I agree with RT. I am mostly a set and forget on these projects. I hardly check this form. I would like to have my quad do all projects, can't cause they don't care about quads on this projects.When I got the quad a 2 months ago. It got a lot of errors on this project. cause my default all project on my dual core. So that was the first time I look on the forum.I only got a the silver badge cause of my dual core. I onlystarted doing DC computing this year. Not going to report error. Don't want to babysit Boinc. Just want to set in and forget it.

No problem.

hmmm "... despite the computer science being flawed."

Lets emphasize unless there's proof to the contrary: Because some HPF2 work units fail to run successfully on a limited number of set ups for unknown reason the science is NOT flawed. The executable has an oversensitivity to an unknown configuration [my bet is in the permissions area]

Meantime, I think to have stumbled on a configuration that makes RICE run successfully now on my duo XP with 6.2.28 (which runs HPF2 at 100%). Finally had a trial job go past the first checkpoint and finish in the standard 7 hours and now sitting with 8 others of the quorum in PV. Will try more later, in combo with HPf2 after a boot to see if it was a fluke or has turned stable.

Let us hope that soon the filter is put in place so devices do not get these jobs where they always or frequently fail on [playing the scratched lacquered record].

edit:

So far so good, boot and RICE / HPF2 running concurrently, first checkpoint logged with a Write to Disk setting of 300 seconds i.e. the science must have run at least 5 minutes before the first checkpoint writes are allowed, then spaced by at least 5 minutes afterwards, if there is one to write:

It seems that knreed is now going to be working on solving the problem (from a post in another thread -btw knreed let me know if you want me to test something).

I cannot believe the waste: an incredible amount of time, thousands of hours of computer time, BOINC downloads and reinstalls (forward then backward) OS reinstalls, directory changes, thousands of downloads and uploads of errors, many "hop on your right foot while holding your left hand behind your back with your right eye closed... and see if that will fix it." posts on the forums.

It would have saved all these people an incredible amount (plus got a lot more done for WCG projects) if you just buckled down and fixed it when it first started happening over a year ago rather than waste all this time, effort and resources and only then actually try to fix it?

The scientist comment was a bit like the comment by D Eisenhower just before D-Day that some percentage of casualities was expected and acceptable. (Common position taken in wars by all sides). That all makes sense from the perspective of the scientist or the General. But, not if you are one of the casualties. In case of war, you do not have a choice but here you do .

The mission of WCG is "... to create the world's largest public computing grid to tackle projects that benefit humanity." How many people has this caviler attitude driven away from HPF2 and more importantly, how many from WCG? Is it one.., ten.., a thousand..., more? How do you know? How much advertising, and recruiting effort has this offset since volunteers are hard to get and easy to lose?

I suggest that it will take less than 1% of the time and effort to fix it than has been expended making excuses, denying its importance and trying to sweep it under the carpet. And that is Sad, Not Funny.

RaymondFO:
That is very interesting. Let me get this straight: If you run a max of 4 simultaneous HPF2 WUs on this I7 with W7, all is well, but if you run more, failures occur?

I have an I7 920 that I run 8 WUs on all the time. I did this with a version (I forget which) of Vista and am now doing it with the Windows 7 Release Candidate (64bit). I am using BOINC 6.6.38 and am exclusively running the two cancer projects at the moment.

Now, it has been a while but I was getting these errors on all four of my quads. Only one was an I7 which has hyper-threading but they were all Vista 64bit...till I changed the I7 to W7 and all were getting a fairly high error rate. So.. either something has changed or you are indeed lucky. Perhaps this insight of yours (about "native" cores) will help knreed in his search for the real solution.
Regards