5.12 Completion Times
Message boards :
Number crunching :
5.12 Completion Times
Message board moderation
Previous · 1 · 2 · 3 · Next
| Author | Message |
|---|---|
W-K 666  Send message Joined: 18 May 99 Posts: 13795 Credit: 40,757,560 RAC: 151 
 |
I'll let you know soon, my cashe has 12 of the 4.18 versions left, and about 15 of the enhanced version units. I currently complete one of the 4.18 versions between 2.0 hours and 2.2 hours...so we will see what happens...... When comparing enhanced times you must look at the AR, without that info you could think that the optimisation is doing nothing or is out of this world. For some guidance see Inglesides Deadlines post Andy ID: 302446 ·  |
 RePhLeX Send message Joined: 3 Dec 05 Posts: 76 Credit: 128,962 RAC: 0
|
I'll let you know soon, my cashe has 12 of the 4.18 versions left, and about 15 of the enhanced version units. I currently complete one of the 4.18 versions between 2.0 hours and 2.2 hours...so we will see what happens...... Its gonna be a big increase in time then. My 4.18s take 40ish minutes currently. The new enhanced ones so far have been taking 5+hours. I have just put crunch3rs opt client on so i will post new times asap but i dont think they will fall by much..if they fall by half, thats still quite poor in comparison to the 4.18s. ID: 302442 · |
sterling0466  Send message Joined: 5 Oct 00 Posts: 204 Credit: 742,621 RAC: 0 
|
I'll let you know soon, my cashe has 12 of the 4.18 versions left, and about 15 of the enhanced version units. I currently complete one of the 4.18 versions between 2.0 hours and 2.2 hours...so we will see what happens...... ID: 302358 · |
Pappa  Send message Joined: 9 Jan 00 Posts: 2562 Credit: 12,301,681 RAC: 0
|
sterling0466 I have no knowledge of your FX-51, I would say let it run and then let us know how it went.... We would have loved to have it in Enhanced Beta... I guess the bottom line is that the orignal estamates for completion times were 10 times a long! Through the effort of the Volunteers that has bee reduced to about 2-4 times as long depending on the Angle Range and the computer... Optimized Applications aside... Credits until "everyone" is crunching Enhanced is going to look like a Roller Coaster Ride! Pappa O.k., so what does this mean for us with an FX-51 64 bit processor? Should we see slower or faster times since these processors are dual core and have multiple avenues to transfer and complete calculations? Just wondering since my cashe is almost empty of the standard workunits and starting to fill with the new enhanced workunits. Please consider a Donation to the Seti Project. ID: 302343 · |
zoom3+1=4  Send message Joined: 30 Nov 03 Posts: 63167 Credit: 55,293,173 RAC: 111
|
Well I don't know about the FX cpu as I just run Opterons(Good Luck), My 1st fully completed Experimental Work Unit took a little over 3 hours Now they are completing in 2:04:00 and 2:02:00 or so, I'd changed My mind as Joker2 was out of 4.18 units(Although Joker3 is getting all It wants for some reason, Go figure ;)) as I found Crunch3r's 5.12 SSE3 app for the Athlon64 cpus and decided to give It a try, Ok It's 3 times the processing time and I guess the same amount of credits or something, I wish someone could put that one to sleep. Mind You Joker2 uses an Opteron 165 that is running at 2.65GHz and It buzzsaws through Predictor in about 26 Minutes and Einstein in about 32 Minutes. Oh and My Boinc setup is BOINC client version 5.3.12.tx36 for windows_intelx86 overlaid on top of Boinc 5.3.16 with Truxes Calibrating functions turned on only for Einstein@Home,Predictor @ Home(That is how It's spelled out in His truxoft_prefs.xml file). Seti doesn't get any calibration with 5.12 at all of course. Who says It isn't customizeable under Windows XP x32 or x64? ;) I run both OSes. truxoft_prefs.xml Contents: <truXoft_conf> <calibrate_credit>Einstein@Home,Predictor @ Home</calibrate_credit> </truXoft_conf> Oh and I know that the reason for not calibrating under Seti enhanced is that It will mess the units up. Maybe Crunch3r can optimize It futher on the app to cut the time down to nearly an hour, I'd hope. As now It's taking 3 times as long to crunch a WU. My Amazon Wishlist The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's ID: 302278 · |
|
sterling0466 Send message Joined: 5 Oct 00 Posts: 204 Credit: 742,621 RAC: 0
|
O.k., so what does this mean for us with an FX-51 64 bit processor? Should we see slower or faster times since these processors are dual core and have multiple avenues to transfer and complete calculations? Just wondering since my cashe is almost empty of the standard workunits and starting to fill with the new enhanced workunits. ID: 302247 · |
bono_vox  Send message Joined: 17 Jul 02 Posts: 13 Credit: 223,782 RAC: 0 
|
The credits are linked to the number of Floating Point OPerations (Fpops) performed, the number of Fpos varies with the Angle_range (AR) of the unit. Over time you will probably get units with high AR >=1 that crunch fast, less Fpops, and will only get 10 credits, and others that take longer (Low AR), more Fpops, and receive 80 credits. Thanks for the information. bono_vox 
ID: 300906 · |
|
bono_vox Send message Joined: 17 Jul 02 Posts: 13 Credit: 223,782 RAC: 0
|
NOT 60-62 credits, It uses a "flops count", so if it takes more flops (floating point operations) you'll get more credit, less flops, less credit, but all who crunch it should be claiming the same credit. OK now I get it. Thanks for the clarification. In fact my newer WUs took less time (and most likely less flops) resulting in fewer GCs. bono_vox
ID: 300903 · |
|
W-K 666 Send message Joined: 18 May 99 Posts: 13795 Credit: 40,757,560 RAC: 151
|
If I understand correctly, no matter what is your computer and how long it takes to process one of these enhanced WUs, you'll get around 60-62 credits? The credits are linked to the number of Floating Point OPerations (Fpops) performed, the number of Fpos varies with the Angle_range (AR) of the unit. Over time you will probably get units with high AR >=1 that crunch fast, less Fpops, and will only get 10 credits, and others that take longer (Low AR), more Fpops, and receive 80 credits. ID: 300898 · |
Astro  Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
If I understand correctly, no matter what is your computer and how long it takes to process one of these enhanced WUs, you'll get around 60-62 credits? NOT 60-62 credits, It uses a "flops count", so if it takes more flops (floating point operations) you'll get more credit, less flops, less credit, but all who crunch it should be claiming the same credit. ID: 300888 · |
|
bono_vox Send message Joined: 17 Jul 02 Posts: 13 Credit: 223,782 RAC: 0
|
If I understand correctly, no matter what is your computer and how long it takes to process one of these enhanced WUs, you'll get around 60-62 credits? Example: wuid=76866800 320145203 1535451 7 May 2006 1:16:47 UTC 8 May 2006 1:42:51 UTC Over Success Done 15,129.31 62.36 62.36 320145204 1771775 7 May 2006 1:16:44 UTC 8 May 2006 18:33:32 UTC Over Success Done 139,429.76 62.36 62.36 320601597 456873 7 May 2006 15:30:49 UTC 8 May 2006 4:37:20 UTC Over Success Done 29,699.06 62.37 62.36 Note the CPU Times (15K, 29K, 139K) all receiving the same amount of credit. bono_vox
ID: 300882 · |
|
Idefix Send message Joined: 7 Sep 99 Posts: 154 Credit: 482,193 RAC: 0 
|
Hi Pappa, It appears that 5.2.13tx36 is attempting to calibrate it... Even though it is only set for "seti@home, einstein@home" and NOT seti_enhanced@home... Of course... The <calibrate_credit>-line in trux's configuration file has *nothing* to do with the name of the application. The name of the *project* is added there. And the project name is still "seti@home". As others already stated, the fpops count is left untouched. And that's the important number for the credit claim in seti_enhanced. So, you don't mess up anything by calibrating seti_enhanced. The only problem will be that you mess up credit claims if you ever run the old 4.18 again, because seti and seti_enhanced need different DCFs (which is used for the calibration). Regards, Carsten ID: 300835 · |
|
RePhLeX Send message Joined: 3 Dec 05 Posts: 76 Credit: 128,962 RAC: 0
|
from 3000 secs to 15000secs per wu.. from 30 cc per wu, to 60. You do the math.. I used to love it when i left my pc running overnight, and got up to see i had crunched maybe 25 wus..was great..now i crunch maybe 3-4.... :-( Gonna get rid of all my enhanced and stick with standard until opt client is released. ID: 300111 · |
Steve Cressman  Send message Joined: 6 Jun 02 Posts: 583 Credit: 65,644 RAC: 0 
|
I have been running mine with calibration on at seti beta for 6 weeks now to test what would happen. And every time it completes a result it claims exactly the same as those that where using the 5.2.13 recomended version. It claims the same amount because Fpops are still counted and that is the value used to make a claim. 98SE XP2500+ @ 2.1 GHz Boinc v5.8.8  And God said"Let there be light."But then the program crashed because he was trying to access the 'light' property of a NULL universe pointer. ID: 299898 · |
|
W-K 666 Send message Joined: 18 May 99 Posts: 13795 Credit: 40,757,560 RAC: 151
|
The only remaining question that comes up is whether or not Trux's BOINC client messes up the calibration/credit claim for the new Seti enhanced? I would have thought the problem is the other way round. As The latest of Trux's clients uses Fpops to calculate credits for enhanced, but I understand that it uses DCF to do the calibration, and on my computer the DCF has gone up from 0.13 to 0.55 Andy ID: 299712 · |
|
Ingleside Send message Joined: 4 Feb 03 Posts: 1546 Credit: 15,832,022 RAC: 29 
 |
The only remaining question that comes up is whether or not Trux's BOINC client messes up the calibration/credit claim for the new Seti enhanced? No, as long as you're running BOINC core client v5.2.6 or later it doesn't matter for Seti_Enhanced whatether your benchmark/cpu-time is being changed to by Trux or any other "optimized" BOINC-clients. ID: 299701 · |
Stefan  Send message Joined: 12 Nov 05 Posts: 226 Credit: 213,560 RAC: 0
|
Stefan Thank you for the reply, it makes things so much clearer. Guess I'll be hitting that no new work button for sure and uninstalling the 5.11, no need to create a mess of credits for a half hour reduction in time, until the finalized opt. app comes out. The only remaining question that comes up is whether or not Trux's BOINC client messes up the calibration/credit claim for the new Seti enhanced? Human Stupidity Is Infinite... 
ID: 299672 · |
Gecko  Send message Joined: 17 Nov 99 Posts: 454 Credit: 6,946,910 RAC: 107
|
Pappa, I've just completed migration to enhanced and downloaded and extracted Crunch3r's opt 5.11 ap. Also edited the app_info.xml file to reflect 5.11 instead of 5.12. Seems to be working fine so far. If I have correctly pieced together the results of everyones efforts in the several threads on the forum, then I want to turn-off calibration on my 5.2.12.tx36 client, correct? Is this as simple as editing the app_info.xml with "calibrate_credit = off". Thanks for you help and a HUGE THANKS to everyone who has diligently worked to get us humble crunchers this far! Regards, Ian ID: 299412 · |
|
Pappa Send message Joined: 9 Jan 00 Posts: 2562 Credit: 12,301,681 RAC: 0
|
Stefan Obviously some people get a bit greedy and want a peek at what will be... My own opinion is that I have had access to those files for a long time (me crunch3r beta tester anyone is welcome to look at my machines here or in Beta). So while looking the Optimized Enhanced Application is Solid... IT is not the final Optimized Enhanced Application that will be here as Crunch3r upgrades his compiler... A month from now the Enhanced Application could have changed once or three times... So running Optimized 5.11 may cause you to change your computer or computers as many times as the changes happen and Crunch3r can catch up with it... Now to be perfectly honest, I will be testing the 5.11 to confirm the app_info.xml that has to be created to move back and forth... Yes I know that it will cause credit headaches... When I do that I will put in place and get work and then say "No New Work" or "No New Tasks" (depending on the version of the BOINC Core Client) and then I have have a machine sit idol until I an there to deal with it... That is part of testing... For those that do that RAC looks like a yoyo... Recommendation: Say no new work and let you cache rundown and then remove it... Are you hurting anything? One of the things that was happening was a credit analysis (several people say that we should be receiving more credit) of Seti BOINC, and Seti Enhanced to insure the correct amount of credit was assigned... That might be hampering that process... So people will do what they will do... At times it is hard to say this person is right and that person is wrong... We know that Optimized Enhanced 5.11 works... You have also helped to prove that... So the choice becomes yours... Pappa Stefan Please consider a Donation to the Seti Project. ID: 299321 · |
|
Stefan Send message Joined: 12 Nov 05 Posts: 226 Credit: 213,560 RAC: 0
|
Stefan I was under the impression that the 5.11 file was for seti enhanced, at least by browsing the Crunch3r site, nevertheless, should I get rid of it or keep it or just wait a bit or what? All this talk about Credit Soup makes me hope this doesn't turn out like the Marx Brother's movie...Duck Soup ;) I guess time, patience, and ingeniuty on someone's part will eventually wade through the mess. All for a good cause though! Human Stupidity Is Infinite...
ID: 299310 · |
©2020 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.