Okay I'm stumped.

I have one machine with 2x 4850x2 in it and I can get only 1 or at most 2 units running on the optimized app at a time. And both run on Device 0.

I have checked the app_info and there is nothing weird in it. I have copied an app info over from another machine running 8 at a time (0.5 on the count x 4 GPU)

I have created a cc_config.xml and set the use all coproc flag.

I have gone from 6.10.18 to 6.10.19 in case it was an issue...

I have reset the project.
I have detached and re-attached the project.

I am now looking for a goat to sacrifice and a calender to show the correct phase of the moon to do it on.

Are you running a Dummy Plug ??? Even though a 4850 X2 is one Card it needs a Dummy Plug for CCC to see it and to Activate it as a Default VGA Monitor ... I have 3 Box's with 2 4850 X2's installed and have to run 3 Dummy Plugs to run all 4 GPU Cores ... You may have done this already but thought I'd bring it up ...

Thanks for the replies.

Sorry I forgot to put that yes all 4 GPUs are shown in BOINC in the status window with there RV type and CAL versions.

Aye Poorboy your not wrong and yes I have 1 monitor and 3 Dummy Plugs in. I can swap the monitor around and the plugs and I still get one running. If I remove a dummy plug the GPU disappears from CCC. When I put the plug back in it reappears. I was trying that last night. Kudos for the great reply though :)

Okay time to hang my head in shame or to use a local saying 'I'm awa tae boil ma heid'.

I noticed that I also did not have a stock of units cached for other projects either. I checked my preferences and sure enough number of days to cache was 0. I set it to the max of 10 and suddenly downloading 47 Milkyway tasks.

So it seems if that setting is 0, and this was a new install on a new OS it defaults to 0, it doesn't even download enough units to keep the GPUs busy....

A quick simple question I'm sure has been asked before, but I can't find it looking through the forums.

My son just upgraded his video card and is sending me his old Radeon 4850. Will the project automatically send me the right app or do I need to use app_info etc.

If so which version should I download? I have an Intel E7200 @ 3.16 Ghz and of course the ATI 4850. Running Windows Vista Ultimate 64bit. I'm just confused between the AMD and ATI versions 0.22 and 0.23.

And as a matter of interest what kind of RAC would the 4850 be capable of?

Thanks for your time.

James

A 4850 should be capable of around 75k or so depending on different factors.

It should send you the correct app if you connect to the project. It would be the ATI app as you are running a newer version of the driver.

Though since you are already running a "optimized app" you would need to either remove the app_info or install the 0.23 ATI app.

---

Well its up and running doing about 220 to 230 seconds per unit. Now if I could just work out how to run gpu units and optimized cpu units at the same time.

Interestingly enough when I run a project on the cpu, times for each workunit increase from 223 seconds up to 240 - 250 seconds. Which is a loss of around 7k points per day. Now I don't know of any cpu project that my processor could get more than 7k per day on so I actually lose points by running another project on the cpu. Any body else see this kind of behavior.

Interestingly enough when I run a project on the cpu, times for each workunit increase from 223 seconds up to 240 - 250 seconds. Which is a loss of around 7k points per day. Now I don't know of any cpu project that my processor could get more than 7k per day on so I actually lose points by running another project on the cpu. Any body else see this kind of behavior.

Depends which CPU project.

With Aqua I see my 4870 doing ~ 203 seconds, but with PrimeGrid it drops to 191 - 193 seconds.

....... Now I don't know of any cpu project that my processor could get more than 7k per day ......

You should do more than that at Aqua by running the FP WU on CPU, worth a try ... there is a new version of Aqua FP running (V2.04) which is now stable. If it does not, it will still be the best pay back for a cpu.

Regards
Zy

It's look like that for this project optimized apps are not longer available. Or in other words, stock apps are the fastest.
I'm right?

I have an Intel Core2 Duo E7600 @ 2 x 3.06 GHz with GTX260-216 and BOINC DLed the stock 0.19 (CPU) and 0.24 (GPU).
The GPU WUs need ~ 10 mins, on the CPU I guess it'll last ~ 10 hours.
The value for the CPU is O.K.?

If I compare with SETI@home the ratio GPU/CPU, the time of the WUs on CPU should be ~ 2 hours (with optimized app) and ~ 4 hours (with stock app) here.

Is the CPU app of this project not well optimized?


Thanks! :-)

---

CPU 3rd party apps are well optimised. Links here.

So I need to make my own app_info.xml file.. Ohh..
I could find somewhere examples?

There are SSE, SSE2, SSE3, x87 apps available, my E7600 have up to SSE4.1 .
After a short Inet search x87 is something between SSE and SSE2?
Or what mean x87?


Thanks! :-)


[EDIT: After DLing of the SSE3 app, I have an app_info.xml file with only CPU entry. How I could add the stock CUDA app?]

---

Use CPU-Z to see your processor capabilites (like it says on the web site), down load the appropriate files and unzip, then read the readme file that is included and install the files - just like it says.

I have edited my upper message..


If I compare with SETI@home (to now I have only experiences with SETI@home ;-) , I need to insert all apps in the app_info.xml file.
So if I don't edit the app_info.xml file, I will trash all DL WUs for the GPU and the PC will not crunch on GPU.

It's the same here, or?

So I need to add the GPU app to the app_info.xml file.
For not to trash the DLed WUs, it would be nice to see an example how to do it.


Thanks! :-)


[EDIT: WinXP Home 32bit, astronomy_0.21_SSE3.exe with milkyway_0.24_windows_intelx86__cuda23.exe]

---

O.K., I made my own app_info.xml and didn't trashed WUs. :-D

I use the following and it work well..

--------------------------------------------------

<app_info> 
    <app>
        <name>milkyway</name>
    </app>
    <file_info>
        <name>astronomy_0.21_SSE3.exe</name>
        <executable/>
    </file_info>
    <app_version>
        <app_name>milkyway</app_name>
        <version_num>19</version_num>
        <file_ref>
           <file_name>astronomy_0.21_SSE3.exe</file_name>
            <main_program/>
        </file_ref>
    </app_version>
    <app>
        <name>milkyway</name>
    </app>
    <file_info>
        <name>milkyway_0.24_windows_intelx86__cuda23.exe</name>
        <executable/>
    </file_info>
    <file_info>
        <name>cudart.dll</name>
        <executable/>
    </file_info>
    <file_info>
        <name>cutil32.dll</name>
        <executable/>
    </file_info>
    <app_version>
        <app_name>milkyway</app_name>
        <version_num>24</version_num>
        <plan_class>cuda23</plan_class>
        <avg_ncpus>0.050000</avg_ncpus>
        <coproc>
            <type>CUDA</type>
            <count>1</count>
        </coproc>
        <file_ref>
            <file_name>milkyway_0.24_windows_intelx86__cuda23.exe</file_name>
            <main_program/>
         </file_ref>
        <file_ref>
	       <file_name>cudart.dll</file_name>
        </file_ref>
        <file_ref>
            <file_name>cutil32.dll</file_name>
        </file_ref>
     </app_version>
</app_info>

--------------------------------------------------

But, I'm little bit confused.
All stock, BOINC DLed automatically CUDA V2.2 and not CUDA V2.3.
It was DLed:
cudart.dll (V2.2)
cutil32.dll

I took cudart.dll (V2.3) from SETI@home and it work.
But, what is cutil32.dll? I don't know this. There are also different versions available? So if, where I could get the 2.3 version?


Thanks! :-)

---

Does anybody have any news about whether or not there will be opti apps developed for CPU machines running the new MW@H .44/.45 apps and/or the N-Body apps?

Thanks, Wes

---

I've just come across an optimised CPU app (astronomy_0.21_SSE3.exe) which i miss filed about a year ago and thus never installed after i upgraded my processor.

Can i still use this?

If yes then my milkyway project folder contains:

milkyway_0.45_windows_intelx86.exe
milkyway_0.50_windows_intelx86_sse2.exe
milkyway_nbody_0.04_windows_intelx86.exe
milkyway_nbody_0.21_windows_intelx86_see2.exe

but the accompanying app_info only references the astronomy_0.21_SSE3 app and the _nbody_0.04 executable.

CPU-Z shows my processor as MMX(+) 3DNow!(+) SSE(1,2,3,4A) x86-64 AMD-V

1. I'm pretty sure it uses the nbody_0.04, but what do i keep, or would a reset project be a better idea.

2. after (1.) is sorted do i just update the app_info?