So, you both see different temperature regime between 3+1 and 4+1 modes while using my mod? Interesting, will check it on my host too.

after more testing i found what is the source of the temp dif there this line in the file app_info.xml
<avg_ncpus>0.040000</avg_ncpus>
<max_ncpus>0.040000</max_ncpus>
meaning the gpu feding at any one time can can use 4% cpu time and over all can only use 4% the cpus time so by changing this
<max_ncpus>0.040000</max_ncpus>
to this
<max_ncpus>0.080000</max_ncpus>
it allows the gpu feeding to bust the data when needed the same cpu time is used just all at once and not broken. this speeds up the gpu task's that is all that is need to be done

<max_ncpus>0.080000</max_ncpus>
it allows the gpu feeding to bust the data when needed the same cpu time is used just all at once and not broken. this speeds up the gpu task's that is all that is need to be done

Done,thanks for the tip.

Just in time too, now running AP and CUDA MB on the AMD, with AP switching with Einstein.

Going well so far.

---

yes, this option defines CPU share but I thought it's only for BOINC scheduling mechanism, not for OS that really does process scheduling (grant CPU cycles to app).
Will do more tests.
For now I discover that at least on my card with big cooler temperature regime is NOT stationary.
Host doing shorty tasks now ~6'50" on each task.
When it finishes one task and starts another temp drops to 51C (44 in idle - for reference).
Then temp increases to 57C and variates in 1-2 C range while task is processed. Then next drop to 51-51C and so on.
So, speaking anout GPU temp lets speak about its peak temperature and ensure that it's peak indeed.

Will try other configs and app_info.xml changes now.

Just installed your mod and now enjoy 10x increase of speed of computing. Thanks a lot %)

But it looks like now I still don't use all cores at 100% efficacy.

PC has 2 cores + GPU (8800).
I have AP and MB units running together. Judging by Task manager AP creates about 45-50% of CPU workload and MB (crunched mostly at GPU) - about 5-10%. So total CPU workload is about 60%. Are there any ways to utilize 100% by parallel computing of one more WU?

Just installed your mod and now enjoy 10x increase of speed of computing. Thanks a lot %)

But it looks like now I still don't use all cores at 100% efficacy.

PC has 2 cores + GPU (8800).
I have AP and MB units running together. Judging by Task manager AP creates about 45-50% of CPU workload and MB (crunched mostly at GPU) - about 5-10%. So total CPU workload is about 60%. Are there any ways to utilize 100% by parallel computing of one more WU?

Try this

---

Just installed your mod and now enjoy 10x increase of speed of computing. Thanks a lot %)

But it looks like now I still don't use all cores at 100% efficacy.

PC has 2 cores + GPU (8800).
I have AP and MB units running together. Judging by Task manager AP creates about 45-50% of CPU workload and MB (crunched mostly at GPU) - about 5-10%. So total CPU workload is about 60%. Are there any ways to utilize 100% by parallel computing of one more WU?

You need to read my first message completely.
Statement "7)"
7) For best CPU and GPU usage I recommend to set number of processors available for BOINC to real_number_of_cores+1. This will mitigate current BOINC bug with CPU+CUDA scheduling and will allow fully load CPU and GPU.
Did you make it?

And about 3+1 config or app_info.xml changes. I tried both, but no GPU peak temp increase was detected. So it seems it doesn't work for my host.
But changeing app_info.xml could still be useful for another hosts.
On my host CUDA app CPU % is about 3% so it fits in original 4% constrains well.

2 Byron and Raistmer:
Yes, I set ncpus to 3 in cc_config.

2 Byron and Raistmer:
Yes, I set ncpus to 3 in cc_config.

So now you should see 2 AP instead of 1 AP (each ~50% CPU) and CUDA MB (some small CPU % )

Ya once i get two 8600gts back from a rma i will have 4gpu's in this computer. Well that be two much for the cpus cores to handle. where as I should put them in a different computer.

The 8600gts's take about 1-3% of the cpu while a higher end card like a gtx 260 uses from 3-9%. So yes once core should be enough.

Also for FYI... I have not done any modifications and boinc is running 4+2.
It just takes it some time to realize its not as high priority as it thinks :)

~Bob

---

Do you Good Search for Seti@Home? http://www.goodsearch.com/?charityid=888957
Or Good Shop? http://www.goodshop.com/?charityid=888957

So now you should see 2 AP instead of 1 AP (each ~50% CPU) and CUDA MB (some small CPU % )

No, I still see only one AP unit and one cuda-MB. Also I see "Your video driver failed and was restarted successfully" each time when I try to open new window while computation is going - and this error leads to failure of computation of current MB unit.

Let's wait till this AP unit is completed... it was started before applications upgrade, maybe that is the reason...

Ya once i get two 8600gts back from a rma i will have 4gpu's in this computer. Well that be two much for the cpus cores to handle. where as I should put them in a different computer.

The 8600gts's take about 1-3% of the cpu while a higher end card like a gtx 260 uses from 3-9%. So yes once core should be enough.

Also for FYI... I have not done any modifications and boinc is running 4+2.
It just takes it some time to realize its not as high priority as it thinks :)

~Bob

Yes, but you still priority mod to feed GPU fully in 4+2 config.

So now you should see 2 AP instead of 1 AP (each ~50% CPU) and CUDA MB (some small CPU % )

No, I still see only one AP unit and one cuda-MB. Also I see "Your video driver failed and was restarted successfully" each time when I try to open new window while computation is going - and this error leads to failure of computation of current MB unit.

Let's wait till this AP unit is completed... it was started before applications upgrade, maybe that is the reason...

Strange... with ncpus==3 you should see 3 tasks running no matter CPU or GPU but at least 3 tasks (maybe even 4 - 3 CPU +1 GPU)....

Yes, but you still priority mod to feed GPU fully in 4+2 config.

Yes I am running priority mod but I was referring to no mods to Boinc for running 4+2.

~Bob

---

Do you Good Search for Seti@Home? http://www.goodsearch.com/?charityid=888957
Or Good Shop? http://www.goodshop.com/?charityid=888957

Hello Raistmer,

thank you for the modified application.

My first four workunits, which were still in the cache when I did the change to your application, were all failing with calculation error and video driver crash (blank screen and driver restartet by OS).

But after checking the results I guess it's the VLAR issue and not related to your modified app, right?

WU true angle range is : 0.009700
WU true angle range is : 0.011063
WU true angle range is : 0.009700
WU true angle range is : 0.010032

I'll try to get some new workunits, hopefully non-VLAR ones.

BTW, please allow me a silly question, just for clarification: your modified application is based on version 6.05 from the main project, not on version 6.06 from Beta, right?

Best regards
Alex

EDIT (some info 'bout the system):
CPU AMD Athlon 64 X2 6000+ dualcore (non-OC)
GPU GTX260-216 ~1.400 MHz (factory-OC)
driver: Nvidia Forceware 180.48 64bit (stock driver)
BOINC: 6.4.5 64bit w/ cc_config tweak (ncpus=3)
OS: Win Vista Home Premium 64bit

But after checking the results I guess it's the VLAR issue and not related to your modified app, right?

Right :)

BTW, please allow me a silly question, just for clarification: your modified application is based on version 6.05 from the main project, not on version 6.06 from Beta, right?

Best regards
Alex

No, not 6.06 or 6.05. I recived sources from Eric and after compilation got exe file with 6.04 in name.
Does it mean that sources slightly outdated or does it mean it's just cosmetic flaw (I also very often forget to increase build number by hands) - I don't know. Some sources available now, I will compare that codebase with my own. If more recent source is available I'm surely rebuild and repack. It will be announced here and on Lunatics of course.

OK, at last I fixed issue with Nr of CPUs... it was my mistake - I put cc_config.xml into BOINC directory, but it should be located in BOINC Data...

But almost gone crazy with video drivers failures. At least half of WUs were spoiled. Also I have a feeling that now these errors happen even when BOINC is not launched or computation is suspended. What is this damned "VLAR issue" and are there any ways to avoid it?

Hm, it's some method but I don't think you like it. You need to check all incoming tasks headers and manually abort VeryLowAngleRange ones....
Too much manual work to be effective unless you write some script for it.
If you familiar with Perl it will be not so hard IMO.

Try to install another driver instead....

Hmm... I'm just a physician interested in PCs and programming %) I'm afraid my knowledge of Perl is not enough to deal with such task %)

What other drivers can you recommend? Currently I ran standard 180.48 package. It looks like this is the only version that really works with my card - I also tried CUDA 2.1 and CUDA 2.0 downloaded from CUDA's page but with them BOINC doesn't identify my card as CUDA-capable.