Latest boinc version thread

[Sunfire]

6.1.14 is now available for Windows 32-bit users.

Download link

This is a development version, so use at your own risk.

Did you tried it?

[sierra_bound]

No, I have not tried it. I'm happy with the BOINC versions (5.10.13, 5.10.39 & 5.10.45) that I'm using now.

[SiGfever]

No, I have not tried it. I'm happy with the BOINC versions I'm using now.

S_B, what versions do you recommend for XP x64 and Server 2003 E 32bit? I am running 5.10.30 now.
Thanks,
John

[sierra_bound]

I use 5.10.39 x64 on my Clovertown system with Server 2003 x64. Combination works really well. That rig averages close to 32K points a day.

5.10.13 x64 and 5.10.45 x64 on two machines that have Server 2008 x64.

5.10.13 x64 on one system with XP Pro x64.

5.10.45 on another machine with XP 32-bit.

I've always liked 5.10.13 because it runs problem-free. I'm testing to see if 5.10.45 does a little better with Server 2008.

I don't use Vista. When Microsoft stops supporting Server 2003 and XP, I will probably switch over to Linux. Server 2008 is okay for crunching. But I would never use it for daily tasks.

[brot]

I don't use Vista. When Microsoft stops supporting Server 2003 and XP, I will probably switch over to Linux.

That means the possibility will rise that i get paired with better pc's, and get more points

[sierra_bound]

Actually there are more BOINC users running Linux than Vista.

http://boincstats.com/stats/host_os_...hp?pr=wcg&st=0

[brot]

When you are look at the average credit/os you know what i mean

But i have to say that since the 5.8 or 5.10 versions, the linux clients got closer to the points/wu value windows gives. Given that you have more wu/day with linux, it is maybe equal. When i (hopefully) finish school this may, i have more time to test things like that

[sierra_bound]

Server 2008 actually produces more points/day than Server 2003 on same system. Unfortunately Server 2008 reminds me too much of Vista. Not surprising since it uses the same kernel as Vista SP1.

This video summarizes my feelings about Vista.

http://www.blimptv.net/mostpopularV1.html

[SiGfever]

Server 2008 actually produces more points/day than Server 2003 on same system. Unfortunately Server 2008 reminds me too much of Vista. Not surprising since it uses the same kernel as Vista SP1.

This video summarizes my feelings about Vista.

http://www.blimptv.net/mostpopularV1.html

S_B, you never cease to amaze me.

[Gamekiller]

It's funny, I've been running Vista for a very long time now on 4 machines and never had any problems. =/

[brot]

I have seen this vid before, but i lost it. Thanks for posting.

I had the same feeling with 98 already, and so i switched to linux. And approx since XP i am totally microsoft free. Well, not totally, if i need win for photoshop cs3 (ps cs2 works in linux now, yay ) i have my trustworthy virtualbox. There windows can do what it want, a single click and everything is in a working condition again....
Welll,...wasnt that thread about boinc versions once?

[sierra_bound]

6.1.14 is now available for Mac OS X users.

Standard GUI

Unix command-line version

Development version. Use at your own risk.

[The Warlock]

Server 2008 actually produces more points/day than Server 2003 on same system. Unfortunately Server 2008 reminds me too much of Vista. Not surprising since it uses the same kernel as Vista SP1.

This video summarizes my feelings about Vista.

http://www.blimptv.net/mostpopularV1.html

DCS, , i like it, you made my day!

[sierra_bound]

6.1.14 is now available for Linux users.

Linux x86

Linux x64

Older Linux x86 & x64

Development version. Use at your own risk.

[sierra_bound]

6.1.15 is now available for Mac OS X users.

Standard GUI

Unix command-line version

Development versions. Use at your own risk.

[brot]

[ebuild U ] sci-misc/boinc-5.10.45

New ebuild in portage for the gentoo-crunchers. Yay.

Edit:
Works. And, if anyone needs a binpkg of it: you may download it from my server

[rcofell]

[ebuild U ] sci-misc/boinc-5.10.45

New ebuild in portage for the gentoo-crunchers. Yay.

Edit:
Works. And, if anyone needs a binpkg of it: you may download it from my server

Hmm, I'm just curious, for the gentoo ebuilds does it make sure to not use any optimizations? I seem to remember compiling the trunk code directly from Berkeley based on their trac wiki instructions, but obviously the instructions were for optimizing the client's benchmarks so I had to drop back to only the -O2 flag, or maybe even not that.

[brot]

http://boinc.berkeley.edu/trac/wiki/CompileClient

That doesnt look like an attemp to optimize the client, even if they use very agressive cflags.
Those cflags get used in the ebuild, too:

src_compile() {
        #upstream recommendation for flags
        append-flags -O3 -funroll-loops -fforce-addr -ffast-math

So it looks like it is normal to build it with flags like that.

[rcofell]

Hmm, because I remember using those flags when I first compiled and I was getting results way above what the normal binaries do. I'll have to see if I still have the results here, or do the testing again.

[sierra_bound]

6.1.15 is now available for Linux users.

Link

Development versions. Use at your own risk.

[WesM63]

I'm still using 6.1.0 on Vista Business x64, is that bad?

[rcofell]

Not really, I'm still running it also. Although I should try one of the newer builds, as I'm hoping they fixed the issue they had with WCG on all the other 6.1.x releases I tried(up to 6.1.8 I'm thinking).

[rcofell]

http://boinc.berkeley.edu/trac/wiki/CompileClient

That doesnt look like an attemp to optimize the client, even if they use very agressive cflags.
Those cflags get used in the ebuild, too:

src_compile() {
        #upstream recommendation for flags
        append-flags -O3 -funroll-loops -fforce-addr -ffast-math

So it looks like it is normal to build it with flags like that.

Well, after hours of frustration over compiling from svn, I've given up doing a comparison by manually compiling it, although maybe I'll make a script to do it for me later if I feel like it... For some reason it refuses to make the self extracting binary, although I guess I finally figured out it was actually compiling correctly( yay ), just not getting to the step of archiving itself, because for some reason boincmgr doesn't compile even though the compiling doesn't end with any major errors... It does that when I compile it normally, but then when I compile it as a release it'll spit out garbage about libcurl having issues...

What I should've done right away was just take Berkeley's version off the site and compare it to the gentoo ebuild. I don't know how to feel, my suspicions were correct, but I wasted 3 hours in the process of getting mad at the source code(yeah, I guess I'm either too stubborn or just get too focused on a subject)...

On second thought it wasn't that hard to compile once I knew that it was actually compiling the client(luckly all you need is the base client to run the benchmark, and maybe only one or two other files thereafter to run with a project, like the one for openssl), so I've included the results.

Linux: 2.6.23-gentoo-r8
Athlon X2 4200+ @ stock (yeah, yeah, it's the school's comp)
All the clients are the x86_64 ones.

5.10.45 binary from Berkeley 
2090 Whetstone
4374 Dhrystone

5.10.45 ebuild with "-O3 -funroll-loops -fforce-addr -ffast-math" flags
2709 Whetstone
6033 Dhrystone

5.10.45 compiled with "-02" flag only
2102 Whetstone
4359 Dhrystone

5.10.45 compiled with "-O1" flag only
2029 Whetstone
4032 Dhrystone

[DDTUNG]

I have had good results with 5.10.39 x64 and my Clovertown rig. Dhrystone benches about 500 points higher than 5.10.13 x64. Work units are being completed faster. This of course means more results returned per day. Quorum results have also been encouraging. Still too early to draw definitive conclusions. But I like what I see so far.

Thanks for the headsup s_b. I am going to give 5.10.39 x64 a try.

DDTUNG

[[XC] Adywebb]

http://www.getdeb.net now has deb's for BOINC 5.10.45 for *buntu