The new version has been uploaded as 0.23 for Windows 32 bit and Linux/GNU 64 bit. It works for me on Linux/GNU 64 bit, but I was unable to test it with Windows (XP isn't too fond of the idea of having an ATI GPU and NVIDIA GPU in the same system).

Please report any issues.

Thanks Anthony Waters it works fine with xp 32:
I was not able to do work with the old wesion.

<core_client_version>6.10.3</core_client_version>
<![CDATA[
<stderr_txt>
Device index specified on the command line was 0
Looking for a Double Precision capable NVIDIA GPU
The device GeForce GTX 260 specified on the command line can be used
Used 129584/917184 memory, 787600 remaining
called boinc_finish

</stderr_txt>
]]>

I just noticed that Cuda 2.2 DLL's were downloaded with the new app.
Were these supposed to be 2.3 DLL's

I finished 4 WU's on my XP64 and validated good. With 2 EVGA 260s, Nvidia 195.62.

However the computer is "barely" useable for the lag, even worse than the old app.

New app does work though. The previous app. gave me nothing but validate errors.

Yep, the lag is terrible, I rather have one third longer application than this
lag.

The larger WUs increased the number of threads, I'll play with some of the settings to reduce the burden and publish 0.24 within the next 24 hours.

Finished my first valid CUDA WUs on WinXP x86!

Well done, folks and thank you for your participation in this interesting little bug hunt.

Keep on crunchin!

---

Version 0.24 has been published, it should reduce the apparent lag when running the CUDA application.

Great news, Anthony. Still crunching the .23 units but can see the .24's downloading.

In other breaking news, my 4GB RAM sticks finally arrived, so I have upgraded a couple of my i7 boxes from 8GB to 10GB, and MW now runs rock solid from not at all (Win 7), and Seti with an ~50% error rate is now down to under 5%.

Of course, this was the box running eight cores of Cosmology.

---

Great news, Anthony. Still crunching the .23 units but can see the .24's downloading.

On one of my systems I have .21, .23 and .24 versions ... ah well...

I turned on one of the systems that was having issues with MW and if it runs them well ... I will turn it on for the other later on ... sadly, will be a bit because I have to run off the other CUDA work on hand ... that stupid Strict FIFO rule ... sigh ... well, maybe I will go in and suspend some tasks to force things ...

Hi,

I think I have a problem:
http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=9243561

Check my host to get the other errored out wus.
In the meantime, I suspended the project not to waste tons on wus.

I'm running Ubuntu 9.10x64 with boinc 6.10.17 and nvidia 190.42 drivers for my GTX275 GPU (896 Mb RAM)

Note : Aqua is running on the CPU

Thanks for your help

---

New .24 working great on XP64, with 2 EVGA 260s 55nm., nvidia 195.62
Lag is better than it has ever been with the previous apps, just a little but tolerable.
Validating good too.

Question: Why is the app call cuda 23 when the cuda dll's are 2.2 ?
just wondering.

I finished 4 WU's on my XP64 and validated good. With 2 EVGA 260s, Nvidia 195.62.

However the computer is "barely" useable for the lag, even worse than the old app.

New app does work though. The previous app. gave me nothing but validate errors.

@ Travis or Anthony: I think a message should be up on the home page that this is fixed. Might help a few out.

Well whatever has been done has fixed this issue for me. Just checked BOINC and sure enough, there's a whole stack of WU's happily crunching away. Doesn't appear to be any computation errors at all, either... at least not from MilkyWay. Still getting the odd one out of Einstein, but that's another story

Thanks all

Hrmm spoke too soon, I think. Getting more frequent with the computation errors across the board, but I am starting to suspect drivers issue., so I'm gonna revert back to 191.07 and let that go for a bit.

Hrmm spoke too soon, I think. Getting more frequent with the computation errors across the board, but I am starting to suspect drivers issue., so I'm gonna revert back to 191.07 and let that go for a bit.

I see

Error executing gpu__integral_kernel3 error message: unknown error

a few other people are experiencing the same thing, I'm thinking it might be something to do with the way the memory is freed when the application is done executing. I'll look into the issue in more depth over the week.