Let us know how it's working here.

---

There are a few long running tasks that error out with a "system freezing" message.

There are a few long running tasks that error out with a "system freezing" message.

Yeah we've noticed this. It's a problem with our flops estimate. It seems like the simulation runtime can vary pretty wildly, but we should have it sorted out pretty soon.

---

Is it possible to "Opt Out" on the nbody simulation / application? I have a 32 bit Intel machine is that is producing constant computation errors, along with Windows application errors. (As an FYI: I have two AMD machines that seem to be working with nbody without problems.)

I thought this would be fixed by now but, apparently, that's not the case.

Is it possible to "Opt Out" on the nbody simulation / application? I have a 32 bit Intel machine is that is producing constant computation errors, along with Windows application errors. (As an FYI: I have two AMD machines that seem to be working with nbody without problems.)

I thought this would be fixed by now but, apparently, that's not the case.

What processor? If it doesn't have SSE2 it won't work. Eventually it won't be sent to those systems.

From my prospective, the type of processor is not the point. I'm donating time on a computer that I don't use. Is that free? No. Even if I never used that particular computer again, the cost is about the equivalent of running a 75 to 100 watt light bulb 24 hours a day, 7 days a week, 365 days a year. So, with 3 machines crunching, over the course of a year (about 250 watts non-stop at an estimate), there's a real tangible cost to me.

The point I'm getting at is, pushing Alpha and pre-Alpha software out to donors of CPU time without giving them a choice in the matter is, well, "uncool" at best. I think the word "arrogant" is a better fit, since there are several (potentially thousands) of donors who are willing to download and test pre-Alpha code and provide feed back for refinement.

In any case, these continuous errors and crashes, over the course of a week or so, are a waste of the machines CPU time. And while I'm sure that no one will care either way, if this irritation doesn't stop soon, I guess I'll direct my excess Computing resources 100% to Einstein@home and call it done. (Einstein@home seems to value their donors, taking a conservative approach.)

For the sake of a year relationship (thus far), I'll give it another week.

The point I'm getting at is, pushing Alpha and pre-Alpha software out to donors of CPU time without giving them a choice in the matter is, well, "uncool" at best. I think the word "arrogant" is a better fit, since there are several (potentially thousands) of donors who are willing to download and test pre-Alpha code and provide feed back for refinement.

In any case, these continuous errors and crashes, over the course of a week or so, are a waste of the machines CPU time. And while I'm sure that no one will care either way, if this irritation doesn't stop soon, I guess I'll direct my excess Computing resources 100% to Einstein@home and call it done. (Einstein@home seems to value their donors, taking a conservative approach.)

For the sake of a year relationship (thus far), I'll give it another week.

We have gone through this before. No one could be bothered to put the 'tests' under the option so users could opt out if they want. As the apps kept crashing numerous computers. (In looking for what that was called I see it was deleted from the MW preferances tab in one of the site updates).

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

"No one could be bothered to put the 'tests' under the option so users could opt out if they want."

I must say, I am amazed..

Even Microsoft (who is not known for their personal service) allows users to "opt out" of their "software improvement" and "quality feedback" programs. And if one really thinks about it, the default should be "opt out".

I can't help but be amazed by the cooperative computing end users (the universities) who will swear that it's a great resource (saving hundreds of thousands in super computing time) but won't take reasonable precautions in preserving and utilizing the resource that users generously provide.

Thus far, I've seen a simple hard drive array failure turn into a 2 or 3 week issue, where users couldn't donate time if they wanted to. As a former LAN Man, I did reasonable hardware health checks and maintained known working spares, to include a RIAD 5 controller because one doesn't have a "hard drive" (array) without one.

In any case;
I believe there are probably several thousand peers, like myself, who have extra PC's with decent performance laying around. Like me, even in this rough economy, they're willing to connect them to the Net and pay the utility company to run them flat out (which results in light bulb like power consumption).
The net result?
The scientific community gets a break on super computing costs and the peer provider gets to "feel good" for donating something.

On the other hand;
Finding one of my machines powered on, but doing nothing because it's constantly crashing is unsat. Why? Time is wasted, power is wasted, and it's completely avoidable.
___________________________________________________

I don't like to complain without offering a solution so;

Drawing from the LAN Man days, where I set up small but complete networks for testing purposes, I think another project should be considered to segregate development efforts from the mainstream. Call it "Milkway-Dev@home", or something like that, and develop software in that domain. Again, there would probably be thousands who would support it, and they'll expect to be dealing with issues. On the other hand, the rest of us won't be irritated by being forced to reboot our machines all the time.

Just a thought.
Regards.
_________________________________

"By definition, it is impossible to plan for the unexpected."