Does this seem right for crunch speed? (PS3, Linux)

Message boards : Number crunching : Does this seem right for crunch speed? (PS3, Linux)
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
OzzFan Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Apr 02
Posts: 15691
Credit: 84,761,841
RAC: 28
United States
Message 927836 - Posted: 21 Aug 2009, 22:23:57 UTC - in response to Message 927825.  

How would one know what the "right" version is?


All current versions are publicly posted here.

Is this saying that 5.00 isn't supported anymore?


Its saying that you have the older v5.00 app, and that if any workunits come through that are required by that app, then you'll be able to work on them. But if a workunit requires the newer v5.05 app, you won't be able to process it as it is not referenced in your app_info.xml file with a suitable application.

What is stopping me from just changing the version info in app_info.xml (and maybe even renaming the AP application) to 5.05?


Nothing is stopping you, technically speaking. However, if you want to make sure the work you do is compatible with the new validator, I wouldn't recommend this practice. Only a v5.05 app will output the correct results that will validate with other v5.05 apps.
ID: 927836 · Report as offensive
Profile KD [SETI.USA]
Avatar

Send message
Joined: 24 Oct 99
Posts: 460
Credit: 2,513,131
RAC: 0
United States
Message 927883 - Posted: 22 Aug 2009, 1:18:23 UTC

Thanks. That explanation makes perfect sense. I won't change it.

So are there still any 5.00 AP WU's going out?

Just wondering if it makes sense ts keep both MB and AP (5.00) going.
I'm interpreting your reply as meaning that it will only download 5.00 AP WU's (if they are available).
ID: 927883 · Report as offensive
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 927894 - Posted: 22 Aug 2009, 2:03:55 UTC - in response to Message 927883.  

Thanks. That explanation makes perfect sense. I won't change it.

So are there still any 5.00 AP WU's going out?

Just wondering if it makes sense ts keep both MB and AP (5.00) going.
I'm interpreting your reply as meaning that it will only download 5.00 AP WU's (if they are available).

No, the "astropulse" application has been deprecated, so even if there were some unresolved WUs the Feeder will not tell the Scheduler about any such tasks. The "astropulse_v5" application (which is version 5.03) is still alive and eligible for any reissues. The "astropulse_v505" application (versions 5.05 or 5.06 so far) is the target for any new AP work split now.

Note it's not the file name which matters, it's the application name. The Applications page shows the "friendly name", but so far the real name for AstroPulse has been the same except for case and using an underscore rather than a space.
                                                                Joe
ID: 927894 · Report as offensive
Profile Gecko
Volunteer tester
Avatar

Send message
Joined: 17 Nov 99
Posts: 454
Credit: 6,946,910
RAC: 47
United States
Message 927988 - Posted: 22 Aug 2009, 13:55:17 UTC

Just some clarity to previous questions.

The PS3 mulibeam App compiled as by Gaurav Khanna does use SPEs and altivec, but is basically a port of stock code w/ whatever vectoring the GCC compiler manages w/ the -altivec flag enabled. It is not based on Alex's optimized source.
Gaurav had to modify stock code to enable & use the SPEs, but that's about it.
Also be aware that Cell uses a virtual 2nd core. So, all you really have is a fully functional single core CPU w/ a virtual 2nd core (think hyper-threading) that basically has to function as a task-manager to the SPEs, main memory etc.

The ars technica articles below are quite good.

Ars technica CBE Part1
Ars technica CBE Part2

A key takeaway is this:

....the PPC core does have a VMX unit. Nonetheless, I expect this VMX to be very simple, and roughly comparable to the Altivec unit o the first G4. Everything on this processor is stripped down to the bare minimum, so don't expect a ton of VMX performance out of it, and definitely not anything comparable to the G5. Furthermore, any Altivec code written for the new G4 or G5 would have to be completely reoptimized due to inorder nature of the PPC core's issue.


The total single precision processing power of the SPEs in aggregate is tremendously powerful, but requires a complete rewrite of code to THIS specific CPU, a daunting challenge.
One reason why there are sparse cell-optimized project support.

An interesting comparison is my single core G4 @ 1.25Ghz running AKv8 runs @ same output the current CBE port.

I love the PPC arch and would have been jazzed to see a highly opt app., but realize it's a single-purpose endeavor of major undertaking.

Cheers,
Gecko.


ID: 927988 · Report as offensive
Profile KD [SETI.USA]
Avatar

Send message
Joined: 24 Oct 99
Posts: 460
Credit: 2,513,131
RAC: 0
United States
Message 928106 - Posted: 23 Aug 2009, 0:49:12 UTC

Thanks for all the replies.

Crunching GPUGrid on the PS3 now...


ID: 928106 · Report as offensive
lemonade lucy

Send message
Joined: 23 Sep 99
Posts: 73
Credit: 1,454,014
RAC: 0
United States
Message 928186 - Posted: 23 Aug 2009, 14:34:35 UTC - in response to Message 928106.  
Last modified: 23 Aug 2009, 14:37:10 UTC

apologies in advance if this posted somewhere in this thread; do not have the time to go through all the messages. certainly a helpful bunch even for the ps3.

do not use dotsch's 6.2xx boinc or any boinc that fires off two concurrently running crunchers. yes, it is the latest boinc port for ps3, but I think it was created before the ps3 apps better utilized the SPEs. if you run two at a time, i.e., use dotsch's boinc, it takes too long with the updated apps. actually, I stopped it after 4 days. from what I read, sounds like resource contention with the apps better using the SPEs.

the gpugrid app crunches in ~25 hrs and claims to utilize 6 SPEs.

you should also notice, with the gpugrid boinc/app, the single process ~150% cpu usage; dotsch's boinc, 2X ~95%.

also, it appears the ps3 is with issue freeing up the SPEs--or the apps cleaning up or something, so running two projects on the ps3 is a problem. significant performance decline switching between projects.

lastly, looking for more ps3 options: yoyo@home has a good thing going but I continued to experience WUs fading to 1-3%--just on the ps3; those WUs took an inordinate time with minimal credit, i.e., problem. it is for these problems I turned to gpugrid and if dissatisfied you might think about yoyo@home. maybe they have that issue resolved; gpugrid is working and steady...so far.
ID: 928186 · Report as offensive
Profile KD [SETI.USA]
Avatar

Send message
Joined: 24 Oct 99
Posts: 460
Credit: 2,513,131
RAC: 0
United States
Message 928312 - Posted: 24 Aug 2009, 1:02:07 UTC
Last modified: 24 Aug 2009, 1:24:02 UTC

My PS3 just made a little under 4,000 credits on GPUGrid in about 24 hours.
RAC already jumped to 366 on first completed WU and based on other PS3's, will eventually level off around 4,500. Too bad it couldn't be on SETI. Going to be a nice RAC jump in Boinc Total though.

My PS3 over there:

http://www.gpugrid.net/show_host_detail.php?hostid=47992

.
ID: 928312 · Report as offensive
lemonade lucy

Send message
Joined: 23 Sep 99
Posts: 73
Credit: 1,454,014
RAC: 0
United States
Message 936672 - Posted: 29 Sep 2009, 13:19:10 UTC - in response to Message 928312.  

so much for that: that iffy bunch, gpugrid, have stopped supporting ps3 and their justification mainly fueled by the slim ps3 not, yet, other os configurable. lame and rather shocking and hit a nerve. come on, it is not that old and still a potent cruncher and the much discussed 3.0update does not impact the fatps3 other os option. ok guys, sorry, you are out. goes to show that not all projects are as committed to providing an opportunity to crunch as do the folk behind seti--I cannot imagine that these people would drop segments of its audience like this especially with such a good box. maybe they do and I missed it but I have some far-from-as-potent junk running seti. as mentioned, however, it is a shame no viable ps3 port for this project, but with gpu taking over it makes less sense to go there than ever.
ID: 936672 · Report as offensive
Profile KD [SETI.USA]
Avatar

Send message
Joined: 24 Oct 99
Posts: 460
Credit: 2,513,131
RAC: 0
United States
Message 936737 - Posted: 29 Sep 2009, 22:47:25 UTC
Last modified: 29 Sep 2009, 22:48:10 UTC

Yep. I was also disappointed in GPUGrid dropping the PS3 too. Goodbye GPUGrid, hardly even knew ya. Moved the PS3 over to PrimeGrid.
ID: 936737 · Report as offensive
Profile Reuben Gathright
Avatar

Send message
Joined: 8 Mar 01
Posts: 213
Credit: 14,594,579
RAC: 0
United States
Message 936747 - Posted: 29 Sep 2009, 23:20:54 UTC

ID: 936747 · Report as offensive
Profile KD [SETI.USA]
Avatar

Send message
Joined: 24 Oct 99
Posts: 460
Credit: 2,513,131
RAC: 0
United States
Message 936754 - Posted: 30 Sep 2009, 0:02:56 UTC - in response to Message 936747.  
Last modified: 30 Sep 2009, 0:09:13 UTC

Great thread here, I certainly want a PS3 now!


LOL, I'm sure you are being sarcastic, but..

...the new "slimline" PS3's that are soon to be sold (if they aren't already?) won't be able to run alternate OS's. Sony is completely dropping support. There won't be an option to load or boot from an alternate OS in the XMB. Running Linux on the PS3 will become a proof-of-concept "sidenote" for the PS3 just as it was for the PS2. The cynical claim that Sony initially provided Linux support for the PS3 (and PS2) so that they could evade certain European taxes on game consoles by claiming that the PS3/PS2 were "computers" since they could run a computer OS.

Those of us with current PS3's are expecting one of the Sony updates to eventually break the ability to run Linux. I know I will be googling to make sure that doesn't happen before I update from now on. Just updated to 3.01 the other day and it hasn't happened yet. If (and when) it does happen, I'm sure someone will figure out a crack. I wouldn't be surprised if someone hasn't already for the new "slimline" PS3.

I only used the PS3 as a BluRay player (it made better $ sense to get a PS3 than a standalone BluRay player at the time), but got a standalone BluRay player once the prices came down, so the PS3's only purpose for me was just to crunch. I also found some other piddly uses for it though. I.e., the PS3 is my CUPS server on my network...
ID: 936754 · Report as offensive
Profile ML1
Volunteer moderator
Volunteer tester

Send message
Joined: 25 Nov 01
Posts: 21725
Credit: 7,508,002
RAC: 20
United Kingdom
Message 936843 - Posted: 30 Sep 2009, 11:38:39 UTC - in response to Message 936754.  
Last modified: 30 Sep 2009, 11:38:52 UTC

... I only used the PS3 as a BluRay player (it made better $ sense to get a PS3 than a standalone BluRay player at the time), ... I also found some other piddly uses for it though. I.e., the PS3 is my CUPS server on my network...

What?!

You don't use it for playing games?...

:-p

Happy crunchin',
Martin
See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
ID: 936843 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 936949 - Posted: 30 Sep 2009, 18:58:59 UTC - in response to Message 936843.  

... I only used the PS3 as a BluRay player (it made better $ sense to get a PS3 than a standalone BluRay player at the time), ... I also found some other piddly uses for it though. I.e., the PS3 is my CUPS server on my network...

What?!

You don't use it for playing games?...

:-p

Happy crunchin',
Martin


WAIT! boinc isn't a game?! heh

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 936949 · Report as offensive
Previous · 1 · 2

Message boards : Number crunching : Does this seem right for crunch speed? (PS3, Linux)


 
©2025 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.