Hi,

my installation (xp32 sp3, 4850, Catalyst 9.9, Q9300) works, but produces calculation errors in pairs ( i use setting count 0.50, f22 ). I canged to these settings, because with original settings my system hangs ten times a day, mostly in combination with a hp-toolbox hangup.
When my system crashes, the screen gets blank and switches to stdby. I can acces my system via remoteconsole, which also brings up my screen again. Calculation of milkyway has stopped. If i stop boinc and restart it again, everthing works fine.
Currently my system produces 50% calculation errors.
Has anyone a good idea?

Regards,
Alexander

from one of your wu's:

<core_client_version>6.10.4</core_client_version>
<![CDATA[
<message>
Unzulässige Funktion. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
Running Milkyway@home ATI GPU application version 0.20 (Win32, SSE2) by Gipsel
setting minimum kernel frequency to 30 Hz
ignoring unknown input argument in app_info.xml: --device
ignoring unknown input argument in app_info.xml: 0
Couldn't find input file [astronomy_parameters.txt] to read astronomy parameters.
APP: error reading astronomy parameters: 1

</stderr_txt>


One thought is when you switched to v.20 did you do the changes the readme file says to do?

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

from one of your wu's:

<core_client_version>6.10.4</core_client_version>

6.10.4 has a known bug causing the errored out WUs (but actually doesn't explain the frequent driver crashes). You should update to 6.10.9 to get rid of the errors where the input files couldn't be read.

If you have stability problems try setting the count to 1, it should improve the stability slightly (but don't remove the actual reason for the crash). If it is still to bad, think about downgrading to Cat8.12. This is the most stable driver release for WinXP.

Hello,

highly effective!
I've upgraded to 6.10.11, i've downgraded to 8.12, changed my settigs to run only one wu at a time and it seems to run stable, at least for the last 2 hours. I will do some experiments with the timing, but anyway, i have a stable system back!

Thanks a lot!

Alexander

My 1st post. I've recently joined as I now have a system with an ATI card in it. I have been getting a few VDU errors with the driver restarting. My setup:

HD4850 card (1Gb)
BOINC 6.10.11
Cat 9.8
Running 0.20 with a single wu.
Win XP (32 bit)

Did the copy of the ATI*.dll's as advised in the readme. Is this still required with BOINC 6.10.11?

I have tried running 2 and 3 wu at a time (changed <coproc> in app_info to 0.5 or 0.33) but they seem to take the same time as doing them one at a time.

My VDU errors usually occur when I click on one of the tabs in BOINC and then it does the popup about reporting the error. Only happens if I have some MW work units running at the same time on GPU. The wu running at the time will proceed up to 100% and then gets a computation error. Do I need to put an entry in the <cmdline> in app_info?

Any help appreciated.
Thanks

---

BOINC blog

Try rolling back to 9.2 or forward to the latest WHQL version

Try rolling back to 9.2 or forward to the latest WHQL version

Well I tried catalyst's 9.1, 9.2 (previously had 9.8 installed from shop) and now on 9.9 (latest and greatest). The 9.9 seems to be a bit more stable as I have only killed the driver once, and despite trying it seems to keep working now.

Seeing as I am in experimental s/w mode I guess I might try BOINC 6.10.13 as well :-)

---

BOINC blog

Try rolling back to 9.2 or forward to the latest WHQL version

Well I tried catalyst's 9.1, 9.2 (previously had 9.8 installed from shop) and now on 9.9 (latest and greatest). The 9.9 seems to be a bit more stable as I have only killed the driver once, and despite trying it seems to keep working now.

Seeing as I am in experimental s/w mode I guess I might try BOINC 6.10.13 as well :-)

I take it back!

I managed to crash the driver about 5 times after I posted this. I will try 8.12 which I saw recommended in another thread. Oh and the later BOINC version made no difference.

---

BOINC blog

I managed to crash the driver about 5 times after I posted this. I will try 8.12 which I saw recommended in another thread. Oh and the later BOINC version made no difference.

Well 8.12 seems to be much better so far. I am also running two wu at a time and nothing for <cmdline>. Current setup is an i7-920 under WinXP. HD4850 at stock speed, BOINC 6.10.13 and Catalyst 8.12 (and none of the ATI add-ons).

I've had it running for a couple of days without any issues yet (touch wood). Its also mentioned in the readme file about Cat 8.12 :-)

---

BOINC blog

Hi,

i can tell you, milkyway 0.20 is working fine with catalyst 8.12, boinc 6.10.11. I had a lot of troubles like anyone here, downgraded catalyst and upgraded boinc, it is running now for more than 2 weeks without any trouble, crunching two wu's at a time.

One hour ago i tried the new 0.20b application. Sorry, not even one wu created results, all of them ended up with errors, right after starting them. Application is still reported as milkyway 0.20, without b.

Has anyone experience with that?


XPpro, SP3, Q9300

Yeah, I tried the new 20b version and it resulted in lots of errors. I haven't got a stable system running on any version though. I have xp32 sp2, ati 4870. And I have tried every Cat 9.9 9.5 9.1 and 8.12. Still not working.

Is there a difference in MW progs? Should I try older 0.19 series?

I normally like to try new versions reasonably quickly, but in the case of 0.20b I decided to wait until it was complete or help was requested in testing it.

As far as I am aware version 0.20 remains the recommended version and the last message regarding the use of 0.20b stated the following:

....."But before you try to download it, it's not complete and not correctly set up yet. So it will probably not work right now. Be a little bit more patient"...... from Message 32135

I normally like to try new versions reasonably quickly, but in the case of 0.20b I decided to wait until it was complete or help was requested in testing it.

As far as I am aware version 0.20 remains the recommended version and the last message regarding the use of 0.20b stated the following:

....."But before you try to download it, it's not complete and not correctly set up yet. So it will probably not work right now. Be a little bit more patient"...... from Message 32135

I must admit I wasn't sure about the status of 0.25b, so I just went ahead and installed it on all my PCs with ATIs, and then proceded to find every one of them failing all WUs with compute errors. I usually test all win_32 op apps before putting them up on www.brilliantsite.com, which I did in this case and 0.20b won't be going up there just yet.

---

One hour ago i tried the new 0.20b application. Sorry, not even one wu created results, all of them ended up with errors, right after starting them. Application is still reported as milkyway 0.20, without b.

If it doesn't report being 0.20b it simply is not 0.20b. Are you sure you downloaded the correct version? Could you post the task details of a failed WU?

Yeah, I tried the new 20b version and it resulted in lots of errors. I haven't got a stable system running on any version though. I have xp32 sp2, ati 4870. And I have tried every Cat 9.9 9.5 9.1 and 8.12. Still not working.

Is there a difference in MW progs? Should I try older 0.19 series?

Yes, of course there there are differences between the MW versions. And no, you shouldn't try v0.19.

You are running WindowsXP, therefore Catalyst 8.12 is recommended. That also means you have to use the Win32_amd or Win32_SSE2_amd version (not the ones with "_ati" at the end). That is all said in the readme by the way.

As far as I am aware version 0.20 remains the recommended version and the last message regarding the use of 0.20b stated the following:

....."But before you try to download it, it's not complete and not correctly set up yet. So it will probably not work right now. Be a little bit more patient"...... from Message 32135

Read that message again! My comment pertained to the ATI version which can be found on the MW server. That is not correctly set up yet.

Maybe all the GPU download links should link to the readme instead, with a link to the actual download only in the readme itself <.<

Read that message again! My comment pertained to the ATI version which can be found on the MW server. That is not correctly set up yet.

My apologies, I misunderstood your message. After reading it again I think you meant the ATI version that will be downloaded automatically by a CAL version aware BOINC version is not yet able to be automatically downloaded.

Even though I misinterpreted the first part, your message still implied to me that most people should be patient and not download 0.20b manually yet. Only those who "can't wait" and are willing to read the readme and are able to understand the correct version to choose should manually download 0.20b.

Because some of the discussion about using different application versions to suit different drivers happened on the Collatz forum and not here, those who have not read about it at Collatz may not be fully aware of what "different versions" means. It appears that some here do not understand about the different "amd" and "ati" versions to suit different drivers and do not realise they have installed the incorrect 0.20b version and that is why they are getting errors.

The readme is getting longer and sometimes people skim quickly without fully understanding or may be impatient to get crunching again and do not read to the end. Not everyone thinks logically, follows instructions well or can understand things easily, there can sometimes be language difficulties as well. Therefore perhaps some have not read or understood the section of the 0.20b readme that includes:

"For Catalyst 8.12 and 9.1 use the application variants with "_amd.exe" at the end, for newer drivers simply use the files ending with "_ati"."

One hour ago i tried the new 0.20b application. Sorry, not even one wu created results, all of them ended up with errors, right after starting them. Application is still reported as milkyway 0.20, without b.

If it doesn't report being 0.20b it simply is not 0.20b. Are you sure you downloaded the correct version? Could you post the task details of a failed WU?

Yes, i'm shure. But anyway, other posts say, this application is faulty and it has been removed from the brilliantsite.com.
Let's wait for the next version 0.25 ...

But anyway, other posts say, this application is faulty and it has been removed from the brilliantsite.com.
Let's wait for the next version 0.25 ...

No, it's not faulty, actually some issues where fixed. I'm quite sure you simply used the wrong version ;)

But nobody is forcing you to use 0.20b of course. Maybe I should wait someone builds v0.25 ;)