FWIW, I'm running BOINC 6.12.34 (x64) and not 7+ because of the problem I described in this thread: https://boinc.bakerlab.org/rosetta/forum_thread.php?id=60579. The reversion to an older NVIDIA driver solved my problem in the current thread, but not in the older thread.

I am running, thanks to Max, several pc's with AMD gpu's, ALSO CRUNCHING but of course for other projects, and am using Boinc version 7.0.40 and am getting credits just fine.

Max I have both an 8 core AMD and two 6 core AMD machines crunching here, all have 16GB of ram in them. I have other machines here too but those are what you were asking Vester about the times, so I thought I would give you another example.

It has 10 GB of RAM because HP built it that way. This computer ran just fine without any errors until I installed BOINC 7.0.28 with Windows 8. I am going to try another project and see if I still have a problem. As it is, BOINC Manager will not reconnect to the client after a restart and if I do a dirty installation, it starts computing and then renders my computer unresponsive until I exit BOINC Manager.

It has 10 GB of RAM because HP built it that way. This computer ran just fine without any errors until I installed BOINC 7.0.28 with Windows 8. I am going to try another project and see if I still have a problem. As it is, BOINC Manager will not reconnect to the client after a restart and if I do a dirty installation, it starts computing and then renders my computer unresponsive until I exit BOINC Manager.

Does 7.0.28 install and work fine on Win 8, I thought a later version was designed for Win 8. Version 7.0.30 has this note in the Developers release notes:
Noteworthy fixes include:
1. Installation issue on Windows 8.

7.0.29 had this note:
client: added the definitions for the new Windows 7/2008r2 preSP1 and Windows 8/2012 SKUs based on the winnt.h from the Windows 8 RC SKD (also added as proof).

You might consider upgrading Boinc to version 7.0.40 or even the newest 7.0.42 BUT be warned they are NOT the 'release' versions so may have bugs in them! I am using 7.0.40 on all but one of my machines, it is using 7.0.42 and have had any problems. You can check all the Boinc release notes here:
http://boinc.berkeley.edu/dev/forum_thread.php?id=6698
It even has the download links in it.

2 Vester
Thanks for the information.

I have just downloaded 7.0.42 from the Berkeley site and will let you know how it goes.

Failure.

I've run distributed computing projects for eleven years and this is the longest-running bug I've seen (worse that United Devices). I ran MemTest86 version 4.20 for over three hours tonight without a problem on my AMD FX-8150. If it makes anyone feel better, BOINC 7.0.42 errors out on WCG also.

Memory test

As mentioned in the sticky, this problem has persisted since about last October without resolution, and I just saw another topic opened regarding errors. I did beta testing at UD, Find-a-Drug, and Folding@home. There doesn't seem to be anyone here actively working on this problem. This computer has run Rosetta@Home without errors on Linux Mint 14 and Windows 7.

The Intel i7 I mentioned before is running Windows 7 and all of the jobs run on it have ended with "client error."

Failure.

I've run distributed computing projects for eleven years and this is the longest-running bug I've seen (worse that United Devices). I ran MemTest86 version 4.20 for over three hours tonight without a problem on my AMD FX-8150. If it makes anyone feel better, BOINC 7.0.42 errors out on WCG also.

Memory test

As mentioned in the sticky, this problem has persisted since about last October without resolution, and I just saw another topic opened regarding errors. I did beta testing at UD, Find-a-Drug, and Folding@home. There doesn't seem to be anyone here actively working on this problem. This computer has run Rosetta@Home without errors on Linux Mint 14 and Windows 7.

The Intel i7 I mentioned before is running Windows 7 and all of the jobs run on it have ended with "client error."

Are you running an Nvidia gpu in the machine too? Because I am running, no Linux for me just Windows, several pc's with AMD gpu's also crunching in them and they are finishing the units just fine.

The HP with AMD FX-8150 eight core processor is running an AMD Radeon HD 7770.

The HP with AMD FX-8150 eight core processor is running an AMD Radeon HD 7770.

So it is only the Linux pc that is having troubles? I see all the Windows pc's are doing just fine and getting credits, is that what you are seeing too? If so there is no more I can do to help you, as I am NOT a Linux guy.

[edit]I JUST saw this at another project...A bug was introduced in the BOINC runtime library on 11 Oct that
will cause the app not to work with the latest (7.0.38+) clients
(they'll exit immediately with "no heartbeat" messages).
This affects Unix (Linux/Mac) only...
I wonder if you DOWNGRADED to an earlier version of Boinc instead of upgraded it would have worked? BE CAREFUL as downgrading Boinc means ALL units you currently have will be IMMEDIATELY aborted!! And that is for ALL projects that you have on that machine!!![/edit]

2 Vester
From my side thing looks like your i7 computer (this one: https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=1587703) obviously hit by well known "GPU bug".
So I suppose it have NV GPU on it? Or may be at least NV-driver installed...
If so workaroud will be downgrade NV drivers (see early posts in this topic).
If it helps please post old and new drivers versions here.

But with FX-8150 computer(https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=1589407) situation is certainly different. It is a pure software errors, so testing hardware is irrelevant for now.

Almost all WUs crash by this error:
couldn't start Input file minirosetta_3.45_windows_x86_64.exe missing or invalid: -163: -163

So it even can not start main R@H executable. Probable something(antivirus for example? or Windows UAC? or other security stuff...) on this computer blocks (or delete) this file. Or it was corrupted at downloading stage...
Fix to this is replace by fresh correct copy of file. Manual Or just do full reset of the project in BOINC.
I suppose you alredy did it (as i see a lot of WUs with "Client detached" in status) and this error is gone for now.

And after reset/detach seems all errors occur on "TOR_aa_hybrid..." WUs only.
It is a general problem with this particular new WU line - many of them are buggy. And it is not related to your computer (i also have few similar errors on TOR WUs, and wingmans in quorium too). So just ignor errors on this WU line.

All other WUs seems still running? (as you set 24 hours target time)
I think most of them will finish normal, so do not try a to change anything on FX-8150 computer. For current moment it is better to wait and see...

I just updated my Nvidia driver to a new version two days ago and now all my workunits are being marked as invalid. From what I've read here my guess is its got to be the driver, because nothing else has changed with my computer.

Kenneth DePrizio
Yes. Try intall "old" NV driver. 306.97 for example.
After it please post for statistic - downgrade help or not in your case...
And driver version before and after.