In 5 days time I will be starting my last two FAMOUS tasks.
No more seem to be forthcoming at present and my Mac is banned from the regional tasks (in spite of the fact that I requested and was sent one EU task by mistake some time back, and it was completed successfully!!!).

So, unless more tasks appear, I will have no "insurance" against the calculation errors while my last 2 FAMOUS tasks are processed.

I know there are other projects, but I was a SETI plus CPDN faithful until SETI lost the plot, and now I am solely interested in CPDN.

Keith

Mac and Linux versions of the HADAM3P regional application are being tested, though it's unlikely they'll be ready before your two FAMOUS tasks finish.

However, the existing FAMOUS work units are currently being examined to produce more work units, so there should be more of those pretty soon.

When your Mac was sent a regional model some time back it wasn't by mistake. For a while the regional models were available for Windows, Linux and Mac. Then the researchers realised that the three operating systems weren't necessarily producing results that were all strictly comparable. So the current Beta testing is to put this aspect right (assuming that this can be done) before releasing a new version for Mac and Linux on this main project.

While waiting for more FAMOUS models to be generated, probably on Friday, no computer needs to be idle. There are lots of useful Boinc projects. For example, one can reduce the computer's work buffer to just 0.25 or 0.5 days (Boinc manager > Advanced menu > Preferences > Network usage > edit Additional work buffer > Click OK. That stops the computer fetching a vast amount of work. If you attach to Einstein, for example, their tasks keep the computer busy with useful tasks but not for too long. As soon as there's new CPDN work available just set the extra project to No new work in the Projects tab.

---

Cpdn news

Dear Keith:

If you are looking for a useful task for a few hours of idle core time you might try Rosetta@home. WU’s can be run in as little as 3 hours. Just set your work buffer for 0.25 days (as Mo advised) and it will download 1 or 2 Wu’s for to idle core. This way you can put idle cores to good use without tying them down for to long.

---

Just a quick summery of the posts over on the CPDNbeta site regarding the Mac and Linux versions of the Regional Models.

The Regional Models seem to be running alright on Linux Ubuntu, but, there are posts reports of immediate crashes on other forms of Linux such as gentoo and fedora. One Mac user reports that the model seem to be running alright, but, he is not getting any graphics. The same problem (no graphic) is reported on Darwin.

---

Hiro's releasing more FAMOUS but in small groups because they have different characteristics. Some do 'one sweep', others do 'two sweeps' though I don't know what this means. The more important groups are being released first.

I don't know whether the supply will be constant so if you have Mac or Linux get some while you can.

---

Cpdn news

Thanks for that Mo & Hiro

I have already increased my cache to maximum and have 4 FAMOUS in reserve with 2 running, so hopefully will be able to sustain a spate of the calculation errors.

Keith

P.S. Thanks for suggestion of Rosetta, Jim, but I prefer to stick to CPDN as long as it survives on Mac OSX.

P.P.S. Mo, would it be helpful to run the regional tasks on my Mac using virtual Windows XP with Parallels?
That is, to see if the same problems arise, whatever they are, and if they are due to OSX or to the MAC computer's chips?

Keith: You should now find that the FAMOUS queue has been replenished.

Keith, if you run Boinc and FAMOUS on the Mac in a partition with XP you should get the Windows compilation which will run more slowly. Some models will still crash but it should be a smaller prorportion.

I don't think there's ever been any suggestion that the cause of the crashes lies in the Mac hardware or OS X. The problem lies in the FAMOUS compilation for Mac which makes the model run faster than on Windows or Linux. I never fully inderstood why the Mac compilation is faster but there were reasons why it couldn't be slowed down.

---

Cpdn news

Hi Mo

OK. I shall not bother with that test for virtual XP on OSX then!!!

One reason why OSX runs faster might be due to the complete lack of crunching details on the graphics display. I feel sure that must save quite a lot of processing as I assume that is still available to Windows Crunchers?.
And I certainly miss the information that we always used to have to be in the know as to the progress of each task, and when it was near it's "save" points for time steps and phases.

Keith

Thanks for that note, Iain.

I had in fact already picked up one task in each of the last 2 days, before they started accumulating in the queue. I need a cache to cover the eventuality for a couple of "calculation errors" making me hit my maximum allowance of 2 tasks per day. As I now have 2 crunching, and 4 held in the cache, I am fairly safe from running dry now.

Keith