GNU/Linux S5R3 "power users" App 4.35 available
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4330
Credit: 251364014
RAC: 36783
22 Feb 2008 10:30:59 UTC
Topic 193528
(moderation:
Built from the same codebase as the 4.33/4.34 Apps and includes assembler-coded prefetching instructions in the "Hough" code.
It also serves as a compatibility test for a new (virtual) build machine I set up. So if you find it crashing instantly, it's probably a bad library reference.
With the larger variety of Apps it had gotten harder to write an app_info.xml that satisfies everyone's needs. You may want to take a look into the file before installing the App (i.e. restarting the client) to see if your previous App version is covered.
Only run this if you're sure of what you're doing, in particular your CPU supports SSE instructions.
Find it on the Power User Apps page.
Happy crunching!
BM
BM
GNU/Linux S5R3 "power users" App 4.35 available
)
For those who used the 64Bit app like me, I added the following to the app_info.xml right before th last tag:
einstein_S5R3
428
6.1.0
einstein_S5R3_4.35_i686-pc-linux-gnu
einstein_S5R3_4.35_graphics_i686-pc-linux-gnu
graphics_app
cu,
Michael
Same thing with app
)
Same thing with app 4.31!
I lost an already finished WU, because I forgot to add this at one host. :(
cu,
Michael
RE: Same thing with app
)
...but that's no problem because Einstein resent WU's. :-)
RE: RE: Same thing with
)
Just almost 16h crunching for nothing...
And what's the difference between new and resent WUs? ;-)
This app runs like hell, really good work! It's very fast even with the shorter WUs.
cu,
Michael
I updated the app_info.xml in
)
I updated the app_info.xml in the package (and the md5sum on the page), and also added the warning I already wrote at the 4.32 thread in the initial post.
The app_info.xml now has antries for:
HTH
BM
BM
RE: Just almost 16h
)
No difference. You will get the same WU's to crunch and past a few hours later, credits would be granted...
I think ziegenmelker was
)
I think ziegenmelker was making a rhetorical question there, rroonnaalldd. :)
No apparent problems so far, switching from 4.27 to 4.35 at 19.9%. Too early to make any judgement on speed improvements.
Soli Deo Gloria
No instant crashes on my box.
)
No instant crashes on my box. We'll see about performance. Since my two current Einstein WUs are only at 23 respectively 6 percent finished one should be able to see a difference in crunch times if the app is much faster (than my previous 4.27).
The first WU crunched pure by
)
The first WU crunched pure by 4.35 is finished.
- 24.8 % runtime
compared to what it should have taken with 4.27 following my formulas. Host is an AMD X2 5000@3,3 GHz, OpenSuse 10.3, 64 Bit.
The WU took 18,632 sec instead of ~24,760 sec.
The relative seq. no. is 23, where 0 would be the trough.
This is a great improvement! Congrats to everybody who takes part in this enhancement!
My AMD XP 3000+ will take some more time to finish its first WU, but its also running very noticeable faster.
cu,
Michael
I have 4 finished results on
)
I have 4 finished results on my X2 now with an average speed increase of 34.6 %.
Speed means [credits/s] or better credits per hour. ;-)
My Athlon XP 3000+ also finished a WU and was 38.6 % faster, but in this case the calculated values are not as precise because of lacking reference WUs.
One result pending, the others are valid.
Great work!
cu,
Michael