I don't know how the 2 Box's I converted to 0.20 App & 6.10.3 Client ran over night but since I've been up this morning they have run like Crap.

I just watched the i7 run out of work for the Umpteenth time before actually getting more work again. It's been calling for work all along but the Server won't send it until it runs out.

The 4850 X2 Box I converted doesn't hardly run at all, most of the time it just sits there with Waiting To Run WU's. It'll run a few WU's then immediately go back into a Waiting To Run State. Probably just go back to 0.19 again as this just ain't working at all from what I can see anyway.

I don't care how fast the New 0.20 App runs the Wu's, if the WU's aren't running 75% of the time then it don't matter how fast it'll run them ... IMO

So I did the MSVC d-l/install/reboot, [wasn't part of the advertising for win '95 that we'd never again have to reboot after installing stuff?], v2.0 install, coproc thing [on 3 boxes] and restarted boinc. all running now, too early to see amazing speed improvements. :)

Note to Cluster - start your next readme with:
'Before you start this make sure you have handy a hammer, a screwdriver, a copy of vcredist_x86.exe, etc.' And then start talking about what we have to do with them.

Anyway, its working well, thanks again.

---

Cheers,

PeterV

.

If I install the .20 CPU app, will I be able to run the stock Cuda app along with it? The opp app stopped me from downloading Cuda WUs b4.
;-p

For those that might want to run CUDA and CPU boinc wont let you download Cuda if CPU opp app is installed. At least thats my experience so far. So with the credit reduction and with the new opp app,they work out almost even if I use all 8 cores for MW. Decisions, decisions???
;-p

---

as you can see from the previous thumbnail, I'm seeing strange behaviour from milkyway 0.20 app. Only one workunit runs at a time and many starts but not finish. As you can also see wall time is a bit high for having done only a fraction of the work of those work-units... What's going on? I must say that when boinc resumes those work-units it starts from 0% and so reported time is always under 60s (ATi 4770)

14.09.2009 15:28:56 cuInit() returned 100
14.09.2009 15:28:56 ATI GPU 0: RV770 (CAL version 1.4.283, 512MB, 1064GFLOPS)

but all WU error
Incorrect function. (0x1) - exit code 1 (0x1)

http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=104481744

6.10.4 has a small problem where it tries to run work in last in first out method and does the waiting to run thing.

Try going back to 6.10.3 instead.

You can adjust how many run in the app_info <coproc> area.

---

Decided to give 0.20 another try after seeing Verstapps post above. So I d/l vcredist, install, reboot and try again. Same result. So then I installed BOINC 6.10.3. Still getting the same results, all WUs error out with the 'output file absent'. I guess it's back to 0.19g for me.

I went back to Client 6.6.20 & App 19g on my i7, I'm making progress I think, I went from Waiting To Run to Ready To Start ... LOL

It is working and fast but everything is working up to par.
Boinc downloads 48 task but only does the bottom 2 then get new ones so it stays at 48. The 3rd from the bottom start when the other 2 are reporting but never completes because when the new task are downloaded the bottom one start and the 3rd from the bottom restarts each time. I have set Milkway to no new task and it worked from the bottom up and every 4th task ran 1 second and got a error. I let them all complete then change it back to allow new task then it reported all the task and downloaded 48 more repeating what it did before. Also the status of the downloaded the 1st is ready to start and the 2nd is waiting to run(0.05CPUs + 1.00 ATI GPUs) and this repeats for all 48.
I will force the task to complete before the expire.

I was having the same problem. As soon as a new Wu was downloaded it would start crunching it and leave the unfinished one sitting there. When it restarts the unfinished Wu it restarts at 0%, wasting time...

Installed 6.6.36 over 6.10.4 and removed the
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>

from both places in App_info.xml.
When restarting the new boinc install all three machines errored out. Had to reinstall Boinc6.6.36 using the repair setting.

Working good so far for me. Runs three at a time now. Also no more errors.


Vista64, HD4870, Boinc 6.6.36(64) ,Milkyway 0.20ati(64)

also on two other machines

Winxp(32), HD4850, Boinc 6.6.36(32),Milkyway 0.20ati(32)

http://milkyway.cs.rpi.edu/milkyway/hosts_user.php?userid=1258

---

Just updated from the ATI 0.19f to 0.20 on a 4850 card and a 3850 card, using Boinc 6.4.7 and catalyst 8.12 driver. Results below.

0.19f

4850

74 secs - 53.46 credit
57 secs - 39.98 credit

3850

210 secs - 53.46 credit
175 secs - 39.98 credit

0.20

4850

51 secs - 53.46 credit
39 secs - 39.98 credit

3850

165 secs - 53.46 credit
113 secs = 39.98 credit

As you can see the performance increase of the new 0.20 version has totally wiped out the last 30% credit reduction. Back to where we were! Yehar! Now to update the CPU machines .....

---

Don't drink water, that's the stuff that rusts pipes

6.10.4 has a small problem where it tries to run work in last in first out method and does the waiting to run thing.

Try going back to 6.10.3 instead.

You can adjust how many run in the app_info <coproc> area.

Thanks a lot
I remember of having found a bug in 6.10.3 which was causing me some problems... I think I'll wait for 6.10.5. Reducing the queue, the problem goes away and, at the end, i'm not putting too much strain on the server because I do not work really much on milkyway!

What about the coproc area? Should I put count on 0.33 if I'm understanding well so that I could have 3 wu at the same time?

Back to where we were!

Not for all hosts and only from credit-hunter point of view.
Project got very nice speedup so we could dream to enjoy 3D Milkyway map much earlier than before ;)

This does not work for me at all.

I've copied the relevant opt files, made sure that MSVC++ 2005 SP1 is installed and rebooted the system.

BOINC then deletes astronomy 0.20 ATI SSE2.exe and brooke32.dll leaving the app info.xml in place then...nothing!

Have I missed something?

running AMD Phenom 9650 Quad, ATI 4870, heaps of RAM and WinXP Pro SP3

---

Hi Raistmer,

Not for all hosts and only from credit-hunter point of view.

I hardly think anyone would put me in that category! It's just nice to juggle the software and get the most out of what we are doing.

I had your same issue. I hadn't read the readme.txt like I was supposed to. Your issue is that there are a few lines in the appinfo.xml file that need to be deleted if you are not using BOINC v 6.10.x

Copied from the readme:
Configuration for older BOINC clients (before 6.10.x):

The included app_info.xml declares the app as GPU app for the new 6.10.x clients. This is done
with a <coproc></coproc> statement, which may confuse older versions as they don't recognize the
type of required GPU. Theses clients may refuse to run the app in such a case. To get it running,
just remove the lines
<coproc>
...
</coproc>
and everything between those tags. It is twice in the .xml file, you have to remove both. After
that you should adjust the value <avg_ncpus>0.05</avg_ncpus> (just above the coproc statement)
to something higher (depends on your needs, but I would recommend 0.5 or even 1).

That should fix your problem. Hope it works.

Had problems with my 3850 and 0.20 so went back, temporarily, to 0.19 - zippo.

Detached and reattached, then downloaded 0.20 again to crunch the 20 odd MW WUs from the normal client (CPU version).

I seem to have problems starting the GPU crunching, and CPU-Z shows the GPU load at 1%-3%. The same seems to be happening with 0.19 and 0.20.

Suggestions to boot the GPU crunching to start. I am leaving BOINC Manager to restart, and the CPU is crunching 3 Malaria while it waits.

---

Go away, I was asleep

with <coproc> <type>ATI</type> <count>1</count> </coproc>
work only 1 WU 0.19g 6.10.4 x64 win7
without work's 3 WU 0.19g 6.10.4 x64 win7 ...
why coproc is need ?

0.20 errors after 1 sec (((
with \ without <coproc>...

I had your same issue. I hadn't read the readme.txt like I was supposed to. Your issue is that there are a few lines in the appinfo.xml file that need to be deleted if you are not using BOINC v 6.10.x

Copied from the readme:
Configuration for older BOINC clients (before 6.10.x):

The included app_info.xml declares the app as GPU app for the new 6.10.x clients. This is done
with a <coproc></coproc> statement, which may confuse older versions as they don't recognize the
type of required GPU. Theses clients may refuse to run the app in such a case. To get it running,
just remove the lines
<coproc>
...
</coproc>
and everything between those tags. It is twice in the .xml file, you have to remove both. After
that you should adjust the value <avg_ncpus>0.05</avg_ncpus> (just above the coproc statement)
to something higher (depends on your needs, but I would recommend 0.5 or even 1).

That should fix your problem. Hope it works.

OK, upgraded to 6.10.4, now I receive the following -
14/09/2009 16:35:58 Milkyway@home Message from server: Your app_info.xml file doesn't have a version of MilkyWay@Home.

---

Strange, I have no problem with boinc 6.10.3, win xp 32 and 0.20 ATI, my mv racs on a 4850 is growing every minute and able to do some cpu work too, like malaria. I am very, very happy! Thank you all....

---

Strange, I have no problem with boinc 6.10.3, win xp 32 and 0.20 ATI, my mv racs on a 4850 is growing every minute and able to do some cpu work too, like malaria. I am very, very happy! Thank you all....

Well i've downgraded to 6.6.36........and all is running merrily away now.

---