What will mbe the consequences for BOINC projects ?
Should i avoid running BOINC apps on a multiprocessor machine ?
Just read:
http://www.physorg.com/news187463445.html
and found:
If you enter the same computer command, you should get back the same result. Unfortunately, that is far from the case with many of today's machines. Beneath their smooth exteriors, modern computers behave in wildly unpredictable ways, said Luis Ceze, a UW assistant professor of computer science and engineering.

"With older, single-processor systems, computers behave exactly the same way as long as you give the same commands. Today's computers are non-deterministic. Even if you give the same set of commands, you might get a different result," Ceze said.

He and UW associate professors of computer science and engineering Mark Oskin and Dan Grossman and UW graduate students Owen Anderson, Tom Bergan, Joseph Devietti, Brandon Lucia and Nick Hunt have developed a way to get modern, multiple-processor computers to behave in predictable ways, by automatically parceling sets of commands and assigning them to specific places.
Sets of commands get calculated simultaneously, so the well-behaved program still runs faster than it would on a single processor.

Next week at the International Conference on Architectural Support for Programming Languages and Operating Systems in Pittsburgh, Bergan will present a software-based version of this system that could be used on existing machines.
It builds on a more general approach the group published last year, which was recently chosen as a top paper for 2009 by the Institute of Electrical and Electronics Engineers' journal
...
multiple processors are responsible for elusive errors that freeze Web browsers and crash programs.

It is not so different from the classic chaos problem in which a butterfly flaps its wings in one place and can cause a hurricane across the globe. Modern shared-memory computers have to shuffle tasks from one place to another.
The speed at which the information travels can be affected by tiny changes, such as the distance between parts in the computer or even the temperature of the wires.
Information can thus arrive in a different order and lead to unexpected errors, even for tasks that ran smoothly hundreds of times before.

Should i avoid running BOINC apps on a multiprocessor machine ?

I guess that if you do not trust it, that you'd best shop on eBay for an older single CPU AMD or Intel and run BOINC on that.

But other than that, there are thousands of multi-core hosts being used out there and I haven't seen any messages of impending doom. And even then, if you suspect impending doom to strike upon you, either run BOINC when the computer is otherwise idle, or run it on a single core only. Or don't run it. That's not up to me to decide for you. :-)

---

I have no worries about doom or breakdowns of my operating system or similar events.
But i have just the fear that the calculation results of my 10 BOINC projects will be contaminated by unpredictable errors.
At present I run LHC@home, boincSIMAP, climateprediction, cosmology@home, World Community Grid, Einstein@home, POEM@home, QMC@home, rosetta and milkyway@home.
LHC@home team told me there was no reason to worry for their application.

Hope so.
Greetings from Germany
Jochen

Even then, isn't it something for the projects to worry about? It's their science that gets the possible corruption.

---

I wouldn't worry about it.

Computers are no more (or no less) deterministic now than they were before.

I've been programming, designing, building, using, and playing with computers since the mid '70's. I've built computers out of everything from telephone relays (look up RODABS if you wish -- or not, it's so old even Google isn't finding it!), to modern LSIs with billions of transistors.

On the face of it, that article sounds so ludicrous that I was checking the date to make sure it wasn't April 1st.

Frankly, the article doesn't make it clear exactly what they claim the problem is. Simply having multiple threads or multiple processors (as every consumer computer utilizing at least hyperthreading has) introduces problems -- but these occur even in single thread processors due to the multi-threading abilities of the operating system. Every computer using any operating system more advanced than MS-DOS has had to deal with these issues, which fall under the category of what's called "concurrency problems".

This is a programming issue, not a hardware fault.

It's something that's been known to computer science for some 30 to 50 years, give or take, and there are, of course, ways to deal with it. Programs, especially those that take advantage of multithreading, need to be written in such a way as to prevent those problems from occurring. It doesn't make the computer unreliable unless the program doesn't account for the concurrency issues. That's a software bug, and the fault of the programmer; it's not due to an unreliable computer.

Unfortunately, the article doesn't really go into any detail at all, perhaps because they're not releasing any details until they do their presentation at the conference next week.

So, on the face of it, that article appears to be complete BS. If not an April 1st joke, it looks like something that would be published at "The Onion".

That being said, it's been my experience that when I think someone else is saying something utterly idiotic, it's usually because I either don't understand what they're saying or I don't have some vital information that they do. That might very well be the case here, since 1) it's not April 1st, 2) it's not published in "The Onion", 3) this is coming from people studying computer science, not agriculture (no offense intended to agriculture; I enjoy eating too.)

But the article itself doesn't really give us much to go on. Rather than speculating on what they're really talking about, if you're really worried, at least wait until they present the details next week.

But even without the details, if you accept the premise of their assertions verbatim, you should still realize that gazillions of WUs have been processed on multi-core computers without error. Even if there is some groundbreaking new truth to their work, this new 'fault' doesn't seem to be having much of an effect on BOINC processing.

So either way, I wouldn't worry about it.