I have tried to understand why your computers are generating invalid results. As part of that effort, I pulled the data from 4 hours worth of workunits that had completed and looked at the ratio of valid to invalid results focusing on narrower subgroups of machines similar to yours. Here are the findings:

All results for workunits completed in past 4 hours

num_results validation_outcome 
----------- ------------------- 
92152       VALID   (99.2%)           
713         INVALID ( 0.8%)          

All results for workunits completed in past 4 hours with quorum of 2 or higher

num_results validation_outcome 
----------- ------------------- 
47276       VALID   (98.6%)            
665         INVALID ( 1.4%)            

All results for workunits completed in past 4 hours with quorum of 2 or higher on Linux

num_results validation_outcome 
----------- ------------------- 
3598        VALID   (99.8%)             
7           INVALID ( 0.2%)             

All results for workunits completed in past 4 hours with quorum of 2 or higher on Linux with AMD processors

num_results validation_outcome 
----------- ------------------- 
687         VALID   (99.6%)             
3           INVALID ( 0.4%)             

All results for workunits completed in past 4 hours with quorum of 2 or higher on Linux with the same types of processors as KerSamson*

num_results validation_outcome 
----------- ------------------- 
78          VALID   (98.7%)           
1           INVALID ( 1.3%)

*These are:

• AMD Athlon(tm) II X4 640 Processor [Family 16 Model 5 Stepping 3]
  
• AMD Phenom(tm) II X6 1055T Processor [Family 16 Model 10 Stepping 0]
  
• AMD Phenom(tm) II X6 1090T Processor [Family 16 Model 10 Stepping 0]

With this information, I cannot find any symptom of a larger issue. I do not dispute that you are seeing the issue on your computers. I have looked at them and seen the invalids previously. However, as far as I can see the issue is isolated to those machines when they run VINA.

Unfortunately, the FightAIDS@Home is about to enter one if its phases of only having VINA work. We run out of AutoDock work in ~10 days. More information will be coming out tomorrow.

However, we are within 2.5 weeks +/- a week of starting the beta test for our next research project (focusing on cancer research) and it will not be VINA based - this should give you a good project to run if you are willing to check back in a few weeks.

One of my systems is also a Phenom 1090T running linux. What I have noticed is that AMD chips tend to be slightly more sensitive to heat issues than do the Intel chips.

Whenever one of my machines, Intel or AMD starts throwing errors or invalids, blowing out the dust bunnies has ALWAYS resolved the issue.

To those of you having problems, I do wish you the best and hope your issues can be resolved.

Hi everybody,
at first, thank you for the kind words many of us wrote after my last post.
My hosts are still crunching, mainly on CEP2 for the AMD and Linux based machines.

Knreed ! I appreciate the time you spent for investigating.
Unfortunately, since I redefined the projects allocation in the device profiles at the beginning of the week, the last days and hours are not really representative.

I do not have an accurate monitoring and recording of invalid WUs. However my estimation is a rate of around 50 invalid WUs within 3 or 4 weeks between end of June and July 27th.
Invalid WUs are in particular frustrating since there is no indication/reason for the invalidity of a WU.
Depending of the project, it represent a lot of hours.

I've tried several times to compute SN2S, DSFL, and GFAM with really similar results:
- Sometime it is good
- Sometime it bad
- Sometime it is really bad.

I observe a similar behaviour with VINA-based FA@H WUs.

Over the years, I did not have any real concerns with invalid or errored WUs (probably in some ppm). At the beginning of a new project, some issues could occur and I accept it.
The current problem with VINA is that suddenly a host (sometime several hosts) does compute correctly but at the end a large part (sometime every WUs) are declared invalid without any reasons.
Temperature could be an issue but the troubles occurred also during the Winter time. Additionally, the machines are well ventilated and the CPU temperatures (for the AMD systems) are below 60°C
- Phenom II x6 1090: 57°C
- Phenom II x6 1055: 50°C
- Athlon II x4 640: 40°C

After I noticed this problem, I avoided VINA-based projects. Today, I do not have any choice excepted CEP2 which is not the easiest project in terms of resources and efficiency (even with SSD).

My post was also caused by WCG's silence. We did not know anything excepted vague indications about some possible future projects. As I read here, I notice that I am not alone with this impression to stay in the "fog".

Again I thank you all for the feedback and I hope that we will experience again more satisfying times in the future.

Cheers,
Yves

I would but I cannot:
http://boinc.berkeley.edu/wiki/Release_Notes_...e_with_Domain_Controllers



These system run AD for a home setup. Due to restrictions on the BOINC client, I'm at the latest that can be installed. Thanks for checking though.

57°C on the x6-1090 is hot, I've got one and even with the stock Amd-cooler the max temperature was around 45°C, as measured by Core Temp and this also claims the absolute max legal temperature is 67°C for this cpu.

With a better cooler, atleast when it comes to fan-noise, the current temperature is 41°C while it's 27°C in the room.


As for Vina, I've not tried the new FAAH-application yet, but atleast for the previous WCG-projects using Vina I don't remember any problems with these on my Amd-X6-1090T running win7/64.