more credits :)
log in

Advanced search

Message boards : Number crunching : more credits :)

1 · 2 · 3 · 4 · Next
Author Message
Profile speedimic
Avatar
Send message
Joined: 22 Feb 08
Posts: 260
Credit: 57,387,048
RAC: 0

Message 36922 - Posted: 4 Mar 2010, 9:15:58 UTC

Usually WUs on my new 5850 take ~2min - now I got some like this in the mix...
Is that intended?

just in case it's purged too quickly: de_14_3s_const_3_391634_1267691174_0
wall clock time: 17.176 seconds
Granted credit 213.759564

____________
mic.


Profile The Gas Giant
Avatar
Send message
Joined: 24 Dec 07
Posts: 1947
Credit: 240,865,573
RAC: 0

Message 36923 - Posted: 4 Mar 2010, 11:16:19 UTC

Shhhhh....they might change the credits.

Profile Vincent JG
Avatar
Send message
Joined: 25 Jan 10
Posts: 127
Credit: 77,983,585
RAC: 39

Message 36924 - Posted: 4 Mar 2010, 11:19:43 UTC - in response to Message 36922.

I just started getting those as well since last night...right after I replaced the GTX260 with a HD5850. Was a little worried until I saw the HD4850 crunching them as well.

About 20 seconds on the HD5850 and around 45 seconds on the HD4850.
____________
"Government big enough to supply everything you need is big enough to take everything you have."
-Thomas Jefferson
American Thinker

John Clark
Send message
Joined: 4 Oct 08
Posts: 1734
Credit: 64,228,409
RAC: 0

Message 36949 - Posted: 4 Mar 2010, 22:50:44 UTC

I thought I would try and get some of these, but they appear to have gone away.

Anyway, I am crunching, but only 1 WU at a time. This is despite my app_info file reading as follows -

<file_info>
<name>astronomy_0.20b_ATI_SSE2_ati.exe</name>
<executable/>
</file_info>
<file_info>
<name>brook32.dll</name>
<executable/>
</file_info>

<app_version>
<app_name>milkyway</app_name>
<version_num>20</version_num>
<flops>1.0e11</flops>
<avg_ncpus>0.01</avg_ncpus>
<max_ncpus>1</max_ncpus>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline>n3</cmdline>
<file_ref>
<file_name>astronomy_0.20b_ATI_SSE2_ati.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>brook32.dll</file_name>
</file_ref>
</app_version>

</app_info>

I remember before crunching Collatz these last few months that I was able to crunch several Milkyway WUs at a time. I remember 1 WU gave a good RAC, but 2 WUs crunched at the same time gave much more and 3 at a time reduced the output.

Is it possible to remind me how I can get 2 WUs crunching together again, please?
____________
Go away, I was asleep


Profile speedimic
Avatar
Send message
Joined: 22 Feb 08
Posts: 260
Credit: 57,387,048
RAC: 0

Message 36950 - Posted: 4 Mar 2010, 23:04:56 UTC - in response to Message 36949.

I thought I would try and get some of these, but they appear to have gone away.

...


hmm... just got another bunch of those de_14_'s

____________
mic.


Profile Cori
Avatar
Send message
Joined: 27 Aug 07
Posts: 647
Credit: 27,592,547
RAC: 0

Message 36951 - Posted: 4 Mar 2010, 23:07:50 UTC - in response to Message 36949.

Is it possible to remind me how I can get 2 WUs crunching together again, please?


Use the count tags! Example:

<coproc>
<type>ATI</type>
<count>0.5</count>
</coproc>

;-)
____________
Lovely greetings, Cori

Profile arkayn
Avatar
Send message
Joined: 14 Feb 09
Posts: 999
Credit: 74,932,619
RAC: 0

Message 36953 - Posted: 5 Mar 2010, 0:00:01 UTC - in response to Message 36949.

I thought I would try and get some of these, but they appear to have gone away.

Anyway, I am crunching, but only 1 WU at a time. This is despite my app_info file reading as follows -

<file_info>
<name>astronomy_0.20b_ATI_SSE2_ati.exe</name>
<executable/>
</file_info>
<file_info>
<name>brook32.dll</name>
<executable/>
</file_info>

<app_version>
<app_name>milkyway</app_name>
<version_num>20</version_num>
<flops>1.0e11</flops>
<avg_ncpus>0.01</avg_ncpus>
<max_ncpus>1</max_ncpus>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline>n3</cmdline>
<file_ref>
<file_name>astronomy_0.20b_ATI_SSE2_ati.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>brook32.dll</file_name>
</file_ref>
</app_version>

</app_info>

I remember before crunching Collatz these last few months that I was able to crunch several Milkyway WUs at a time. I remember 1 WU gave a good RAC, but 2 WUs crunched at the same time gave much more and 3 at a time reduced the output.

Is it possible to remind me how I can get 2 WUs crunching together again, please?


Any more it does not matter if you crunch 1 or 2 at a time, they take an identical time.

1 at a time: 5:06
2 at a time: 10:12

No benefit at all.
____________

John Clark
Send message
Joined: 4 Oct 08
Posts: 1734
Credit: 64,228,409
RAC: 0

Message 36955 - Posted: 5 Mar 2010, 1:04:37 UTC

Thanks Cori, it works.

After running 2 consecutively for a little time I have reverted to 1, just to see how things go overnight before reverting to Collatz for a while.
____________
Go away, I was asleep


Vid Vidmar*
Avatar
Send message
Joined: 29 Aug 07
Posts: 81
Credit: 60,360,858
RAC: 0

Message 36961 - Posted: 5 Mar 2010, 10:15:05 UTC - in response to Message 36953.

...
Any more it does not matter if you crunch 1 or 2 at a time, they take an identical time.

1 at a time: 5:06
2 at a time: 10:12

No benefit at all.


Not true. If you run only 1 app at a time, the GPU will sit idle between WUs. If you run 2 WUs at a time, offset, GPU will never be idle.
BR,
____________

SkyeHunter
Send message
Joined: 6 Mar 09
Posts: 41
Credit: 38,856,291
RAC: 0

Message 36963 - Posted: 5 Mar 2010, 10:45:04 UTC - in response to Message 36922.
Last modified: 5 Mar 2010, 10:45:33 UTC

Same here, not complaining though ;o)

Profile Gavin Shaw
Avatar
Send message
Joined: 16 Jan 08
Posts: 98
Credit: 1,371,299
RAC: 0

Message 37012 - Posted: 6 Mar 2010, 23:46:14 UTC

Got a handful of shorter ones overnight.

For example: de_14_2s_const_3_2276048_1267832764

Only took about 40 to 50 minutes instead of the normal 4 - 4 1/2 hours on my CPU.
They pay good too. But they seem to be in the minority.

____________
Never surrender and never give up. In the darkest hour there is always hope.

schizo1988
Send message
Joined: 26 Nov 08
Posts: 6
Credit: 44,903,350
RAC: 0

Message 37019 - Posted: 7 Mar 2010, 9:31:34 UTC

I hope these de_14's stay around for a while as they do crunch way faster than anything else, and you can score major credits quickly, as it appears they run about 5 times faster that the usual jobs, and that is on everything, my 295's, 5870 's and i7 920's.

On Thursday, I came within 30,000 of having my first 1 million credit day, so close, and not likely to happen again as they are now sending fewer of these really fast crunching jobs. I am actually glad in a way that they are not all this fast, as just one day of nothing but these jobs would be ridiculous, and sort of spoil things by granting too many credits, as I would end up being able to earn more than 4 million credits/day. CPU's are no longer able to produce numbers that make my over-clocking them even seem worth anymore and even doing it to my GPU's is becoming too much trouble as they tend to run hot even at stock speeds and keeping everything not only cool but quiet, can take a lot of work and extra expense for just small increase.

While Milkyway is great for GPU's it is not that great for CPU's, does anyone know of a good CPU project.
____________

Chris S
Avatar
Send message
Joined: 20 Sep 08
Posts: 1380
Credit: 186,119,883
RAC: 14,020

Message 37023 - Posted: 7 Mar 2010, 13:07:26 UTC

I'm a very happy bunny at the moment :-)

I hope it keeps up

____________
Don't drink water, that's the stuff that rusts pipes

Profile speedimic
Avatar
Send message
Joined: 22 Feb 08
Posts: 260
Credit: 57,387,048
RAC: 0

Message 37028 - Posted: 7 Mar 2010, 17:48:18 UTC - in response to Message 37023.

I'm a very happy bunny at the moment :-)

I hope it keeps up


...the Kredit Kops will come to haunt us....

____________
mic.


STE\/E
Send message
Joined: 29 Aug 07
Posts: 486
Credit: 573,620,762
RAC: 0

Message 37048 - Posted: 8 Mar 2010, 12:30:43 UTC
Last modified: 8 Mar 2010, 12:35:09 UTC

The Project needs to do something about people Aborting the longer Wu's and running only the Shorter ones. It's already evident some People are running scripts to abort the Long ones and run only short WU's as their output has Doubled or even Tripled.

Prime Example of running only the Short Wu's > http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=155282&offset=0&show_names=0&state=3 >>> http://milkyway.cs.rpi.edu/milkyway/results.php?hostid=146264&offset=0&show_names=0&state=3 There are others too ...
____________
STE\/E

SkyeHunter
Send message
Joined: 6 Mar 09
Posts: 41
Credit: 38,856,291
RAC: 0

Message 37050 - Posted: 8 Mar 2010, 14:02:05 UTC - in response to Message 37048.

I am afraid that it is technically quiet difficult to rectify this abuse. Honnestly, in that case I won't mind deviding the amount of credit given by these WU's by 5 to 7.

Chris S
Avatar
Send message
Joined: 20 Sep 08
Posts: 1380
Credit: 186,119,883
RAC: 14,020

Message 37053 - Posted: 8 Mar 2010, 14:21:46 UTC

I take what I am given in workunits be they short or long. If some people are filtering out the longer ones then I think that is unfair to the rest of us that are playing straight. :-(


____________
Don't drink water, that's the stuff that rusts pipes

SkyeHunter
Send message
Joined: 6 Mar 09
Posts: 41
Credit: 38,856,291
RAC: 0

Message 37058 - Posted: 8 Mar 2010, 16:08:53 UTC - in response to Message 37050.

Possible solution : Take an abort counter per user, that is reset on a daily basis to the queue length. Every abortion decreases the counter, every result resturned increases the counter. Counter below 0 = No WU ...

If I remind it correctly GPUGrid uses a similar technique...

Is this workable / realistic ?

Other solution, every infriction reported and valid = 1 week ban.

Cons : admin overhead and snitching culture ...

STE\/E
Send message
Joined: 29 Aug 07
Posts: 486
Credit: 573,620,762
RAC: 0

Message 37059 - Posted: 8 Mar 2010, 16:26:50 UTC

Reset the offending Host's to 0 and you wouldn't have the Aborting Problem anymore, at least not on the scale it is now on some Box's. There's no valid excuse for a Computer running only the Shorter Wu's other than some Script running to Abort the Longer Wu's.

They would still try to get away with what they could but couldn't just completely Abort all the longer Wu's and only run the Shorter Wu's.


____________
STE\/E

Profile [AF>HFR>RR] Jim PROFIT
Send message
Joined: 19 Mar 08
Posts: 5
Credit: 232,730,744
RAC: 0

Message 37073 - Posted: 9 Mar 2010, 11:33:46 UTC - in response to Message 37059.

Completely agree with you PoorBoy.

1 · 2 · 3 · 4 · Next
Post to thread

Message boards : Number crunching : more credits :)


Main page · Your account · Message boards


Copyright © 2017 AstroInformatics Group