Welcome to the MacNN Forums.

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

You are here: MacNN Forums > Community > Team MacNN > BOINC for OS X

BOINC for OS X
Thread Tools
jarling
Forum Regular
Join Date: Mar 2002
Location: 42 minutes from the other side of the world
Status: Offline
Reply With Quote
Dec 13, 2003, 03:09 PM
 
BOINC is now available for OS X beta testing here.
     
reader50
Administrator
Join Date: Jun 2000
Location: California
Status: Offline
Reply With Quote
Dec 13, 2003, 03:57 PM
 
Shaktai has set up a Team MacNN for us, but Berkeley has previously said that when SETI-1 ends, all members and teams will automatically be migrated to BOINC (SETI-2). This implies that the beta stats & teams are beta-only, or that early team forming on BOINC failed to do a check against existing SETI-1 teams.

All beta stats may go away when the test is done - has Berkeley said anything one way or another?
     
Shaktai
Mac Elite
Join Date: Jan 2002
Location: Mile High City
Status: Offline
Reply With Quote
Dec 14, 2003, 01:32 AM
 
Originally posted by reader50:

All beta stats may go away when the test is done - has Berkeley said anything one way or another?
The only thing that I have heard is that the "beta stats" may go away, but nothing definite. I figured as long as I am testing, no harm in adding the team. If the stats go away, fine, it is just one of the perils of beta testing. However if they figure a way to carry them over to reward the beta testers, then all the better. Beta testing is for helping to create a solid client. I am getting ready to update to the latest client version because of upload problems. Hopefully that will fix things. The current client is apparently terminal only. No graphics.

The client runs fine on my iBook 600 so far.
     
chris v
Addicted to MacNN
Join Date: Jan 2001
Location: The Sar Chasm
Status: Offline
Reply With Quote
Dec 14, 2003, 08:06 PM
 
?

The download page says the OS X app should launch automatically upon download and ask for an id #. I have a file on my desktop called boinc_2.12_powerpc-apple-darwin7.0.0 but no associated read-me file, and can't seem to get it to do much of anything. Nothing launched and asked me for anything. Top doesn't reveal any boinc processes running, and my limited vocabulary of UNIX commands hasn't succeeded in getting anything out of the file in the Terminal.

What am I missing here? There's f**k-all info on their website as to how to operate the client.

OS 10.3.1

CV

When a true genius appears in the world you may know him by this sign, that the dunces are all in confederacy against him. -- Jonathan Swift.
     
bradoesch
Professional Poster
Join Date: Jun 2000
Status: Offline
Reply With Quote
Dec 14, 2003, 11:00 PM
 
Originally posted by chris v:
?

The download page says the OS X app should launch automatically upon download and ask for an id #. I have a file on my desktop called boinc_2.12_powerpc-apple-darwin7.0.0 but no associated read-me file, and can't seem to get it to do much of anything. Nothing launched and asked me for anything. Top doesn't reveal any boinc processes running, and my limited vocabulary of UNIX commands hasn't succeeded in getting anything out of the file in the Terminal.

What am I missing here? There's f**k-all info on their website as to how to operate the client.

OS 10.3.1

CV
Fire up the Terminal and change to the directory where you downloaded and uncompressed the download. Then type ./ hit tab and press enter. Tab will autocomplete the rest of the file name for you. Yah, pretty stupid setup.
     
chris v
Addicted to MacNN
Join Date: Jan 2001
Location: The Sar Chasm
Status: Offline
Reply With Quote
Dec 14, 2003, 11:35 PM
 
Originally posted by bradoesch:
Fire up the Terminal and change to the directory where you downloaded and uncompressed the download. Then type ./ hit tab and press enter. Tab will autocomplete the rest of the file name for you. Yah, pretty stupid setup.
[Chris-Vreelands-Computer:~/Desktop] chrisvre% cd ~/Desktop/boinc
[Chris-Vreelands-Computer:~/Desktop/boinc] chrisvre% ./ boinc_2.12_powerpc-apple-darwin7.0.0
tcsh: ./: Permission denied.
[Chris-Vreelands-Computer:~/Desktop/boinc] chrisvre%


??

CV

When a true genius appears in the world you may know him by this sign, that the dunces are all in confederacy against him. -- Jonathan Swift.
     
reader50
Administrator
Join Date: Jun 2000
Location: California
Status: Offline
Reply With Quote
Dec 15, 2003, 12:10 AM
 
[Minerva:~] root# cd desktop/boinc
[Minerva:~/desktop/boinc] root# chmod +x boinc_2.12
[Minerva:~/desktop/boinc] root# ./boinc_2.12

Filename shortened because I didn't feel like typing the long version each time.
'chmod' stands for Change Mode, the '+x' part adds execute permission to the file.
     
Shaktai
Mac Elite
Join Date: Jan 2002
Location: Mile High City
Status: Offline
Reply With Quote
Dec 15, 2003, 12:42 AM
 
Keep in mind this is beta. There is going to be some problems and the OS-X client is essentially just a variant of the Linux client. No graphical interface yet except for Windows.

The forum is here:
http://setiboinc.ssl.berkeley.edu/ap/forum/

For problems and technical help go here:
http://setiboinc.ssl.berkeley.edu/ap.../help_desk.php

BOINC is progressing but it is still a bit of a frontier. Be patient. Nothing easy about beta testing.
     
chris v
Addicted to MacNN
Join Date: Jan 2001
Location: The Sar Chasm
Status: Offline
Reply With Quote
Dec 15, 2003, 08:05 AM
 
Being a 'nix *diot doesn't help, either. As usual, y'all are being most helpful.

CV

When a true genius appears in the world you may know him by this sign, that the dunces are all in confederacy against him. -- Jonathan Swift.
     
chris v
Addicted to MacNN
Join Date: Jan 2001
Location: The Sar Chasm
Status: Offline
Reply With Quote
Dec 15, 2003, 08:54 AM
 
Okay, i would appear to be running an Astropulse unit. Found some more readme pages on the web, but there doesn't appear to be a -verbose mode, like with the SETI CLI. So Ihave no clue how far along the W/ is, or anything. I can see it running in top, though. Looks like I downloaded 6 Astropulse W/U's and an ap_2.14 .app.

What have you seen as far as average time for results? Looks like you've turned in quite a few. And hey, we're just two teams back of GayAsians!

CV

When a true genius appears in the world you may know him by this sign, that the dunces are all in confederacy against him. -- Jonathan Swift.
     
Microns
Junior Member
Join Date: Mar 2002
Location: In the tall grass prairie next to a compass plant
Status: Offline
Reply With Quote
Dec 15, 2003, 09:06 PM
 
If the MacOS X client is like the Linux and Solaris cli versions then you should be able to monitor progress by looking at the "prog=" line in the state.sah file that is located in the directory containing the BOINC program. No "-verbose" required. It also appears that this client version does not work with MacOS 10.2.8 (Darwin 6.8) but only with MacOS 10.3 (Darwin 7.0.0).
     
chris v
Addicted to MacNN
Join Date: Jan 2001
Location: The Sar Chasm
Status: Offline
Reply With Quote
Dec 15, 2003, 09:25 PM
 
There's a file called clent_state.xml, but I'll be d*mned if I can find any english in it.

It's had about 9.5 hrs CPU time so far, and no results sent, yet.

CV

When a true genius appears in the world you may know him by this sign, that the dunces are all in confederacy against him. -- Jonathan Swift.
     
chris v
Addicted to MacNN
Join Date: Jan 2001
Location: The Sar Chasm
Status: Offline
Reply With Quote
Dec 16, 2003, 07:36 PM
 
Okay, the client looks to have sent a result around 3 this morning, but their stats page doesn't think so now almost a day later.

Until this thing gets a little less cryptic, I think I'll stick with something that I can actually see running.

CV

When a true genius appears in the world you may know him by this sign, that the dunces are all in confederacy against him. -- Jonathan Swift.
     
Scotttheking
Moderator Emeritus
Join Date: Dec 2000
Location: College Park, MD
Status: Offline
Reply With Quote
Dec 16, 2003, 10:56 PM
 
Originally posted by Microns:
It also appears that this client version does not work with MacOS 10.2.8 (Darwin 6.8) but only with MacOS 10.3 (Darwin 7.0.0).
I'll confirm that. I'd say it's a "minor" problem.
My website
Help me pay for college. Click for more info.
     
Shaktai
Mac Elite
Join Date: Jan 2002
Location: Mile High City
Status: Offline
Reply With Quote
Dec 18, 2003, 11:16 PM
 
Originally posted by chris v:
Okay, the client looks to have sent a result around 3 this morning, but their stats page doesn't think so now almost a day later.

Until this thing gets a little less cryptic, I think I'll stick with something that I can actually see running.

CV
Yes I can confirm that stats from the mac clients don't seem to be recording. Been too busy to post about it in the BOINC forums. There are still some ongoing stats issues with BOINC.

I've said it before but it bears repeating, this is beta, and it still feels like "early" beta. Progress is being made but it is not close to primetime ready yet. I do not reccommend the current beta for the faint of heart or the impatient.
     
reader50
Administrator
Join Date: Jun 2000
Location: California
Status: Offline
Reply With Quote
Dec 19, 2003, 01:06 AM
 
According to the BOINC stats rules, each client reports to the server how much CPU time it spent on a unit. Each work unit is sent out to two clients. Once both results come back (and agree), whichever client claimed the least credit - that credit is awarded to both accounts.

This will go a long way towards eliminating the cheaters, but also makes it likely that your slower box will only get the credit that a faster box gets for the same work unit. And it means a variable delay from when a unit is returned until it is credited.

If the Berkeley server is reserving work units by platform, then the delay in stats being awarded may be because very few Macs are working on the pool of Mac work units - or that someone with a slow older Mac is the other person crunching your unit. If that other person doesn't finish or return the unit, after a certain number of days it will be reissued to someone else. Either way, no stats will be awarded until someone else crunches it with the same results as you got.

Our delays could be a client bug as assumed, or a side effect of the new stats design.
     
Shaktai
Mac Elite
Join Date: Jan 2002
Location: Mile High City
Status: Offline
Reply With Quote
Dec 19, 2003, 07:34 PM
 
Originally posted by reader50:
According to the BOINC stats rules, each client reports to the server how much CPU time it spent on a unit. Each work unit is sent out to two clients. Once both results come back (and agree), whichever client claimed the least credit - that credit is awarded to both accounts.

This will go a long way towards eliminating the cheaters, but also makes it likely that your slower box will only get the credit that a faster box gets for the same work unit. And it means a variable delay from when a unit is returned until it is credited.

If the Berkeley server is reserving work units by platform, then the delay in stats being awarded may be because very few Macs are working on the pool of Mac work units - or that someone with a slow older Mac is the other person crunching your unit. If that other person doesn't finish or return the unit, after a certain number of days it will be reissued to someone else. Either way, no stats will be awarded until someone else crunches it with the same results as you got.

Our delays could be a client bug as assumed, or a side effect of the new stats design.
You are right. I forgot about that. Durned CRS....
     
chris v
Addicted to MacNN
Join Date: Jan 2001
Location: The Sar Chasm
Status: Offline
Reply With Quote
Dec 20, 2003, 09:21 AM
 
The pages must not update too often, either. After about five days, if fnally shows some work claimed, but still zero awarded. I guess Reader 50's note about the stats only being awarded after being crunched by two dif. clients would explain some of it.

I think I'll at least hang back until we see some kind of progress meter.

CV

When a true genius appears in the world you may know him by this sign, that the dunces are all in confederacy against him. -- Jonathan Swift.
     
reader50
Administrator
Join Date: Jun 2000
Location: California
Status: Offline
Reply With Quote
Jan 3, 2004, 12:30 AM
 
Has anyone completed some cross-platform benching to see if the G3/G4/G5 is competitive? If BOINC is using AltiVec, that would be ideal. Optimized code may be more important, but AltiVec can make up for a lot.

I've been looking over the different team ranks, and we're getting killed in SETI. Right now, our project breakdown by client optimization looks something like this:
  1. Folding@home - heavily optimized, outcrunches other platforms.
  2. RC5-72 - optimized (except the G5), outcrunches most other platforms.
  3. D2OL - has a 'hidden' vecmath library, PPC G4 very competitive. Assume the G5 is at least as competitive - not sure about the G3.
  4. SETI@home - no optimization, no AltiVec, neither is planned. PPC at serious disadvantage against cheap high-MHz PC chips.
  5. dFold - somewhat optimized code, but no AltiVec and none planned. Flakey clients, zeroed stats.
  6. Ubero - unoptimized java, no AltiVec at present. Old work units.
  7. BOINC - ???
Perhaps coincidentally, that client order is pretty close to MacNN's member participation order in the projects. We're doing well in the first 3 projects, and sinking in the next three. Of those three, SETI (BOINC) is the only one that we can do much about on this end (such as through beta feedback) - dFold and Ubero need to address client/work issues on the project end to become attractive again.

If BOINC is competitive, we could do some planning ahead to jump in once it comes out of beta. I'm skipping 10.3, so cannot bench the current client. Also, no x86 boxes here for comparison benching.

We need to work on SETI if we can, I don't like being passed.
( Last edited by reader50; Jan 3, 2004 at 12:43 AM. )
     
Microns
Junior Member
Join Date: Mar 2002
Location: In the tall grass prairie next to a compass plant
Status: Offline
Reply With Quote
Jan 11, 2004, 06:45 PM
 
Originally posted by Scotttheking:
I'll confirm that. I'd say it's a "minor" problem.
Fixed.

Go here: http://setiboinc.ssl.berkeley.edu/ap/

Download the 2.17 client. Install will update to 2.18.

Begin crunching. No worries so for on my 10.2.8 system...
     
Shaktai
Mac Elite
Join Date: Jan 2002
Location: Mile High City
Status: Offline
Reply With Quote
Jan 11, 2004, 11:18 PM
 
Originally posted by reader50:
Has anyone completed some cross-platform benching to see if the G3/G4/G5 is competitive? If BOINC is using AltiVec, that would be ideal. Optimized code may be more important, but AltiVec can make up for a lot.

If BOINC is competitive, we could do some planning ahead to jump in once it comes out of beta. I'm skipping 10.3, so cannot bench the current client. Also, no x86 boxes here for comparison benching.
Okay. I had some problems and had to reload BOINC on my iBook 600. I'll run some cross platform tests for 3-4 days. I just updated my PowerMac 800 to 10.3.2 so that it and the iBook have the same OS and ran benchmarks. The benchmarks don't seem to equate to real world performance but here they are including my Athlon 2400.

iBook 600 - OS 10.3.2
Floating Point: 142.02 million ops/sec
Integer: 100.31 million ops/sec
Mem bandwidth: 262.41 MB/sec

PowerMac 800 - OS 10.3.2
Floating Point: 241 million ops/sec
Integer: 131 million ops/sec
Mem bandwidth: 551 MB/sec

Athlon 2400 - Win XP home edition
Floating point - 216.72 million ops/sec
integer - 90.88 million ops/sec
Mem bandwidth - 638.68 MB/sec

In order to get "real world" numbers I am getting ready to start clean on all three boxes with the latest clients. I will run all three 24/7 for at least three days. Then we should hopefully see some real numbers if all three units show in the stats.

Based on messages on the message board the G5's are screaming along very well, and a 455 mhz G4 will take about twice as long as an Athlon 2000 for a work unit. However I doubt the "consistency" of the forum test results.

NOTE: I got lucky and got all SETI units for all three clients. That is good so there are no astropulse units to alter the results from the Athlon. The Macs get only SETI units so far.
( Last edited by Shaktai; Jan 11, 2004 at 11:56 PM. )
     
Microns
Junior Member
Join Date: Mar 2002
Location: In the tall grass prairie next to a compass plant
Status: Offline
Reply With Quote
Jan 12, 2004, 09:25 PM
 
Originally posted by reader50:
Has anyone completed some cross-platform benching to see if the G3/G4/G5 is competitive? If BOINC is using AltiVec, that would be ideal. Optimized code may be more important, but AltiVec can make up for a lot...

snip...
The only BIONC related cross-platform benchmarking thread that I am aware of is here: http://setiboinc.ssl.berkeley.edu/ap...ad.php?id=1019
     
Shaktai
Mac Elite
Join Date: Jan 2002
Location: Mile High City
Status: Offline
Reply With Quote
Jan 12, 2004, 11:23 PM
 
Okay, I am giving up on trying to benchmark for the moment. The database is all screwed up and nothing is recording. When they get it sorted out, I'll try again.

Best estimates that I can come up with right now is that in real world results, my Athlon 2400 is just a little over twice as fast as my PowerMac 800. The difference is roughly equal to the CPU megahertz difference, with the G4 possibly having a very slight edge. The G4 has a little bit of an advantage over the G3, at equal megahertz.

For now, raw megahertz and memory bandwidth still seem to rule SETI. However this may be good for the G5's.

If I come up with any more, I will let you know. Until they get the stats database fixed, it is very hard to do much of anything benchmarking wise. It has been messed up for a few days.
     
Microns
Junior Member
Join Date: Mar 2002
Location: In the tall grass prairie next to a compass plant
Status: Offline
Reply With Quote
Jan 18, 2004, 07:20 PM
 
Originally posted by Microns:
The only BIONC related cross-platform benchmarking thread that I am aware of is here: http://setiboinc.ssl.berkeley.edu/ap...ad.php?id=1019
There is a little more BOINC benchmark information located here:

http://www.setiatwork.de/boinc-benchmark.htm

It is limited to Intel and AMD (no powerPC) but Linux as well as Micro$oft.
     
   
Thread Tools
 
Forum Links
Forum Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Top
Privacy Policy
All times are GMT -4. The time now is 03:35 AM.
All contents of these forums © 1995-2017 MacNN. All rights reserved.
Branding + Design: www.gesamtbild.com
vBulletin v.3.8.8 © 2000-2017, Jelsoft Enterprises Ltd.,