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 > Enhanced Optimized

Enhanced Optimized (Page 19)
Thread Tools
Junior Member
Join Date: Jun 2005
Status: Offline
Reply With Quote
Apr 27, 2007, 04:10 PM
 
well I have problems getting the pre-release V8 running on my 8-core mac pro

tried everything im running boinc 5.4.9 via the command line (i have tried 5.8.17 as well with less success)

using pr7

this is the error i get:

2007-04-26 12:15:26 [SETI@home] Process creation (../../projects/setiathome.berkeley.edu/seti_boinc) failed: Error -1
2007-04-26 12:15:26 [SETI@home] Unrecoverable error for result 09ja04aa.27470.10576.415910.3.150_1 (process exited with code 13 (0xd))
2007-04-26 12:15:26 [SETI@home] Unrecoverable error for result 09ja04aa.27470.10576.415910.3.150_1 (process exited with code 13 (0xd))
2007-04-26 12:15:26 [SETI@home] Deferring scheduler requests for 1 minutes and 0 seconds
2007-04-26 12:15:26 [SETI@home] Process creation (../../projects/setiathome.berkeley.edu/seti_boinc) failed: Error -1
2007-04-26 12:15:26 [---] Rescheduling CPU: application exited
2007-04-26 12:15:26 [SETI@home] Process creation (../../projects/setiathome.berkeley.edu/seti_boinc) failed: Error -1

i have checked and re-checked permissions etc but cant think of anything else to try

i also used version 5.8.17 which gave the following error:

Thu Apr 26 22:21:51 2007|SETI@home|Restarting task 12fe05ab.23077.4738.104832.3.52_1 using setiathome_enhanced version 513
Thu Apr 26 22:21:51 2007|SETI@home|Restarting task 12fe05ab.23077.4738.104832.3.68_0 using setiathome_enhanced version 513
Thu Apr 26 22:21:52 2007|SETI@home|Task 12fe05ab.23077.4738.104832.3.68_0 exited with zero status but no 'finished' file
Thu Apr 26 22:21:52 2007|SETI@home|If this happens repeatedly you may need to reset the project.

repeatedly

perhaps one of you kind gentlemen can email me the previous pre-release just to check if it just this version

i must say its rather frustrating (but damn impressive !!) watching zombie et al racing towards 10k and i'm probably gonna hit 6k :-) ho hum.... im sure i can cope...

so any ideas people ?

regards

adrian
63. (1) (b) "music" includes sounds wholly or predominantly characterised by the emission of a succession of repetitive beats
     
Junior Member
Join Date: Feb 2001
Location: Austria, Europe
Status: Offline
Reply With Quote
Apr 27, 2007, 04:27 PM
 
Originally Posted by alexkan, Apr 12, 2007, 06:23 AM View Post
... Give me a couple days to get the PowerPC builds up and running. As you may already know, there are things that I did for v7.2 for x86 that weren't in v7.1 for PPC, so these alone should provide some performance gains. It remains to be seen whether the cache tuning I did for v8 will help PPC, so that means I'll need PPC testers once the code is almost ready. For that matter, I need testers with Core Duos, since I haven't assessed the performance impact on that architecture, either.

I realize my performance tuning has been targeted more towards the leaderboards lately, but hopefully what I've learned there will soon bear fruit for everyone else as well!
Hi alex,

any news on the PPC-version?
     
Dedicated MacNNer
Join Date: Oct 2005
Location: Switzerland
Status: Offline
Reply With Quote
Apr 27, 2007, 06:37 PM
 
adrian - your problem may be related to insufficient shared memory available. See this page for details.

HTH,

Ron
     
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Apr 27, 2007, 06:43 PM
 
Originally Posted by adream View Post
i also used version 5.8.17 which gave the following error:

Thu Apr 26 22:21:51 2007|SETI@home|Restarting task 12fe05ab.23077.4738.104832.3.52_1 using setiathome_enhanced version 513
Thu Apr 26 22:21:51 2007|SETI@home|Restarting task 12fe05ab.23077.4738.104832.3.68_0 using setiathome_enhanced version 513
Thu Apr 26 22:21:52 2007|SETI@home|Task 12fe05ab.23077.4738.104832.3.68_0 exited with zero status but no 'finished' file
Thu Apr 26 22:21:52 2007|SETI@home|If this happens repeatedly you may need to reset the project.

repeatedly
This happens to mine occasionally too. Basically BOINC Manager freezes up, I get the spinning pinwheel of death, all the CPU meters zero out. It lasts for about 10 seconds, then everything resumes like normal, and the I get:

Fri Apr 27 15:33:08 2007||Can't rename client_state_next.xml to client_state.xml; check file and directory permissions
Fri Apr 27 15:33:08 2007||[error] Couldn't write state file: system rename
Fri Apr 27 15:34:08 2007|SETI@home|Task 29se04ab.22429.32801.978420.3.12_2 exited with zero status but no 'finished' file
Fri Apr 27 15:34:08 2007|SETI@home|If this happens repeatedly you may need to reset the project.
Fri Apr 27 15:34:08 2007|SETI@home|Restarting task 29se04ab.22429.32801.978420.3.12_2 using setiathome_enhanced version 513

(repeat the last three lines 7 more times, on for each WU running)

But like I said, everything resumes as if it never happened. No error'ed out WUs, no restarting the WUs from 0.
     
Junior Member
Join Date: Jun 2005
Status: Offline
Reply With Quote
Apr 27, 2007, 08:52 PM
 
ok heres the situation

i have managed to get pr6 running fine, no problems

so it looks like a weird zip/unzip problem i have tried every way of unzipping pr7 and it doesnt work...

anyone fancy emailing me their own pr7 rather than alex's zip file ?



halimedia... yes i have sorted out my shared memory problems (that was the first hurdle last week)

thanks for your help people

regards
adrian
( Last edited by adream; Apr 28, 2007 at 03:34 PM. Reason: removal of my email address :-0)
63. (1) (b) "music" includes sounds wholly or predominantly characterised by the emission of a succession of repetitive beats
     
Dedicated MacNNer
Join Date: Oct 2005
Location: Switzerland
Status: Offline
Reply With Quote
Apr 28, 2007, 02:33 AM
 
Sounds like you guys might be having permission issues, preventing certain files to be written. I'm running v5.4.9 of the CLI-client, and by manually setting permissions of my boinc directory, I have solved such issues before. Here's what worked for me:

sudo chown -R myuser:admin myboincdir
sudo chmod -R 755 myboincdir

where myboincdir is the name of the boinc client _and_ working directory. If you use sepparate boinc directories for the client and all boinc data, you may have to apply these changes to both directories. Also, it's important to note that boinc clients of the 5.8.x-variety can now setup their own unprivileged user under which to run. The script that does this takes care of any permission issues, so the above steps will not be needed.

Lastly, if you're using a v8 pre-release, make sure you adapt your app_info.xml file to include the correct name of the v8 executable.

HTH,

Ron
     
Junior Member
Join Date: Jun 2005
Status: Offline
Reply With Quote
Apr 28, 2007, 05:39 AM
 
ok problem solved here !

just needed to make the bugger executable :-)

chmod +x seti_boinc sorted it out...

should have tried this earlier but, was suffering from a mild case of stupidity... oh well

thanks for everyones help again, especially peter,zombie and ron

regards

adrian
63. (1) (b) "music" includes sounds wholly or predominantly characterised by the emission of a succession of repetitive beats
     
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Apr 28, 2007, 09:55 AM
 
FYI, for ownership, the owner should be boinc_master, not myuser. This worked for me:

cd into the S@H folder, then

sudo chown boinc_master:boinc_project *

In any case, the error I listed earlier is apparently not hurting anything. It's just annoying.
     
Dedicated MacNNer
Join Date: Oct 2005
Location: Switzerland
Status: Offline
Reply With Quote
Apr 28, 2007, 03:07 PM
 
Originally Posted by zombie67 View Post
FYI, for ownership, the owner should be boinc_master, not myuser.
The option of creating boinc-specific users and groups was only introduced with v5.8.x of the client, correct?

TIA,

Ron
     
Fresh-Faced Recruit
Join Date: Nov 2006
Status: Offline
Reply With Quote
Apr 29, 2007, 05:43 PM
 
Originally Posted by halimedia View Post
The option of creating boinc-specific users and groups was only introduced with v5.8.x of the client, correct?
AFAICT it’s not an option, rather part of a programme to improve security.
     
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Apr 29, 2007, 07:28 PM
 
It is my understanding that all file should be owned by "boinc_master" and group "boinc_project". Are you saying they should be something different? This is for 5.8.17.
     
Dedicated MacNNer
Join Date: Aug 2005
Location: Golden Valley, AZ
Status: Offline
Reply With Quote
May 1, 2007, 11:19 PM
 
My iMac just cracked into the top 1000 computers, it is at 927 and working its way up. RAC is 1435 and climbing.
     
Dedicated MacNNer
Join Date: Oct 2005
Location: Switzerland
Status: Offline
Reply With Quote
May 2, 2007, 02:30 AM
 
Originally Posted by zombie67 View Post
It is my understanding that all file should be owned by "boinc_master" and group "boinc_project". Are you saying they should be something different? This is for 5.8.17.
Nope, for any boinc client version since 5.5.4, these are the correct user and group settings, AFAIK - at least if you're running in 'sandboxed' mode. I'm not very well versed with this situation, since I'm still using 5.4.9, but it is my understanding that sandboxing the client is optional at least with the CLI-version. I don't know if this also applies to the GUI-version, though.

Here's more detailed info on the sandbox-design: Sandbox design

HTH,

Ron
     
Dedicated MacNNer
Join Date: Aug 2005
Location: Golden Valley, AZ
Status: Offline
Reply With Quote
May 9, 2007, 01:46 AM
 
My iMac peaked up at about 1836 RAC but has dropped back down to 1717 as of this post. But I have climbed all the way up to 274 on the top computer list!!
     
Baninated
Join Date: May 2007
Status: Offline
Reply With Quote
May 9, 2007, 01:58 AM
 
In a way, considering your age, the break-up was good. I guess
that's life.


_______________
Amphy
Vauxhall Tigra Specifications 2007
     
Dedicated MacNNer
Join Date: Jun 2006
Location: Chicago
Status: Offline
Reply With Quote
May 9, 2007, 10:59 AM
 
Originally Posted by DefaultName View Post
In a way, considering your age, the break-up was good. I guess
that's life.
Alright...you've got us all wondering...Which post was this in reply to?

QS
     
Mac Elite
Join Date: Apr 2000
Location: Minneapolis, MN USA
Status: Offline
Reply With Quote
May 13, 2007, 11:00 AM
 
I was wondering that myself. Maybe his old Macintosh?

Anyone hear from Alex? - it's been a month. Hope he's okay.

Interesting news though - I added another gigabyte to my G5 so I now have
3.5 gigs of ram and it literally shaved nearly 20 minutes off of my average
work unit processing time. Anyone else notice adding ram speeds things
up considerably?
     
Dedicated MacNNer
Join Date: Aug 2005
Location: Golden Valley, AZ
Status: Offline
Reply With Quote
May 13, 2007, 10:25 PM
 
It is a spammer, it is in the signature.
     
Forum Regular
Join Date: Aug 2005
Location: Cupertino, CA
Status: Offline
Reply With Quote
May 14, 2007, 02:53 AM
 
Originally Posted by Todd Madson View Post
Anyone hear from Alex? - it's been a month. Hope he's okay.
Yes, I'm still alive. Unfortunately, performance on v8 is not yet where I want it to be on PowerPC, so I'm going to take some time to investigate whether or not I'm overlooking something. I also haven't gotten around to tuning for Core Duo, and it's possible that the gains on that architecture will be modest as well.
Interesting news though - I added another gigabyte to my G5 so I now have
3.5 gigs of ram and it literally shaved nearly 20 minutes off of my average
work unit processing time. Anyone else notice adding ram speeds things
up considerably?
First of all, are you comparing work units that have the same angle range and claimed credit? I seem to remember pointing this out in the past, but comparing two WUs with different angle range is essentially meaningless. Comparing average WU completion time is similarly meaningless, since the distribution of angle ranges in the work that SETI sends out also varies over time.

If you're seeing a difference on similar WUs, the only case that I can think of where adding RAM would help is if it increases available memory bandwidth. If your DIMMs weren't installed in matched pairs previously and they are now, it's possible that you've gone from single-channel to dual-channel mode.
     
Dedicated MacNNer
Join Date: Nov 2004
Location: Virginia
Status: Offline
Reply With Quote
May 18, 2007, 01:19 PM
 
Alex: Here's a validate error: http://setiathome.berkeley.edu/resul...ltid=533097235
First one I've received in a long time. MacBookPro Core Duo, 2 GB RAM, OS X 10.4.9, running Alex Khan's seti_enhanced-i386-v7.2-coreduo-nographics via BOINC Manager 5.4.9

C
     
Junior Member
Join Date: Feb 2001
Location: Austria, Europe
Status: Offline
Reply With Quote
May 18, 2007, 03:26 PM
 
There seems to be a problem with the app_info.xml file that prevents BOINC from downloading new work: [FIX] For those using optimized SETI app and can't get new work
     
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
May 18, 2007, 05:32 PM
 
Originally Posted by Gulliver64 View Post
There seems to be a problem with the app_info.xml file that prevents BOINC from downloading new work: [FIX] For those using optimized SETI app and can't get new work
Or even here:

http://forums.macnn.com/72/team-macn...2/#post3383563

     
Dedicated MacNNer
Join Date: Aug 2005
Location: Golden Valley, AZ
Status: Offline
Reply With Quote
May 19, 2007, 12:09 AM
 
So far so good on 3 of 4 machines.

Of course my iMac is almost done with that initial run of work and I will have to do it again tonight.
     
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
May 19, 2007, 12:15 AM
 
If you are using 5.9.11, you can use the field in your account settings for an additional 1-10 days of work. I have mine set to .1 (connect time) + 5 additional days. Just be careful. It is possible to get more than you can return before the deadline. For example, if you somehow downloaded 5 days worth of WUs, all of which have a 4 day deadline.
     
Dedicated MacNNer
Join Date: Aug 2005
Location: Golden Valley, AZ
Status: Offline
Reply With Quote
May 19, 2007, 01:43 AM
 
I am on dial-up and every time I try to download one of the betas I can only get about 200 k and then it says it has finished the download.

I am still using 5.8.17 on my 4 computers.
     
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
May 24, 2007, 03:46 PM
 
Looks like I have some competition from adream (team MacCafe):

Top computers

He is also running a pre-release version of alexkan's application. Not sure if that's intentional, or if we have a leak...
     
Junior Member
Join Date: Nov 2003
Status: Offline
Reply With Quote
May 24, 2007, 04:38 PM
 
Speaking of Alexkan's pre-release versions... anyone hear from Alex lately? My Quad-core MacPro is itching to get some of that action (nope, didn't contact him about joining in the Beta festivities, real life interferes with my ability to give him meaningful stats on a regular basis).
----------------------------------------------------
Jedi's Lair: Reviews, Tips, and the RickyCam
----------------------------------------------------
Jedi's Photos: Living life one shutter click at a time...
     
Junior Member
Join Date: Jun 2005
Status: Offline
Reply With Quote
May 24, 2007, 06:53 PM
 
he heh... not a leak :-)

i got the same client from alex that you got.... and very grateful too... the battle between our machines is a close one eh? as we have identical gear (mine has 5gb or ram)

the reason i dont crunch seti for macnn anymore is that i set MacCafe up years ago and once people joined i felt obliged to return to the team and crunch there, its a friendly UK based mac yahoo group

so whos the money on, which of the two machines will settle out highest? is there a few more coming up behind us, either way the top 20 should be all mac pretty soon i reckon :-) and thats great !

hope to see alexs client out soon , all depending on how the ppc code is going..

regards

adream
63. (1) (b) "music" includes sounds wholly or predominantly characterised by the emission of a succession of repetitive beats
     
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
May 24, 2007, 08:38 PM
 
Cool. Competition is always good! FWIW, mine is not a dedicated cruncher. I use it as my daily work/play machine. Nothing strenuous. Just email/web/NeoOffice/iTunes/etc.

Speaking of competition...Where's E.T from Tellus?? He was #1 before the shutdown. Now I think the few credits/day he is getting are just due to pending results that finally validate. No new crunching going on. Someone needs to tell him that SETI is back up!
     
Dedicated MacNNer
Join Date: Aug 2005
Location: Golden Valley, AZ
Status: Offline
Reply With Quote
May 24, 2007, 10:58 PM
 
When I get back from my vacation I hope that I will be able to get my iMac back crunching full time.
     
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
May 25, 2007, 01:56 AM
 
Tick Tick Tick....
     
Dedicated MacNNer
Join Date: Aug 2005
Location: Golden Valley, AZ
Status: Offline
Reply With Quote
May 25, 2007, 01:59 AM
 
I won't be back to my computer until Monday evening. At least I have my laptop with me.
     
Fresh-Faced Recruit
Join Date: May 2007
Status: Offline
Reply With Quote
May 25, 2007, 10:58 AM
 
Hello, my brother have a MACpro 2,66ghz and he use the last intel app of Alexkan for SETI;
His WU are about 6800 seconds for 61 credits i think that is slow because i see in statistics of SETI MAcpro 3ghz complete WU in 3300 seconds for 61 credits .

So , have we got other optimization to do ? does there need wisdom file for MACpro like my PowermacG5 ? or other optimization ?
     
Dedicated MacNNer
Join Date: Aug 2005
Location: Golden Valley, AZ
Status: Offline
Reply With Quote
May 31, 2007, 01:30 AM
 
Looks like things are starting to crank back up, now all I need is a nice beta Alexkan worker for my iMac.
     
Fresh-Faced Recruit
Join Date: May 2007
Status: Offline
Reply With Quote
May 31, 2007, 06:51 AM
 
No there is nothing that work in the same time ...
so i don't understand why there is this large différence between same machine with same worker
     
Fresh-Faced Recruit
Join Date: May 2007
Status: Offline
Reply With Quote
Jun 4, 2007, 11:00 AM
 
So no other optimization ?
i'm surprized my G5 is faster at equal clock frequency than Xeon although x86 app is compiled with intel compiler instead gcc like for powerpc !
     
Forum Regular
Join Date: Aug 2005
Location: Cupertino, CA
Status: Offline
Reply With Quote
Jun 4, 2007, 12:22 PM
 
Originally Posted by CRISTOBOOL View Post
Hello, my brother have a MACpro 2,66ghz and he use the last intel app of Alexkan for SETI;
His WU are about 6800 seconds for 61 credits i think that is slow because i see in statistics of SETI MAcpro 3ghz complete WU in 3300 seconds for 61 credits .

So , have we got other optimization to do ? does there need wisdom file for MACpro like my PowermacG5 ? or other optimization ?
Make sure your brother's Mac Pro has all four memory channels occupied with matching FB-DIMMs. SETI is sensitive to memory bandwidth, so having only two of the four channels populated (which is often the case with stock Mac Pros) is disastrous for crunching performance.

Also, make sure you're looking at times for machines running v7.2, instead of those running the v8 prerelease. v7.2 machines are more likely to be in the upper 3000s for 61-credit WUs. (No, v8 isn't ready yet, and won't be until I understand what PowerPC needs to get the same sorts of performance boosts.)
     
Fresh-Faced Recruit
Join Date: May 2007
Status: Offline
Reply With Quote
Jun 4, 2007, 02:15 PM
 
ok thanks ; there are only 1 Go of RAM so 2 FB-DIMMs on the macpro and the G5 have 1,5Go with 4 FB-DIMMs.
     
Junior Member
Join Date: Jun 2006
Status: Offline
Reply With Quote
Jun 5, 2007, 04:34 AM
 
Originally Posted by alexkan View Post
(No, v8 isn't ready yet, and won't be until I understand what PowerPC needs to get the same sorts of performance boosts.)
Let me start off by saying that I could not code my way out of a wet paper bag. Also, I am a PPC Luddite (had four PPC Macs, just added an upgraded PCI PM) and think it's great that you are wanting to give both platforms equal gains in performance.

Having said that, is it realistic to expect the PPC version to realize the same gains that x86 yields? Are there not limitations inherent in every design? I mean look at what's being done with the Cell processor and GPUs, I never expected to see a video game console doing DC, much less doing it better than a PC can even if it's abilities are rather narrow.

I guess my question is, as noble as your efforts are, is the Team MacNN effort and DC as a whole, being held back your desire to keep the x86 and PPC workers "equal"?

Chris

Please know that I mean no offense whatsoever. I have nothing but the highest admiration for your altruistic efforts.
     
Fresh-Faced Recruit
Join Date: May 2007
Status: Offline
Reply With Quote
Jun 5, 2007, 07:01 AM
 
I think that without another compiler (here gcc) we can't expect large gains for the PowerPC app (because alexkan have done a very good work in optimizing the code); the best compiler for PowerPC on Mac OS X should be the IBM XL C/C++ compiler, maybe with this compiler we can expect good gains for PowerPC app.
For the x86 app ; the best compiler is the intel compiler of course , already used by alexkan.
     
Mac Elite
Join Date: Apr 2000
Location: Minneapolis, MN USA
Status: Offline
Reply With Quote
Jun 5, 2007, 04:47 PM
 
I suspect one of the issues with the PPC G5 design was (in the initial iteration) small L2 cache size.

The later iteration had a larger cache and produced generally better performance, even with
slower clock speed (for example, old 2.5/2.7 G5s versus 2.0/2.3 DC versions). Those later
iterations also had faster memory as well which probably helped some.
     
Forum Regular
Join Date: Aug 2005
Location: Cupertino, CA
Status: Offline
Reply With Quote
Jun 6, 2007, 02:35 AM
 
Originally Posted by ChillyWilly5280 View Post
Having said that, is it realistic to expect the PPC version to realize the same gains that x86 yields? Are there not limitations inherent in every design? I mean look at what's being done with the Cell processor and GPUs, I never expected to see a video game console doing DC, much less doing it better than a PC can even if it's abilities are rather narrow.

I guess my question is, as noble as your efforts are, is the Team MacNN effort and DC as a whole, being held back your desire to keep the x86 and PPC workers "equal"?
I think you've asking a very good question. As things are now, I actually maintain two Xcode projects in order to build with ICC for x86 and GCC for PPC, so perhaps I should just release the two Xcode projects separately. Releasing a single tarball would mean either code duplication or requiring people who download the source to fiddle with the Xcode project to get everything building the way I did. Perhaps what I should do now is get an x86 release ready first. For that I'll need Core Duo testers to see how v8 stacks up. I expect the gains to be more modest than on v8, but I don't think there's that much more I can do about. Same rules apply as before--PM me if you have the hardware in question, know how to take Shark traces, and can compile a table of WU times for different ARs on your machine.

As for the other questions you posed, I think what I said about hoping for the PPC version to get the same sort of performance gain was misleading, since there are certain performance gains that are very specific to Core2. Todd is right to point out the effects of L2 cache on performance within the G5 line, but these effects extend to x86 as well--one of the bigger wins in v8 for Core2 is only possible because its L2 is large enough--even Core Duos have too little L2 to benefit. However, there are still some hot loops on PPC which are not as well-tuned as I originally thought, and this is what I'm currently working on fixing.

Something to think about, though, if you're concerned about Team MacNN and DC efforts: if you're crunching for SETI, you're probably best served by going all-Core2, both in terms of RAC per machine and RAC per watt. I suspect G5s don't fare too well in the RAC-per-watt metric, and G4s really don't fare well in either metric.

Still, if you're going to crunch with these machines, I might as well make them run as fast as they can.
     
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Jun 6, 2007, 03:10 AM
 
I'd provide my Core Duo 17" MBP, but both cores are already allocated to CPDN. We all know that's too long already.

Originally Posted by alexkan View Post
Something to think about, though, if you're concerned about Team MacNN and DC efforts: if you're crunching for SETI, you're probably best served by going all-Core2, both in terms of RAC per machine and RAC per watt. I suspect G5s don't fare too well in the RAC-per-watt metric, and G4s really don't fare well in either metric.
On the other hand, all PPC machines are even *worse* on every other project. So while gains due to your applications aren't as much with PPC, it is still better than everything else!

So THANKS!

Now, can anyone explain to me why my RAC is 500-600 less than another machine just like mine?!
     
Dedicated MacNNer
Join Date: Aug 2005
Location: Golden Valley, AZ
Status: Offline
Reply With Quote
Jun 6, 2007, 09:32 AM
 
We cannot even tell what speed the processor is in adreams Mac Pro, he might have replaced the original ones with upgrades.

Looks like I will not be able to participate as I have no clue what a shark trace is, I can do the table work though.
     
Junior Member
Join Date: Jun 2005
Status: Offline
Reply With Quote
Jun 6, 2007, 10:08 AM
 
my 8-core is a standard mac pro xeon machine with 5GB ram

im using 5.4.9 as the boinc manager (cos its alot less fussy about the location of the boinc data folder). unfortunately it doesn't report the processors like the later versions of boinc

regards

adream
63. (1) (b) "music" includes sounds wholly or predominantly characterised by the emission of a succession of repetitive beats
     
Senior User
Join Date: Jun 2006
Location: Dublin, CA
Status: Offline
Reply With Quote
Jun 6, 2007, 11:29 AM
 
Originally Posted by arkayn View Post
We cannot even tell what speed the processor is in adreams Mac Pro, he might have replaced the original ones with upgrades.
Heh. Not a chance. There is nothing to upgrade to (yet). We are already using the fastest Xeons out there (5365). In fact, you can't even buy them individually (yet).
     
Junior Member
Join Date: Jun 2005
Status: Offline
Reply With Quote
Jun 6, 2007, 11:37 AM
 
intestingly theres a v8 being reviewed here briefly

Intel power PC platforms: Core 2 Extreme QX6800 and V8 | Reg Hardware
63. (1) (b) "music" includes sounds wholly or predominantly characterised by the emission of a succession of repetitive beats
     
Mac Elite
Join Date: Apr 2000
Location: Minneapolis, MN USA
Status: Offline
Reply With Quote
Jun 7, 2007, 11:19 AM
 
Has anyone tried the new Alex client on a Core2 Extreme chip? I'd be real curious to see how that
performance fares vs the Xeons of the same cache size. I realize the Extremes are optimized for
gaming but I suspect it would do better than Core2Duo possibly, possibly not.
     
Dedicated MacNNer
Join Date: Aug 2005
Location: Golden Valley, AZ
Status: Offline
Reply With Quote
Jun 13, 2007, 12:01 AM
 
I think my iMac has caught back up to where it was over a month ago, now lets see how much farther it will go.

Currently sitting at 412th on the list.
     
Forum Regular
Join Date: Oct 2005
Location: Las Vegas, NV
Status: Offline
Reply With Quote
Jun 17, 2007, 12:14 PM
 
Originally Posted by alexkan View Post
I think you've asking a very good question. As things are now, I actually maintain two Xcode projects in order to build with ICC for x86 and GCC for PPC, so perhaps I should just release the two Xcode projects separately.
Hi Alex,

Curious if ICC 10.0 is making it any easier to work between x86 and PPC or are many of your improvements/approaches are already beyond the basic features and conveniences of 10.0 to where it's not so much of a help.

Best regards,
Ian
     
 
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
Trackbacks are On
Pingbacks are On
Refbacks are On
Top
Privacy Policy
All times are GMT -4. The time now is 04:51 PM.
All contents of these forums © 1995-2015 MacNN. All rights reserved.
Branding + Design: www.gesamtbild.com
vBulletin v.3.8.8 © 2000-2015, Jelsoft Enterprises Ltd., Content Relevant URLs by vBSEO 3.3.2