements for CUDA and ATI applications on the way out, so hopefully it won't be too bad. As usual, try to keep the forum flaming about the credit changes somewhat civil.

There's no mentioning of SETI and no word about not "get[ting] the credit for your labors".

Gruß,
Gundolf

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

SETI@home classic workunits 3,758
SETI@home classic CPU time 66,520 hours

Just FYI, copied from the MilkyWay homepage:

Credit Reduction

September 11, 2009

Milkyway's credit awarded has been a bit too high for awhile now, and we're bringing it down to be in line more with other projects. We've been using a credit multiplier of 7.5 per double precision FLOP and other projects are using around 5.4 per double precision FLOP. We're reducing our multipler to this, so credit should be down around 20-30%. I know everyone hates credit reductions, but it was really needed. We also have some performance improvements for CUDA and ATI applications on the way out, so hopefully it won't be too bad. As usual, try to keep the forum flaming about the credit changes somewhat civil.

There's no mentioning of SETI and no word about not "get[ting] the credit for your labors".

Gruß,
Gundolf

It has been so long since money has been on a gold standard that we seem to have forgotten what "gold standard" means.

It means that a unit of currency is directly related to a single commodity. For money, that's traditionally been Gold or Silver, and each bank note represented a specific amount of precious metal in a vault somewhere.

The definition of a Cobblestone is a "gold standard" in that it represents 1/100th of the daily output of a computer with specific characteristics.

Sounds like Milkyway is moving toward that "gold standard" and trying to make their credit comparable to other projects.

I know nothing of what lead to their decision, but peer pressure may have been involved.

Credit police? What can they do? Projects are independent, there is no central management to pull the plug.

---

Getting back to the topic at hand...

Eric said there is an effort underway by the crew at Lunatics to try and create an Astropulse app that will run on the ATI cards.

I would have thought it may have been easier to take the cuda app and create a CAL/Brook+ app out of it, but it seems not.

---

BOINC blog

I would have thought it may have been easier to take the cuda app and create a CAL/Brook+ app out of it, but it seems not.

It depends algorithm of what app you already understand better :p
Also, CUDA != Brook+.
You can't do many things that possible in CUDA kernels in Brook kernels.
So, app should be rewritten anyway...

[Ah, almost forgot ;) There is no "BroFFT" or whatever. No FFT library for ATI GPUs. Open field here ;) ]

How do I know if my ATI card can work with their flavor of CUDA?

---

30+ Computers heading our way! Currently at the "Zomg we need to talk to our tech expert at the co-op about this first!!!" stage. 16 Lab machines and 14+ Staff machines each with 2.2Ghz CPUs and 256MB ram. Think they balance? The RAM certainly is bad

How do I know if my ATI card can work with their flavor of CUDA?

For projects that support ATI GPUs it seems that some need double precision floating point operation support. You can check to see if your card does here at ATI Stream SDK System Requirements. Until said application is actually released with requirements you may not know for sure.

---

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

I would have thought it may have been easier to take the cuda app and create a CAL/Brook+ app out of it, but it seems not.

It depends algorithm of what app you already understand better :p
Also, CUDA != Brook+.
You can't do many things that possible in CUDA kernels in Brook kernels.
So, app should be rewritten anyway...

[Ah, almost forgot ;) There is no "BroFFT" or whatever. No FFT library for ATI GPUs. Open field here ;) ]

Raistmer is any one looking at porting to Mac OSX 10.6 now that it supports both Nvidia and ATI GPU crunching. I say on one web site that Snow Leopard supports Open CL to support dynamic computation between CPU and GPU based on system load. From reading the article with a recompile of the basic app it uses dynamic JIT to run either on the CPU or GPU. Is it being looked at? My current Mac does not support it but my next one should. My current Mac has a crappy intel GPU but my next one should have a Nvidia GPU.
Links to some info on Open CL and GCD (Grand Central Dispatch) on OSX 10.6 http://arstechnica.com/apple/reviews/2009/08/mac-os-x-10-6.ars this is a link to the Ars Technica 10.6 review
now the links to the specific pages in the review. also a link to the new compiler shipping with the 10.6 xTools 3.2?
Compiler info
http://arstechnica.com/apple/reviews/2009/08/mac-os-x-10-6.ars/9
Link to the new Concurrency framework for the OS link
http://arstechnica.com/apple/reviews/2009/08/mac-os-x-10-6.ars/12
Link to Open CL stuff and a bit about tying it into GCD and the ability to run the came code on GPU and CPU based on free compute resources.
http://arstechnica.com/apple/reviews/2009/08/mac-os-x-10-6.ars/14

Edit the Ars Technica review runs to 26 pages with links to apple dev docs and other resources. Also the follow up discussion has some good points in it.

---

Just to fuel speculation I noticed this when putting 6.10.4 on for test.

9/16/2009 7:15:40 PM Starting BOINC client version 6.10.4 for windows_intelx86
9/16/2009 7:15:40 PM log flags: task, file_xfer, sched_ops
9/16/2009 7:15:40 PM Libraries: libcurl/7.19.4 OpenSSL/0.9.8k zlib/1.2.3
9/16/2009 7:15:40 PM Data directory: Z:\BOINC\DATA
9/16/2009 7:15:40 PM Running under account Tester
9/16/2009 7:15:40 PM Processor: 2 GenuineIntel Intel(R) Core(TM)2 Duo CPU E8400 @ 3.00GHz [x86 Family 6 Model 23 Stepping 6]
9/16/2009 7:15:40 PM Processor: 6.00 MB cache
9/16/2009 7:15:40 PM Processor features: fpu tsc sse sse2 pni mmx
9/16/2009 7:15:40 PM OS: Microsoft Windows Vista: Ultimate x86 Edition, Service Pack 2, (06.00.6002.00)
9/16/2009 7:15:40 PM Memory: 3.25 GB physical, 4.19 GB virtual
9/16/2009 7:15:40 PM Disk: 4.00 GB total, 1.99 GB free
9/16/2009 7:15:40 PM Local time is UTC -4 hours
9/16/2009 7:15:40 PM Can't load library nvcuda.dll
9/16/2009 7:15:40 PM ATI GPU 0: RV770 (CAL version 1.4.317, 512MB, 1200GFLOPS)
9/16/2009 7:15:40 PM SETI@home Found app_info.xml; using anonymous platform
9/16/2009 7:15:40 PM Not using a proxy
9/16/2009 7:15:40 PM Version change (6.6.36 -> 6.10.4)
9/16/2009 7:15:40 PM SETI@home URL: http://setiathome.berkeley.edu/; Computer ID: 5011745; location: home; project prefs: home
9/16/2009 7:15:40 PM SETI@home General prefs: from SETI@home (last modified 10-Sep-2009 22:34:36)
9/16/2009 7:15:40 PM SETI@home Computer location: home
9/16/2009 7:15:40 PM SETI@home General prefs: no separate prefs for home; using your defaults
9/16/2009 7:15:40 PM Reading preferences override file
9/16/2009 7:15:40 PM Preferences limit memory usage when active to 3325.57MB
9/16/2009 7:15:40 PM Preferences limit memory usage when idle to 3325.57MB
9/16/2009 7:15:40 PM Preferences limit disk usage to 2.11GB
9/16/2009 7:15:40 PM Running CPU benchmarks
9/16/2009 7:15:40 PM Suspending computation - running CPU benchmarks
9/16/2009 7:16:11 PM Benchmark results:
9/16/2009 7:16:11 PM Number of CPUs: 2
9/16/2009 7:16:11 PM 3146 floating point MIPS (Whetstone) per CPU
9/16/2009 7:16:11 PM 6359 integer MIPS (Dhrystone) per CPU
9/16/2009 7:16:11 PM SETI@home Restarting task 25au09ab.16172.17654.6.10.168_1 using setiathome_enhanced version 603
9/16/2009 7:16:11 PM SETI@home Restarting task 25au09aa.15546.17659.15.10.13_1 using setiathome_enhanced version 603
9/16/2009 7:16:16 PM SETI@home Sending scheduler request: To fetch work.
9/16/2009 7:16:16 PM SETI@home Requesting new tasks for GPU
9/16/2009 7:16:21 PM SETI@home Scheduler request completed: got 0 new tasks
9/16/2009 7:16:21 PM SETI@home Message from server: No work sent
9/16/2009 7:17:36 PM SETI@home Sending scheduler request: To fetch work.
9/16/2009 7:17:36 PM SETI@home Requesting new tasks for GPU
9/16/2009 7:17:41 PM SETI@home Scheduler request completed: got 0 new tasks
9/16/2009 7:17:41 PM SETI@home Message from server: No work sent
9/16/2009 7:18:56 PM SETI@home Sending scheduler request: To fetch work.
9/16/2009 7:18:56 PM SETI@home Requesting new tasks for GPU
9/16/2009 7:19:01 PM SETI@home Scheduler request completed: got 0 new tasks
9/16/2009 7:19:01 PM SETI@home Message from server: No work sent

The odd bit being I don't have anything in my app_info.xml for any GPUs. Also I have GPU processing disabled in my account settings.

I've just started looking at it. So it might be nothing. Just amused me so far lol.

---

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

Just to fuel speculation I noticed this when putting 6.10.4 on for test.

<snip>

9/16/2009 7:18:56 PM SETI@home Sending scheduler request: To fetch work.
9/16/2009 7:18:56 PM SETI@home Requesting new tasks for GPU
9/16/2009 7:19:01 PM SETI@home Scheduler request completed: got 0 new tasks
9/16/2009 7:19:01 PM SETI@home Message from server: No work sent

The odd bit being I don't have anything in my app_info.xml for any GPUs. Also I have GPU processing disabled in my account settings.

I've just started looking at it. So it might be nothing. Just amused me so far lol.

If you have an app_info it overrides the web setting, so having told it you don't want gpu on the web site gets overridden.

The fact that its found the gpu means it will ask for work for it. If you don't want BOINC using it you can code <no_gpus>1</no_gpus> in cc_config, however that means it won't use the gpu for ANY of your projects. In fact BOINC will treat it like you don't have any gpu at all.

---

BOINC blog

Why deny the GPU work? If SETI wants to shove work onto your card, let it!

Or would that cause problems if there is a lack of application for CAL?

---

30+ Computers heading our way! Currently at the "Zomg we need to talk to our tech expert at the co-op about this first!!!" stage. 16 Lab machines and 14+ Staff machines each with 2.2Ghz CPUs and 256MB ram. Think they balance? The RAM certainly is bad

Just to fuel speculation I noticed this when putting 6.10.4 on for test.

<snip>

9/16/2009 7:18:56 PM SETI@home Sending scheduler request: To fetch work.
9/16/2009 7:18:56 PM SETI@home Requesting new tasks for GPU
9/16/2009 7:19:01 PM SETI@home Scheduler request completed: got 0 new tasks
9/16/2009 7:19:01 PM SETI@home Message from server: No work sent

The odd bit being I don't have anything in my app_info.xml for any GPUs. Also I have GPU processing disabled in my account settings.

I've just started looking at it. So it might be nothing. Just amused me so far lol.

If you have an app_info it overrides the web setting, so having told it you don't want gpu on the web site gets overridden.

The fact that its found the gpu means it will ask for work for it. If you don't want BOINC using it you can code <no_gpus>1</no_gpus> in cc_config, however that means it won't use the gpu for ANY of your projects. In fact BOINC will treat it like you don't have any gpu at all.

I had disabled the cc_config on that system prior to the upgrade. However, I had left app_info.xml enabled to finish out the tasks before switching to the standard apps for test.

I was just not expecting seti@home requests for gpu tasks with an ati gpu. As there is not currently an application for it. Just one of those "wait, what, lol" moments.

---

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

Why deny the GPU work? If SETI wants to shove work onto your card, let it!

Or would that cause problems if there is a lack of application for CAL?

If it got any tasks and tried to run them, they'd all just error out.

-Dave

---

Switched over to the standard applications & it still requests work but gracefully handles the event.

9/18/2009 1:44:30 PM SETI@home Sending scheduler request: To fetch work.
9/18/2009 1:44:30 PM SETI@home Requesting new tasks for GPU
9/18/2009 1:44:35 PM SETI@home Scheduler request completed: got 0 new tasks
9/18/2009 1:44:35 PM SETI@home Message from server: No work sent
9/18/2009 1:44:35 PM SETI@home Message from server: Your computer has no NVIDIA GPU

I'm going to have to grab 6.10.5, since they released that yesterday, and try it before I submit that to the alpha list. The "Use Graphics Processing Unit (GPU) if available" value is being ignored by this version in all cases, where the "Use Central Processing Unit (CPU)" value is OK.

---

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