Credit
log in

Advanced search

Message boards : Number crunching : Credit

Previous · 1 · 2 · 3 · 4 · 5 · 6 . . . 10 · Next
Author Message
Profile Beyond
Avatar
Send message
Joined: 30 Jul 09
Posts: 213
Credit: 939,653,252
RAC: 3,969
Message 1644 - Posted: 21 Sep 2009, 16:07:49 UTC

Is there going to be an x86_32 v2.01 CUDA app? Still getting occasional validate errors on some WUs in the 32bit CUDA v2.00:

http://boinc.thesonntags.com/collatz/workunit.php?wuid=695367

Haven't had the problem with the x86_64 CUDA v2.01 or ATI WUs.

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 1
Message 1653 - Posted: 21 Sep 2009, 18:04:37 UTC - in response to Message 1644.

Is there going to be an x86_32 v2.01 CUDA app? Still getting occasional validate errors on some WUs in the 32bit CUDA v2.00:

http://boinc.thesonntags.com/collatz/workunit.php?wuid=695367

Haven't had the problem with the x86_64 CUDA v2.01 or ATI WUs.


Added.

Profile TomaszPawel
Avatar
Send message
Joined: 13 Jul 09
Posts: 29
Credit: 23,946,954
RAC: 0
Message 1657 - Posted: 21 Sep 2009, 18:35:47 UTC - in response to Message 1653.

Slicker,

I would like to thank you for the good 2.01 CUDA application and good number of credit you are giving for WU.

For GeForce users Collatz Conjecture is now best credit per WU project.

Thanks.

My GeForce now support Collatz Conjecture as main project!

2 day's of crunching and no errors :) Nice work!
____________
POLISH NATIONAL TEAM - Join! Crunch! Win!

Profile Beyond
Avatar
Send message
Joined: 30 Jul 09
Posts: 213
Credit: 939,653,252
RAC: 3,969
Message 1685 - Posted: 21 Sep 2009, 23:14:02 UTC - in response to Message 1653.

Is there going to be an x86_32 v2.01 CUDA app? Still getting occasional validate errors on some WUs in the 32bit CUDA v2.00:

http://boinc.thesonntags.com/collatz/workunit.php?wuid=695367

Haven't had the problem with the x86_64 CUDA v2.01 or ATI WUs.

Added.

Thanks! I can see the v2.01 WUs working their way through now. You're doing an impressive job with this project. Kudos!

Senilix
Send message
Joined: 30 Jul 09
Posts: 16
Credit: 281,188,980
RAC: 295,386
Message 1728 - Posted: 22 Sep 2009, 12:35:06 UTC

I hope it will we a while before DA detects this:



Looks like u will be urged to cut your credits by a factor of at least 5 in the near future...

Profile STE\/E
Avatar
Send message
Joined: 12 Jul 09
Posts: 581
Credit: 761,710,729
RAC: 0
Message 1731 - Posted: 22 Sep 2009, 12:41:52 UTC - in response to Message 1728.

I hope it will we a while before DA detects this:



Looks like u will be urged to cut your credits by a factor of at least 5 in the near future...


I think that was just an Anomaly caused by the 2 WU's that were Granted an Enormous amount of Credit which has been Corrected. That or just the Surge in Participants while Milkyway was down.

Profile Gipsel
Volunteer moderator
Project developer
Project tester
Send message
Joined: 2 Jul 09
Posts: 279
Credit: 77,274,766
RAC: 78,139
Message 1732 - Posted: 22 Sep 2009, 12:44:37 UTC - in response to Message 1728.
Last modified: 22 Sep 2009, 12:45:12 UTC

I hope it will we a while before DA detects this:

[..]

Looks like u will be urged to cut your credits by a factor of at least 5 in the near future...

I guess quite a lot of that surge were caused by all the ATI users switching from MW as it was down for the weekend. But it may give the project some kind of a kick start ;)

John Clark
Send message
Joined: 21 Sep 09
Posts: 548
Credit: 56,516,565
RAC: 0
Message 1737 - Posted: 22 Sep 2009, 13:00:38 UTC
Last modified: 22 Sep 2009, 13:01:23 UTC

I would support that theory Gipsel.

Now MW is back running will the surge drop back to the original point as people try and rebuild their RACs? If so, then the ATI GPU theory will be shown to be correct.

Also, there is the high paying duplex WU thingee, now corrected. However, if that gets in to the BOINC stats it remains there as a distortion (right or wrong).

Profile Gipsel
Volunteer moderator
Project developer
Project tester
Send message
Joined: 2 Jul 09
Posts: 279
Credit: 77,274,766
RAC: 78,139
Message 1741 - Posted: 22 Sep 2009, 13:25:37 UTC - in response to Message 1737.
Last modified: 22 Sep 2009, 13:28:44 UTC

Now MW is back running will the surge drop back to the original point as people try and rebuild their RACs?

I would think some of them will stay. Even if they just let a small resource share active here as a backup, it will help the project to get off the ground much faster.

Also, there is the high paying duplex WU thingee, now corrected. However, if that gets in to the BOINC stats it remains there as a distortion (right or wrong).

But that was 111 million credits (times two, as it was awarded to two guys) or something in that range. It is definitely not included in the BoincStats numbers. Guess it was fixed before the stats export was read by BoincStats.

Profile STE\/E
Avatar
Send message
Joined: 12 Jul 09
Posts: 581
Credit: 761,710,729
RAC: 0
Message 1742 - Posted: 22 Sep 2009, 13:28:59 UTC

It looks like some of the Bigger Milkyway users are hanging around to some extent today yet so we'll have to see if the Spike stays or Drops or what ...

John Clark
Send message
Joined: 21 Sep 09
Posts: 548
Credit: 56,516,565
RAC: 0
Message 1761 - Posted: 22 Sep 2009, 15:45:45 UTC

I am very sure that most of the ATI players will stay in the long run. Some may turn back to MW for a few days to rebuild, and then let BOINC Manager manage the GPU between MW and Collatz on whatever their individual project share link is.

Bymark
Avatar
Send message
Joined: 28 Jul 09
Posts: 78
Credit: 588,916,663
RAC: 1,182,288
Message 1777 - Posted: 22 Sep 2009, 16:44:55 UTC - in response to Message 1761.
Last modified: 22 Sep 2009, 16:50:23 UTC

I am very sure that most of the ATI players will stay in the long run. Some may turn back to MW for a few days to rebuild, and then let BOINC Manager manage the GPU between MW and Collatz on whatever their individual project share link is.


I am not so sure check my message:
Thats why my rac here is increasing, I have now 5 more 4850 working here:

http://boinc.thesonntags.com/collatz/forum_thread.php?id=32
____________

Profile Chris S
Avatar
Send message
Joined: 12 Jul 09
Posts: 257
Credit: 89,016,367
RAC: 0
Message 1778 - Posted: 22 Sep 2009, 16:54:20 UTC

We will see see what we see. ATI crunchers are voracious beasties, if this project offers decent returns they will stay. I'm experimenting at the moment with trying the Collatz cpu opti apps to see what they return on some of my slower machines.

Profile Slicker
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 11 Jun 09
Posts: 2525
Credit: 740,580,099
RAC: 1
Message 1783 - Posted: 22 Sep 2009, 17:01:42 UTC - in response to Message 1778.

We will see see what we see. ATI crunchers are voracious beasties, if this project offers decent returns they will stay. I'm experimenting at the moment with trying the Collatz cpu opti apps to see what they return on some of my slower machines.


Get a 64 bit op sys and you will see a huge increase on CPU apps. The v2.0 app does all 64 bit integer math so the 32 bit version is half the speed.

Profile kashi
Send message
Joined: 28 Jul 09
Posts: 164
Credit: 100,303,718
RAC: 0
Message 1784 - Posted: 22 Sep 2009, 17:04:05 UTC

The validator not working for some time at the same time as a large number of new users were joining here can also cause the credit for that day to appear much higher than it would be otherwise. I know I got a large lot of credit at once when the validator started working again after it had been down for a while. So would everyone else who had completed work that had not yet been validated and credited.

There was a fairly large increase in the number of hosts recently. Being new they would naturally be active hosts and many of them with withdrawal symptoms from missing their daily dose of MilkyWay would be very keen hosts as well:

Daily Hosts Added

I know I'll be giving some of my resources to this project as well as MilkyWay and not only when MilkyWay is down. I'm back on MilkyWay at the moment, but I'll be back here regularly after the database is moved. Besides, depending on the price, in the next few weeks I may have something I want to try out that will give a result like this:

Device 0: ATI Radeon HD 5800 (RV870) 1024 MB local RAM (remote 1855 MB cached + 1855 MB uncached)
GPU core clock: 850 MHz, memory clock: 1200 MHz
1600 shader units organized in 20 SIMDs with 16 VLIW units (5-issue), wavefront size 64 threads
supporting double precision

GPU time: 293 seconds

Haha, not that I'm a BOINC addict or anything. :)

Profile Gipsel
Volunteer moderator
Project developer
Project tester
Send message
Joined: 2 Jul 09
Posts: 279
Credit: 77,274,766
RAC: 78,139
Message 1794 - Posted: 22 Sep 2009, 17:55:37 UTC - in response to Message 1784.

depending on the price, in the next few weeks I may have something I want to try out that will give a result like this:

Device 0: ATI Radeon HD 5800 (RV870) 1024 MB local RAM (remote 1855 MB cached + 1855 MB uncached)
GPU core clock: 850 MHz, memory clock: 1200 MHz
1600 shader units organized in 20 SIMDs with 16 VLIW units (5-issue), wavefront size 64 threads
supporting double precision

GPU time: 293 seconds

Haha, not that I'm a BOINC addict or anything. :)

I seriously doubt that you will see such times if ATI hasn't managed to increase the memory controller efficiency quite a bit. For such times the memory controller would need to sustain roughly 120GB/s bandwidth with virtually random accesses to the 16MB lookup table (i.e. much larger than all caches). That's hardly going to happen. So GT300 will likely catch up (whenever it arrives).
The situation with the MW code looks better though ;) But nvidia also promised a major boost in double precision speed. But that is necessary as a HD5870 is between factor 6 and factor 12 (twelve!) times as fast as a GTX285 in double precision.

By the way, if someone sees a results of an unknown ATI card, to be exact "unknown ATI Radeon (RV700 class)", it could very well be one of those new HD5800 cards. But we will see.

Profile Chris S
Avatar
Send message
Joined: 12 Jul 09
Posts: 257
Credit: 89,016,367
RAC: 0
Message 1800 - Posted: 22 Sep 2009, 18:22:53 UTC - in response to Message 1783.

Hi Slicker, nice to make your acquaintance.

At the moment I run 1 x 4850 and 2 x 3850 cards, plus all of the rigs (10) are P4's with Win XP prof 32. From what I can see right across Boinc over a number of projects that I support, 64 bit computing is a bit of a minefield. By that I mean people keep trying various combinations of Boinc and drivers to see what works and what doesn't.

Windoze 7 plus Intel I7 have only served to muddy the waters as far as I can see. If I could get hold of a copy of Win XP 64 cheaper than the normal £95 price in the UK I might try it on one machine. I wont touch Vista with a 20 foot barge pole.

I like what I see here, I'll stick around for a while ... ;-))

____________
Why is there only one Monopolies Commission?

[SETI.USA]Tank_Master
Avatar
Send message
Joined: 12 Jul 09
Posts: 18
Credit: 35,564,306
RAC: 0
Message 1809 - Posted: 22 Sep 2009, 19:27:01 UTC

heres my win7x64 on i7(@3.6GHz) comp (with geforce 250)
http://boinc.thesonntags.com/collatz/show_host_detail.php?hostid=361

current RAC of 21k, though it probably will go higher as it runs more 2.0 apps. at least until Aqua CPU come back.... (GPU will always be Collatz)

Profile kashi
Send message
Joined: 28 Jul 09
Posts: 164
Credit: 100,303,718
RAC: 0
Message 1846 - Posted: 23 Sep 2009, 2:39:52 UTC - in response to Message 1794.


I seriously doubt that you will see such times if ATI hasn't managed to increase the memory controller efficiency quite a bit. For such times the memory controller would need to sustain roughly 120GB/s bandwidth with virtually random accesses to the 16MB lookup table (i.e. much larger than all caches). That's hardly going to happen. So GT300 will likely catch up (whenever it arrives).
The situation with the MW code looks better though ;) But nvidia also promised a major boost in double precision speed. But that is necessary as a HD5870 is between factor 6 and factor 12 (twelve!) times as fast as a GTX285 in double precision.

By the way, if someone sees a results of an unknown ATI card, to be exact "unknown ATI Radeon (RV700 class)", it could very well be one of those new HD5800 cards. But we will see.


Yes I'm a bit ahead of myself there. First prices are from AU$369 for HD 5850 and from AU$519 for HD 5870. Hmm, that means the HD 5850 is currently better crunching value. May be easier to keep cool too.

I'm thinking that there may be some setting that can be configured to suit the new cards. Something like the -k parameter on the 1.10 application. Anyway best leave speculation until someone actually has one, this thread is meant to be about credit after all.


Senilix
Send message
Joined: 30 Jul 09
Posts: 16
Credit: 281,188,980
RAC: 295,386
Message 3522 - Posted: 10 Nov 2009, 21:47:42 UTC

I noticed that since some time today all of my work units are getting 800.01 credits - instead of a variable amount depending on the number of calculated steps.

Was the credit system of Collatz changed to a fixed one?

Previous · 1 · 2 · 3 · 4 · 5 · 6 . . . 10 · Next
Post to thread

Message boards : Number crunching : Credit


Main page · Your account · Message boards


Copyright © 2018 Jon Sonntag; All rights reserved.