Message boards :
Number crunching :
Rescheduling Hosts - Bad Practice
Message board moderation
Previous · 1 . . . 6 · 7 · 8 · 9 · 10 · 11 · Next
Author | Message |
---|---|
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
When the awarded credits for AP drop to the point where it is not an advantage to crunch them VS MB work, this discussion shall all be a moot point. I crunch 24/7, and as long as the servers shall send me something to work on, the kitties shall be content to know that they are participating in what I consider to be one of the most significant attempts by mankind to find out just where we stand in the makings of the universe. Succeed or not, at least I can say that I did, indeed, make the attempt. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
James Sotherden Send message Joined: 16 May 99 Posts: 10436 Credit: 110,373,059 RAC: 54 |
I can see both sides of this. I still have no idea what my two I7's are capable of. When we have AP's the RAC climbs. When no AP it goes south fast. Before ver7 came out I never saw swings of what we see now. I see see swings of close tp 5,000 RAC on my machines. And I just have below middle of the range machines. I can see why its called credit screw. The name fits 100%. That being said. I agree with Mark I will run what I get untill this project closes the door. [/quote] Old James |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
I can see both sides of this. I still have no idea what my two I7's are capable of. When we have AP's the RAC climbs. When no AP it goes south fast. Before ver7 came out I never saw swings of what we see now. I see see swings of close tp 5,000 RAC on my machines. And I just have below middle of the range machines. Before we saw v7 of MB I would see swings of 10-15K of my total RAC. I still see the same swing. I wouldn't say anything has changed there. The only difference is that the numbers are a bit lower now. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
SongBird Send message Joined: 23 Oct 01 Posts: 104 Credit: 164,826,157 RAC: 297 |
Is this machine rescheduling? |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Is this machine rescheduling? I see no sign of it. Just running optimised applications for CPU and Intel GPU, each on the advertised resource. No harm in that. |
SongBird Send message Joined: 23 Oct 01 Posts: 104 Credit: 164,826,157 RAC: 297 |
It has 109 tasks In progress. I thought the whole point is that there is a maximum of a 100. Anyway. This is actually my machine. I was doing some stuff I do not fully understand in order to use a beta Intel_GPU app and was just wondering if I've been in fact rescheduling... As you might have picked up I'm not much of a specialist in all this :D Thanks! |
Wiggo Send message Joined: 24 Jan 00 Posts: 34772 Credit: 261,360,520 RAC: 489 |
It has 109 tasks In progress. I thought the whole point is that there is a maximum of a 100. Actually you are allowed 100 for CPU and 100 for GPU. Cheers. |
ausymark Send message Joined: 9 Aug 99 Posts: 95 Credit: 10,175,128 RAC: 0 |
To the developers, just a thought on how to handle credit calculation/allocation that should solve everyones issues - assuming its possible. Here is my thought. When Boinc starts it knows the processing Power of the CPU and GPU's of its computer. There does not need to be any estimates of how long a work unit will take, what we do need is something like an Odometer within the seti apps(s) that calculates and tallies how much processing power is being used at that point on that work unit (calculated every second for example), then tallies up the total processing power used to fully process that work unit. (Hence my Odometer analogy - or even 'total trip fuel used' meter - whatever analogy works best for you) Also by doing it this way fixes any issues we have of diminishing credits due to more efficient ways of processing the work units in software. So for example, lets say we have a work unit that takes 3 seconds to complete and the processor can process at 600 MIPS per core, and we are using one core (for the sake of simplicity in the example). Second SETI CPU Loading MIPS Total Mips 1 50% 300 300 2 100% 600 900 3 60 360 1260 Total Work Unit Processing: 1260 MI (Million Instructions) Allocate Credits for 1260 MI (Whatever that conversion rate may be) So thats my idea, it could be used for all BOINC projects. Thoughts? Cheers Mark |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
That is, in fact, exactly how the previous credit system used to work. If you look at one of my tasks at random, like task 3358432916, you'll see a line in the middle of the output: Flopcounter: 16102518492340.953000 That's exactly your odometer (though I'm not sure what the significance of one-thousandth of a floating point operation is). But if you look at your own recent AP results, you won't see that line (I think it might have been there once, but was taken out because there was already more than enough clutter). And thereby hangs the problem. The scheme only works if every programmer, at every project, adds in the odometer mechanism. I'm told it isn't as easy as it sounds. You don't want to actually 'count' flops, because that would take almost as much computing power as running the job - which would be horribly wasteful. Instead, you have to go through all the significant chunks of computer code, and assign each one a realistic 'count' you can add up to make the overall total. The present scheme was designed to save that extra programming work. It would be a nice idea if it worked, but unfortunately it doesn't, quite - yet. |
ausymark Send message Joined: 9 Aug 99 Posts: 95 Credit: 10,175,128 RAC: 0 |
Hi Richard Well thats the thing, we wouldnt need to calculate the flops at any instant as they are calculated when boinc first starts and runs the benchmarks on the PC. Then all one has to do is work out the % load that the seti program is having on the core(s), which Im guessing can be done fairly easily, and if not use the total core load, less precise but possibly still more accurate than what is currently being used. That load on the CPU to calculate the 'odometer'should be very minimal in the scheme of things these days. (Linux and Windows system monitors do that with negligible load.) And to reduce manipulation of the system have BOINC run the benchmark program randomly again every 24 to 40 hours. Perhaps im over simplifying it and dont know the complexities involved, but i guess the question is - would it be a more accurate system than what we currently have assuming it doesnt adversely affect affect crunching ability by more than half of one percent? Cheers Mark |
Wiggo Send message Joined: 24 Jan 00 Posts: 34772 Credit: 261,360,520 RAC: 489 |
|
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Now you're asking for something slightly different. Have a look at http://boinc.berkeley.edu/trac/wiki/CreditNew Your first suggestion - the odometer analogy - is what is described as "The second credit system" - equal pay for equal work done. Your second suggestion - relying on the benchmark - goes back to "The first credit system". The problem we found with that was that it makes no allowance for the efficiency of the programming. If you install an optimised application, the benchmark remains the same, and the processing time goes down, so you are awarded less credit per task. That breaks the rule of "equal pay for equal work". The benchmark is a speedometer, not an odometer - and it's not a very good one. |
ausymark Send message Joined: 9 Aug 99 Posts: 95 Credit: 10,175,128 RAC: 0 |
Hi Richard I glanced over the link and see that its nowhere near as simple as I had assumed... I will have a good look at it tomorrow to see if I get any flashes of brilliance, or just zombie eyes lol Cheers :) Mark |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
And another host for Wiggo's list, one with 1250+ tasks and most of them returning as error: https://setiathome.berkeley.edu/results.php?hostid=7090798 |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
And another host for Wiggo's list, one with 1250+ tasks and most of them returning as error: https://setiathome.berkeley.edu/results.php?hostid=7090798 That host has In progress (33) & does not have any aborts. Just a lot of errors. While 20% of their tasks are errors it is better than I have seen on other machines. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
juan BFP Send message Joined: 16 Mar 07 Posts: 9786 Credit: 572,710,851 RAC: 3,799 |
We all are out of AP WU to DL then i see this hosts: http://setiathome.berkeley.edu/results.php?hostid=6637324 with 846 in progress AP WU http://setiathome.berkeley.edu/results.php?hostid=6016862 with 1180. http://setiathome.berkeley.edu/results.php?hostid=6794209 with 696. Hard to belive that could be right, even with all hosts producing few errors. All are anonymous. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Then one compute number of tasks that gone to plain stock CPU hosts that just waste of electricity and (in case of Linux) just trashing tasks... one would thank these 3 for saving at least some tasks ;) SETI apps news We're not gonna fight them. We're gonna transcend them. |
Batter Up Send message Joined: 5 May 99 Posts: 1946 Credit: 24,860,347 RAC: 0 |
It is obvious that the powers that be condone this type of behavior so it must still get the job done for them. I found the script that reassigns CPU to GPU but cobblestones are worth less than crypto-currency so why bother. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
The limits were imposed due to some of the faster hosts having a "State: All" number of around 10,000 or more. Those sort of numbers really affect the database. The previously mentioned hosts are just above 1,000, well below some of the other hosts that are 'following the limits'. Until someone approaches around the 5k mark, there really isn't an issue. Unless there is some actual need to hoard that many tasks, it's just a waste of time & energy on their part. |
Batter Up Send message Joined: 5 May 99 Posts: 1946 Credit: 24,860,347 RAC: 0 |
Unless there is some actual need to hoard that many tasks, it's just a waste of time & energy on their part. If cobblestones are the goal reassigning and hoarding AP WU so one never runs out is as good as running another GPU. |
©2024 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.