@Paul:

Further above in this thread, Raistmer said:

6) 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.
Here is example of minimal cc_config.xml file you need

<cc_config>
<options>
<ncpus>enter_number_of_cores+1_value_here</ncpus>
</options>
</cc_config>

You should put it in BOINCdata folder.

Create a new file using Notepad using the lines above (copy and paste), replace "enter_number_of_cores+1" with the integer value applicable to your CPU (thus use 2 if single core, 3 if dual core, 5 if quad core), then in Notepad choose Save As with filename "cc_config.xml" ensuring Save as type is "All files" (so that you save it with .xml extension, not as .txt). Put that cc_config.xml file in your Boinc folder (default folder path is C:\Documents and Settings\All Users\Application Data\BOINC).

Post back if you have any other questions, or to tell us how it works for you.

Thanks I made the file like you said and I quite BOINC and
Put it in the data folder along with Raistmers's files and restarted BOINC and I got errors WU's? So I suspended all WU's so they will not go bad.
I guess I will have to remove the added files and reinstall BOINC. It trashed all my WU's and I can't get any more till tomorrow.

Thanks I made the file like you said and I quite BOINC and
Put it in the data folder along with Raistmers's files and restarted BOINC and I got errors WU's? So I suspended all WU's so they will not go bad.
I guess I will have to remove the added files and reinstall BOINC. It trashed all my WU's and I can't get any more till tomorrow.

Hi,
If I'm reading the part of what you said that I have in bold text correctly, you may have moved Raistmer's app to the wrong folder, which could explain why the WU's got trashed.
I believe for your OS Raistmer's app should go in the seti folder at C:\Documents and Settings\All Users\Application Data\BOINC\projects\setiathome.berkeley.edu
The cc_config.xml file you created goes in C:\Documents and Settings\All Users\Application Data\BOINC

---

I reverted from 6.6.2 to 6.6.0 (and Raistmer said 6.4.5 is OK too)

For V8 "team" builds even 5.45 should be fine. BOINC doesn't need to be aware of GPU. CUDA app does it's own GPU detection (as it does in all prev builds too BTW, BOINC's CUDA -awareness is needed only for correct scheduling - something that at laest BOINC 6.4.5 can't do anyway ;) ).

One more time (for record ;) ):

all files from *.rar archive should go in BOINCdata\projects\setiathome.berkeley.edu\
folder.
cc_config.xml should be created in BOINCdata\
folder.
If you use BOINC version prior 6.x.x - replace BOINCdata with BOINC.

(new AMD-related combo is ready to release, but >8 hours I can't get any MB task from project so still can't validate if it will run - release delayed...)

Question: Should this be able to feed my quaddie plus *both* halves of my GTX295 i.e. running 4 CPU cores + 2 GPU? It could just be my Vista x64 that's preventing it, but I would like to confirm whether or not the app should be capable of it?

BTW - kudos to all who contributed to this version. It's running like a dream on Boinc 6.6.0.

F.

---

Question: Should this be able to feed my quaddie plus *both* halves of my GTX295 i.e. running 4 CPU cores + 2 GPU? It could just be my Vista x64 that's preventing it, but I would like to confirm whether or not the app should be capable of it?

BTW - kudos to all who contributed to this version. It's running like a dream on Boinc 6.6.0.

F.

No, as I said only first GPU will be used.

Question: Should this be able to feed my quaddie plus *both* halves of my GTX295 i.e. running 4 CPU cores + 2 GPU? It could just be my Vista x64 that's preventing it, but I would like to confirm whether or not the app should be capable of it?

BTW - kudos to all who contributed to this version. It's running like a dream on Boinc 6.6.0.

F.

No, as I said only first GPU will be used.

Ahh. I missed that caveat. Thanks for the confirmation - and thanks for all your work on this.

F.

---

New combo for AMD CPUs available.

---

"The most exciting phrase to hear in science, the one that heralds new discoveries, is not Eureka! (I found it!) but rather, 'hmm... that's funny...'" -- Isaac Asimov

But if I want to elaborate Seti only with my GPU and not with CPU?

---

But if I want to elaborate Seti only with my GPU and not with CPU?

LoL, so don't use this mod. Pretty simple.

Hi
Many thanks for the good work, Raistmer!

No, as I said only first GPU will be used.

Is that also when you have 2 separate graphics cards with V8a?
2 of my quads has dual gr.cards and both (with ncpu=7) are running:
1 GPU at 100% + 2 CPU at 100% + 4 CPU at 50% (with ncpu=5 all 4 CPU are 100%)
(In another tread someone said that connecting 2 monitors could solve the problem but it didn't)

They are also not doing any AP's. The ones I had when I upgraded to V8a are still idle after 2 days.
How can I get them working without going back to earlier version?

(Earlier version was working very good with both GPU crunching Cuda and 4 CPU-core was doing AP'S.)

I'm using 181.22 and
Q9400 / 9500GT (2x) /v6.4.5 /XP-64b (can not get 6.6.2 working here)
Q6600 / 8400GS (2x) /v6.6.2 /XP-32b

(I hope the information is not mentioned earlier (if so I apologize) I am trying to read all information but as my first language is not english it sometimes is hard to understand everything.)

Thanks!

---

//Peter

But if I want to elaborate Seti only with my GPU and not with CPU?

LoL, so don't use this mod. Pretty simple.

Ok
The V7 vlar kill is the latest package for gpu only?



P.S. sorry for my english.....

---

No, as I said only first GPU will be used.

Is that also when you have 2 separate graphics cards with V8a?

Because this question become common here and on Lunatics forum I will give more extended answer.

This combo relies on CUDA MB ability to detect GPU with CUDA support.
It will use first found. So it will not use second and so on. Moreover, if one CUDA MB app is launched there will be no attempts to launch second one until first exits. So you could treat this as "by design".

But things not so bad you one could think. There are V7x of my combos that rely on BOINC GPU scheduling ability - they can use both GPU cards and all CPU cores provided you use SETI AP opt app too.

That is, summary:
V8x combos: Can do SETI MB on CPU and GPU but can use only one GPU card.
V7x combos: Can do SETI MB on GPU only (but can use CPU for SETI AP ) but can use all GPUs in system.

For all dual/many GPU owners I recommend to use V7x packages, not V8x ones. That way you will get best performance from your hosts.

But if I want to elaborate Seti only with my GPU and not with CPU?

LoL, so don't use this mod. Pretty simple.

The V7 vlar kill is the latest package for gpu only?

Yes. You can use VLAR kill, or VLAR process correctly ;) modifications - both available.

That is, summary:
V8x combos: Can do SETI MB on CPU and GPU but can use only one GPU card.
V7x combos: Can do SETI MB on GPU only (but can use CPU for SETI AP ) but can use all GPUs in system.

For all dual/many GPU owners I recommend to use V7x packages, not V8x ones. That way you will get best performance from your hosts.

Many thanks for your answer!
Have just installed BOINC 6.6.3 and it installed without the problems 6.6.2 had and is working very good on 4 quad's and 1 i7 with your V8.
But I reverted back to v.7 (tried both versions) on the multi graphics PC but they crashed Boinc Manager so I had to install V8 again to get it running again.

I have 181.22 driver.

Any suggestion?

---

//Peter

Have just installed BOINC 6.6.3 and it installed without the problems 6.6.2 had and is working very good on 4 quad's and 1 i7 with your V8.
But I reverted back to v.7 (tried both versions) on the multi graphics PC but they crashed Boinc Manager so I had to install V8 again to get it running again.

I have 181.22 driver.

Tried the same on another PC with dual cards and it did not crash Boink but I can not see any information in "function"-tab. (I can see in "message"-tab that it started 4 AP and 2 Cuda)
This one is a Q6600 with dual 8400GS (XP 64bit)
The one who crashed is Q6600 with dual 9500GT (XP 32bit)
both cards is Gigabyte and mobo is Gigabyte GA-X48

Glad for any suggestions

---

//Peter

I didn't use new BOINC versions so have no idea why science app can crash BOINC manager - this question should be forwarded to BOINC supporters or maybe someone who tried new BOINC could give some ideas.

I didn't use new BOINC versions so have no idea why science app can crash BOINC manager - this question should be forwarded to BOINC supporters or maybe someone who tried new BOINC could give some ideas.

Same thing happend in 6.6.2 but not in 6.6.0 and only on 2 of 5 PC

I'll try your suggestion.

Thanks

---

//Peter

Not sure if this has been posted before, but this looks like a good place to log it.

Installed V8a-SSE3-AMD package on my AMD X2 system.
Running XP Pro SP3, 181.20 drivers.
5% cpu OC but no memory or GPU OC.
BOINC 6.4.5.
I have ncpus set to 3.
gpu is 9400GT.
Running SETI MB and Einstein...no AP.

I've backed it off due to the following symptoms I've found and am running stock 6.08 CUDA and E@H only for now.

1. If running multiple MB WUs (gpu + 1 or more CPU) and BOINC is shutdown/restarted, frequently one of the CPU WUs runs on the GPU after restart instead of the WU that earlier ran on gpu. They appear to validate correctly, so nothing bad here...yet.

2. Due to resource imbalance (probably STD, not LTD), E@H will crowd out all the MB WUs and run 3 instances at the same time. This causes severe cpu contention and the gpu is idle. I tried cutting down ncpus value to 2 and the third E@H WU suspended, but MB CUDA would not restart. This is unacceptable.

The lesson from this: you can only run MB or MB+AP(?) on a system set up this way.