FreeBSD S5R3 App 4.18 available for Beta test
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4312
Credit: 250205517
RAC: 35421
5 Nov 2007 16:25:51 UTC
Topic 193294
(moderation:
A FreeBSD x86 App is available from our Beta Test page.
This has been built due to somewhat popular demand and as I had a FreeBSD 5.5 environment ready. I hadn't time to put special effort in optimization, it's just what I could do while waiting for other jobs to finish. I expect the Linux app in emulation will run faster than this native FreeBSD app.
Give it a try. I'm curious how the latest BOINC library version works on FreeBSD.
BM
BM
FreeBSD S5R3 App 4.18 available for Beta test
)
It took a bit of effort to get everything running (more because of the somewhat different core client and file structure than the science app) but now everything seems to be fine. :-)
I got a S5R3 WU which should be ready some time tomorrow or the day after, depending on how accurate the estimate proves to be (currently 31 hours, but the WU has only just started). Naturally, I can't say anything about performance yet, but I'll have a good comparison later since I have some fresh Windows results from the same box. Basically, I'm glad I could rejoin so quickly after the OS change :-) thanks again for that, Bernd.
Sometime today, I hope to
)
Sometime today, I hope to have DesktopBSD running on a spare SGI 1200 server, with a dual P-III 700 setup. I'll give this a try and let you know how it works out.
This is awesome! Maybe I got
)
This is awesome! Maybe I got a bit lucky with the frequency band, but the WU crunched under FreeBSD was a good deal faster than under Windows. 17.3 creds/hour compared to 15.1 under Windows :-) Even if it was coincidence, 17.3 is not half bad for such a box (3500+, not overclocked).
Bernd, if that is what you can do "in your spare time completely without optimization", I shudder to think of the later versions :-D Thanks a lot for your effort! This put my box back on track big time...
I guess you got lucky. With
)
I guess you got lucky. With "no special effort" I meant that I just used the standard gcc-3.3 compiler that came with FreeBSD 5.5. It might be worth a look if some gcc-4 really makes faster code of our sources.
BM
BM
RE: This is awesome! Maybe
)
Hi Annika!
I've finally gotten DesktopBSD installed on my SGI box, and the BOINC client installed from the FreeBSD repository. But, when I start BOINC, I get an error message that the client can't connect to the host. Did you have to do anything special to get it to work for you?
Oh yes, don't we all love
)
Oh yes, don't we all love this error... In your BOINC directory, there should be a file called "gui_auth_cfg" (or similar, can't check right now). This contains a random hexadecimal key. Open the file with a text editor of your choice (I think you need root for that) and copy the key. Then go into BOINC manager, advanced, select computer. As computer you type in "localhost". Then you paste the key into the "password" window. Now you should be able to connect to the manager.
RE: A FreeBSD x86 App is
)
I downloaded the new application manually (the ports tree is frozen for now). The md5sum differs from what is mentioned on the beta page:
MD5 (einstein_S5R3_4.18_i386-unknown-freebsd.tar.gz) = adef424a702f7b122148bb5b77e370a8
SIZE = 1193948
Beta page: md5sum: 4fc760fec66070231cccb6041f8c2050
The contents look alright according to tar:
drwxr-xr-x 0 bema bema 0 Feb 16 2006 projects/
drwxr-xr-x 0 bema bema 0 Nov 5 16:58 projects/einstein.phys.uwm.edu/
-rwxr-xr-x 0 bema bema 3248091 Nov 5 16:38 projects/einstein.phys.uwm.edu/einstein_S5R3_4.18_i386-unknown-freebsd
-rw-rw-r-- 0 bema bema 441 Nov 5 16:57 projects/einstein.phys.uwm.edu/app_info.xml
I downloaded the file twice in succession to exclude some transient network error. So which MD5 value is correct?
Yes, the MD5sum on the page
)
Yes, the MD5sum on the page is probably wrong. I'll fix that ASAP.
Edit: Fixed.
Thanks for the report.
BM
BM
RE: A FreeBSD x86 App is
)
So far so good :) I lost one work-unit but that was caused by some "interesting issues" of the kernel scheduler. See http://einsteinathome.org/host/825677
RE: I lost one work-unit
)
It looks like you lost the S5R2 task that was still in the pipeline. Honestly I didn't expect people to still have one of these, and the app_info.xml isn't prepared for this.
BM
BM