[zombie67]

For those interested, and have been frustrated that no mac application is available, now is your chance. They have a beta project running to develop an (intel) OSX application:

http://climateapps1.oucs.ox.ac.uk/beta/

Team MacNN is already #2 there, but needs a lot of help to make #1.

[ritzl]

Thanks.

[zombie67]

Rumor has it that the intel/osx version gets much better credits than the windows or linux versions.

[beadman]

Any numbers in that rumor, such as credits/hour?

beadman

[zombie67]

Originally Posted by beadman 

Any numbers in that rumor, such as credits/hour?

beadman

Sorry, I can't recall where I read about it. Too many forums!

[reader50]

I'd be happy to crunch some Climate, as soon as they release more PPC work.

[zombie67]

Originally Posted by reader50 

I'd be happy to crunch some Climate, as soon as they release more PPC work.

Yeah. Even the beta is intel-only.

[Billy]

Originally Posted by beadman 

Any numbers in that rumor, such as credits/hour?

beadman

I am getting 7776 credits for 3 trickles which are 158,193 CPUsec on my 1.84 Mhz Core Duo iMac. This is 77760 timesteps, so 1 credit per 10 timesteps. Does it matter since the credits are not reported under Boinc?

[zombie67]

Originally Posted by Billy 

I am getting 7776 credits for 3 trickles which are 158,193 CPUsec on my 1.84 Mhz Core Duo iMac. This is 77760 timesteps, so 1 credit per 10 timesteps. Does it matter since the credits are not reported under Boinc?

Once the application migrates to the regular project, then the credits start to matter.

[beadman]

Originally Posted by Billy 

I am getting 7776 credits for 3 trickles which are 158,193 CPUsec on my 1.84 Mhz Core Duo iMac. This is 77760 timesteps, so 1 credit per 10 timesteps. Does it matter since the credits are not reported under Boinc?

Well, if I did the math right, that's around 44 CPU hours, so you're getting around 176 credits per CPU hour, which is a lot more than you'd get with SETI...

beadman

[Billy]

Originally Posted by zombie67 

Once the application migrates to the regular project, then the credits start to matter.

There is a new application posted, so you need to upgrade.

17 Jan, 2007
We've just put up beta versions (CPDN version 5.20) of Windows, Intel Mac, and Linux executables, which should be optimized for SSE2 (i.e. Pentium4 on up) and have greatly reduced disk & file I/O (90% reduction)! The workunits are an assortment of finished BBC experiment runs, so we can cross compare and ensure that the optimizations are not changing the physics of the models.

Note: There were some errors in the workunits on this beta test -- so if you have a workunit that doesn't start with 'hadcm3transb1' then you will want to detach/reattach to get a proper workunit and the new 5.20 version. Sorry for the inconvenience!

Also note that graphs should now appear, you can see from your 'Results' list (which is how I discovered the problem with workunits above! ;-)

[ritzl]

I've crunched 178 hours for 36K credit (202 credit/hour) on a macintel core duo 1.86 running Climate Prediction 33% of the time. That was with a superbench 5.49 client so I don't know if that had any effect. I switched back to the original BOINC client to see if it did. Will report back on that.

I detached, because I thought this was outrageously high credit, and reattached to get 5.20 worker which is supposed to be faster.

It seems like Climate Prediction can really generate the credit.

[zombie67]

Originally Posted by ritzl 

It seems like Climate Prediction can really generate the credit.

FYI, the regular, non-beta application generates normal amounts of credit. So when the mac version migrates to to the regular project, expect the credits to be reduced to be comparable.

[zombie67]

Climateprediction.net Beta

They are fixing the credit problem, and will start awarding credits against the normal CPDN account. So that should show up on external stat pages.

[ritzl]

Thanks zombie67. My credit was adjusted a couple of days ago. The superbench had no effect that I can discern.