the other day i did an astropulse work unit with v512 with only 20 seconds run time, and 3 seconds of cpu use. i thought i had a problem, or maybe my computer transformed overnight into a super computer, but them my wingman validated my result!

http://setiathome.berkeley.edu/workunit.php?wuid=790928136

turns out to be a 100% blanked work unit. still i got 0.04 credit for it so it isnt all bad
/sarcastic

cant the servers weed out work units that are like this? it's a waste of bandwidth and storage space. infact they should weed out all work units over a certain precent of blanking. that would help to cut down on server and internet connection load.

the other day i did an astropulse work unit with v512 with only 20 seconds run time, and 3 seconds of cpu use. i thought i had a problem, or maybe my computer transformed overnight into a super computer, but them my wingman validated my result!

http://setiathome.berkeley.edu/workunit.php?wuid=790928136

turns out to be a 100% blanked work unit. still i got 0.04 credit for it so it isnt all bad
/sarcastic

cant the servers weed out work units that are like this? it's a waste of bandwidth and storage space. infact they should weed out all work units over a certain precent of blanking. that would help to cut down on server and internet connection load.

They could spend the resources to do that...or just send them out to the largest distributed computing programme in the world to do it for them.

Just a thought ;)

---

Executive Director GPU Users Group Inc. -
brad@gpuug.org

I tested also this new app..

With my E7600 (Duo-CPU) with GTX260 OC (WinXP 32bit, 275.33 driver).

1 WU/GPU:
Very low CPU support, like with CUDA (~ 5 to ~ 10 % CPU-Core).
GPU load vary ~ 50 to ~ 80 %.

2 WUs/GPU:
Every app/WU get ~ 50 % of one CPU-Core, so ~ 50 % of whole CPU only for AP OpenCL apps support.
GPU load vary also ~ 50 to 80 %. But the usage last longer.

Someone other saw also this high CPU support if 2+ WUs/GPU (maybe - only on non Fermi GPUs? / only with 275.33 driver?)?
I guess maybe the system is overloaded and the CPU is in a loop and wait for further calculation..


- Best regards! - Sutaru Tsureku, team seti.international founder. - Optimize your PC for higher RAC. - SETI@home needs your help. -

---

I only see the high CPU load in combination openCL + 275.33. And it's the same with the Milkyway-OpenCL Application, so i say, its a driver problem.

---

- Performance is not a simple linear function of the number of CPUs you throw at the problem. -

I tested also this new app..

With my E7600 (Duo-CPU) with GTX260 OC (WinXP 32bit, 275.33 driver).

1 WU/GPU:
Very low CPU support, like with CUDA (~ 5 to ~ 10 % CPU-Core).
GPU load vary ~ 50 to ~ 80 %.

2 WUs/GPU:
Every app/WU get ~ 50 % of one CPU-Core, so ~ 50 % of whole CPU only for AP OpenCL apps support.
GPU load vary also ~ 50 to 80 %. But the usage last longer.

Someone other saw also this high CPU support if 2+ WUs/GPU (maybe - only on non Fermi GPUs? / only with 275.33 driver?)?
I guess maybe the system is overloaded and the CPU is in a loop and wait for further calculation..


- Best regards! - Sutaru Tsureku, team seti.international founder. - Optimize your PC for higher RAC. - SETI@home needs your help. -

@Sutaru Tsureku

Hi teammate
It's me Paul Harris rank number 4 in team.
I was wondering what I have to do to get this running on my machine and will this interfere with my Lunatics app I am now running. I would like to run astropulse on my gpu's.

---

Hi,

to now the Astropulse OpenCL app for nVIDIA GPUs is still beta.

From my experiences (on my system (E7600 + GTX260 OC)) an AP WU last ~ 1 to ~ 4 hours with this currently app.

So currently I guess the max performance (RAC) would be still only with Multibeam CUDA on nVIDIA GPUs (at least non Fermi GPUs).

I don't know how it's with Fermi (GTX4xx+) GPUs.
Maybe someone could jump in here?

If you like to test this beta app, we could help you to add this app.


- Best regards! - Sutaru Tsureku, team seti.international founder. - Optimize your PC for higher RAC. - SETI@home needs your help. -

---

Hi Paul, first I would suggest you upgrade your drivers to 26x.xx the newest 275.xx drivers seem to be giving problems to a lot of people and your 258.xx might be a bit too far back.Grab these files http://files.mail.ru/W3B0CG from Raistmer and drop them in your project data folder. Be sure to read the first post in this thread carefully then copy either the app_info segment Raistmer gave in it or mine a bit later in the thread and place in your existing app_info file.Once you get it running check it to make sure everything is working ok and come on back to this thread if you have any problems.

With the way things are going it might take awhile to get some AP work. Also, once you do get some work, expect it to go into high priority as the to completion guestimate will be way off.

Hope that helps you some. My little GTS 450 is running them in about 3 and a half hours more or less depending on blanking. 460s are reporting a lot faster than that. As it comes it should run one instance on each of your 460s but it can be changed to where it is running two or more APs and also MBs at the same time on each GPU. I'm sure others will be along to give better advice on how to set that up.

---

PROUD MEMBER OF Team Starfire World BOINC

Hi Paul, first I would suggest you upgrade your drivers to 26x.xx the newest 275.xx drivers seem to be giving problems to a lot of people and your 258.xx might be a bit too far back.Grab these files http://files.mail.ru/W3B0CG from Raistmer and drop them in your project data folder. Be sure to read the first post in this thread carefully then copy either the app_info segment Raistmer gave in it or mine a bit later in the thread and place in your existing app_info file.Once you get it running check it to make sure everything is working ok and come on back to this thread if you have any problems.

With the way things are going it might take awhile to get some AP work. Also, once you do get some work, expect it to go into high priority as the to completion guestimate will be way off.

Hope that helps you some. My little GTS 450 is running them in about 3 and a half hours more or less depending on blanking. 460s are reporting a lot faster than that. As it comes it should run one instance on each of your 460s but it can be changed to where it is running two or more APs and also MBs at the same time on each GPU. I'm sure others will be along to give better advice on how to set that up.

@perryjay
Hi thanks for the help. I am following your instructions.
First I have downloaded the Raistmer file but where is the project data folder I use 6.10.60 default install.

---

@perryjay
Hi thanks for the help. I am following your instructions.
First I have downloaded the Raistmer file but where is the project data folder I use 6.10.60 default install.

Your data directory path is listed in the first few Boinc messages when you restart Boinc...find that first.
Then click on 'projects', then click on 'setiathome.berkeley.edu....that's your project data folder for Seti.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Mine is in C://program data/BOINC/projects/setiathome. Yours may be someplace else depending on your OS and how you installed. You should be able to find the right path in the first few messages in your BOINC Manager. Drop the files in the project folder and put the app_info segment in the app_info. It's best to just add it and leave the regular AP app in there too so that you can finish any CPU APs you may have. You can also continue running APs on your CPU if you wish.

---

PROUD MEMBER OF Team Starfire World BOINC

@perryjay; Mark
I have 2 boinc folders in 2 different locations
C:/documents and settings/all users/boinc/

C:/program files/boinc

I believe you mean C:/documents and settings/all users/boinc/projects/setiathome.berkley.edu

And add the app_info from you or raistmer to my app_info file

Is this right?
I also need to close boinc before I edit my app_info file the restart boinc.
Also I need to change the <count>1</count> from 1 to .5 to use 2 wu per card

---

@perryjay; Mark
I have 2 boinc folders in 2 different locations
C:/documents and settings/all users/boinc/

C:/program files/boinc

I believe you mean C:/documents and settings/all users/boinc/projects/setiathome.berkley.edu

And add the app_info from you or raistmer to my app_info file

Is this right?
I also need to close boinc before I edit my app_info file the restart boinc.
Also I need to change the <count>1</count> from 1 to .5 to use 2 wu per card

Yeah, the docs and settings folder is the data directory.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Okay, I'm running windows 7 64bit so it probably is in a different location from what you are running. Drop the three files from Raistmer in the same folder where your app_info file is located. After that, just put the fragment in your app_info file. If you want to try two at a time yes, you change the count to .5 but you also have to change the instance per device to 2.

---

PROUD MEMBER OF Team Starfire World BOINC

@perryjay and Mark

Thanks for the help
The instance per device to 2 where on the app_info file is that at on your example or on my app_info file.

---

@perryjay

I think I found the instances on Raistmers app_info
<cmdline>-instances_per_device 1 -hp -unroll 10</cmdline>
I need to change the 1 to a 2?

---

@perryjay and Mark
I followed all the instructions and all is well I do not have any ap cuda wu in my cache yet

thanks for all the help

---

Sorry Paul, I was away for a while. Yes, if you want to run more than one at a time on each GPU you have to change the count to .5 and the instance to 2 for two at a time or .33 and 3 if you want to try three at a time.

Just for information, I have mine set to .33 for both the AP app and the MB apps but the per device set at 2 on the AP app so that I can run either one AP and two MBs or two APs and one MB or three MBs. This way it might try to start a third AP but it won't go anywhere. It will just count up the elapsed time on the third AP without any work actually being done. When it finishes one of the other APs it will start up the third and elapsed time will reset to zero and start again.

---

PROUD MEMBER OF Team Starfire World BOINC

Paul, with MB CUDA you need to change only
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>

to

<coproc>
<type>CUDA</type>
<count>0.5</count>
</coproc>

for 2 MB WUs/GPU.

With AP OpenCL you need to change this upper mentioned entry and
<cmdline>-instances_per_device 2

for 2 AP WUs/GPU.

It looks like to now noone said, that you need to unpack the downloaded .RAR file with e.g. 7-Zip, then this 3 files to the project folder.

Paul's GTX460's have 738 MB VRAM. So the max would be 2 WUs/GPU (MB CUDA and AP OpenCL).
But, I don't know if this is enough for AP, e.g. if I let run 2 AP WUs on my GTX260 OC ~ 770 MB VRAM is used (with <cmdline>-instances_per_device 2 -hp -no_cpu_lock -unroll 10 -ffa_block 8192 -ffa_block_fetch 4096</cmdline>).
So maybe you should start with 1 WU/GPU for AP. And then look e.g. with help of GPU-Z how high the VRAM load is.

Your project settings should be:
Run only the selected applications SETI@home Enhanced: no
Astropulse v5: no
Astropulse v5.05: yes
If no work for selected applications is available, accept work from other applications? yes

Then your machine ask first for AP WUs, if no available you get MB WUs.
Then the chance to get AP WUs is higher.


- Best regards! - Sutaru Tsureku, team seti.international founder. - Optimize your PC for higher RAC. - SETI@home needs your help. -

Additional..

If you use Raistmer's app_info.xml entry for AP..

You should delete or change the

<flops>30987654321</flops>

entry.

If you use already flops, then you should take maybe first the half flops of your MB CUDA entry.

If this flops entry is too big, then you could get -177 errors.


- Best regards! - Sutaru Tsureku, team seti.international founder. - Optimize your PC for higher RAC. - SETI@home needs your help. -

---

Paul, with MB CUDA you need to change only
<coproc>
<type>CUDA</type>
<count>1</count>
</coproc>

to

<coproc>
<type>CUDA</type>
<count>0.5</count>
</coproc>

for 2 MB WUs/GPU.

With AP OpenCL you need to change this upper mentioned entry and
<cmdline>-instances_per_device 2

for 2 AP WUs/GPU.

It looks like to now noone said, that you need to unpack the downloaded .RAR file with e.g. 7-Zip, then this 3 files to the project folder.

Paul's GTX460's have 738 MB VRAM. So the max would be 2 WUs/GPU (MB CUDA and AP OpenCL).
But, I don't know if this is enough for AP, e.g. if I let run 2 AP WUs on my GTX260 OC ~ 770 MB VRAM is used (with <cmdline>-instances_per_device 2 -hp -no_cpu_lock -unroll 10 -ffa_block 8192 -ffa_block_fetch 4096</cmdline>).
So maybe you should start with 1 WU/GPU for AP. And then look e.g. with help of GPU-Z how high the VRAM load is.

Your project settings should be:
Run only the selected applications SETI@home Enhanced: no
Astropulse v5: no
Astropulse v5.05: yes
If no work for selected applications is available, accept work from other applications? yes

Then your machine ask first for AP WUs, if no available you get MB WUs.
Then the chance to get AP WUs is higher.


- Best regards! - Sutaru Tsureku, team seti.international founder. - Optimize your PC for higher RAC. - SETI@home needs your help. -

@Sutaru Tsureku
OK thanks I have made those changes
Bye Paul

---