Posts by William


log in
1) Message boards : Number crunching : GBT ('guppi') .vlar tasks will be send to GPUs, what you think about this? (Message 1788949)
Posted 6 days ago by Profile WilliamProject Donor
What one who interested in project results, not in competition for imaginary toaster, will chose?


It depends on whether such person would consider the results as the only part of the experiment, or if they consider the pain of change as part of the science too.

IOW: probably making the choices en-masse that feel right as individuals, add together to guide where things will head next. At some point the imaginary toaster/credits will be important, but for now it's more about understanding who's affected most.

Nash's Equilibrium (game theory) would suggest the most helpful situation would be for everyone to run as horribly as possible, so as to expose all the weaknesses. For some 'horribly as possible' would mean aborting, for others rescheduling, for others crunching CPU only. Probably since there aren't known 'best options' yet, I'd say we're likely to end up with a good outcome.

Ever the optimist, eh?
OTOH, maybe it's that bad, it can only get better...
2) Message boards : Number crunching : OpenCL MB v8.12 issues thread (Message 1788413)
Posted 8 days ago by Profile WilliamProject Donor
It's not openCL workunits it's openCL app.
You can use the lunatics installer and specify exactly what apps you want to run.


Once the WU is downloaded, it is assigned an app. It can't be ran by any other app at that time. The assignment might happen before it is downloaded, I don't know for sure. I really doubt too many people were very confused by what I said. If anyone was, let me know and I'll send you my snail mail addy in PM so you can send me hate mail that way and keep it out of the forum.

Also, I did basically ask for any way to avoid the openCL, but I will only stay with stock apps from now on. I've had way too much trouble from 3rd party apps.

You can run the stock apps under anonymous platform and restrict the set to what you want. It's just a lot more work than installer, since you need to write your own app_info.xml

And you can reassign WUs after they've been downloaded but it requires stopping boinc, tampering with client_state.xml and restarting. One could even write a batch file and automate the process.

Which 3rd party apps did you have trouble with if I may ask? Seeing as Raistmer's and Jason's work eventually becomes stock...
3) Message boards : Number crunching : OpenCL MB v8.12 issues thread (Message 1788404)
Posted 8 days ago by Profile WilliamProject Donor
I just noticed that running legacy with setiathome_8.12_windows_intelx86__opencl_nvidia_sah.exe uses up 1.0 (100%) CPU per WU despite the app_config.xml settings of <app_config>.


And it is for this reason I wish to eliminate any opencl WUs from being downloaded. Is this even possible? If not, I reckon I'll finish what WUs I have and that'll be it for me and SETI@home.

It's not openCL workunits it's openCL app.
You can use the lunatics installer and specify exactly what apps you want to run.
4) Message boards : Cafe SETI : Raccoon Update XXIII - All Are Welcome In The Critter Cafe (Message 1788389)
Posted 8 days ago by Profile WilliamProject Donor
There seemed to be an awful racket outside our first floor window, which upon investigation turned out to be sparrows nesting in the wooden compartment between gutter, wall and roof.
5) Message boards : Number crunching : Average Credit Decreasing? (Message 1787236)
Posted 13 days ago by Profile WilliamProject Donor
how was that about a wise (wo)man accepting the things she/he cannot change?

*ducks*
6) Message boards : Number crunching : Average Credit Decreasing? (Message 1786958)
Posted 14 days ago by Profile WilliamProject Donor
Got a spreadsheet somewhere on Googledocs for analysing Albert data [Edit: may have been seti-beta CPU, will see], Will dig it out on the weekend.

Data from Eve, now sadly demised. Beta iirc, only place that keeps results almost indefinitely and is best if you need to compare stuff.

I'll rummage.

edit: on second thought, you may have a spreadsheet, I've got the original data... somewhere. dang.
edit 2: I did find the spreadsheet though. I can't find the original graph on the cloud. Must be somewhere though...
7) Message boards : Number crunching : Average Credit Decreasing? (Message 1786647)
Posted 15 days ago by Profile WilliamProject Donor
True. It *appears* and is árbitrary (actually chaotic), because of the faults.

Hold that thought. "Chaotic" is a good scientific (mathematical) word. Just don't call it random in David's hearing (he's touchy about that).

'Three is Chaos'

No it's not random, but the code amounts to fairly unpredictable credit awards (though as for any chaotic system, there are patterns [*]) and to the layman unpredictable is pretty much the same as random. We say chaotic, when we know you could calculate it.
'random' is a word we use when there don't seem to be any rules. Very few things are truly random. Blame physics for that ;)



[*] and if you were in possession of every single variable used in CN you could indeed calculate how much credit you'll be getting. CN IS a calculation after all.
8) Message boards : Number crunching : Munin setiathome has now also graphs for V8 (Message 1783422)
Posted 27 days ago by Profile WilliamProject Donor
That's because of the xml error Andy reported a few days ago.
9) Message boards : Number crunching : Task Preference Not Being Honored (Message 1783418)
Posted 27 days ago by Profile WilliamProject Donor
Nonetheless he should be getting v8 if no AP are available with the first setting.
10) Message boards : Number crunching : SETI V8 CPU Wierdness (Message 1783416)
Posted 27 days ago by Profile WilliamProject Donor
signal cor caroli i have the image

Might as well try to signal cori celesti... ;)
11) Message boards : Number crunching : Average Credit Decreasing? (Message 1783206)
Posted 27 days ago by Profile WilliamProject Donor
In easier words, it is supposed to take the time you took for a task, award you credit according to how efficient you were (compared to all tasks and tasks in your group of tasks) in crunching the task, [I think in theory, you are supposed to get a bonus if you are faster than the average] and then use the time (in relation to how 'big' the task was) to estimate your speed for future tasks (APR).

I'm not sure that was an easier explanation of what it's supposed to do...

In praxi, we have noticed that it falls sadly short of both making a good estimate of your speed (it lacks adaptability) and of awarding credit relative to your efficiency.

In short, the way in which it was implemented (coded) cannot work. It leads to chaotic behaviour.
12) Message boards : Number crunching : SETI/BOINC Milestones [ v2.0 ] - XXIX (Message 1783161)
Posted 28 days ago by Profile WilliamProject Donor
My BOINC milestone: 10 years as your BOINC moderator. Go on, say something nice about me. :-)

My signature image links there as well.

wow. congrats. I tried to read the text on your sig earlier ;)
13) Message boards : Number crunching : V8 CUDA for Linux? (Message 1783156)
Posted 28 days ago by Profile WilliamProject Donor
Greetings All

Is user Ralph Hyre (beta) Workunit ID# http://setiweb.ssl.berkeley.edu/beta/result.php?resultid=23347475
using a special app, as he is running a version of Linux and is getting what looks to be like OpenCL workunits.

I am a bit confused, or is it because he is a beta tester??????

Regards

SETI@home v8 v8.07 (opencl_nvidia_SoG) x86_64-pc-linux-gnu

whatever he is running meets the specs for that stock app. Or met them in march, he doesn't seem to be active any more.
14) Message boards : Number crunching : Help! No GPU work for Days (Message 1783141)
Posted 28 days ago by Profile WilliamProject Donor
also, once you are empty and have repeated 'no task' events, boinc will go into increased backoffs (up to 24h). [mea culpa. it's meant to stop boinc asking for tasks too often]
under normal conditions it can take a few attempts to top up - if you are on a 24h backoff...

each tie a task is reported, those backoffs are cleared, so it doesn't happen if you have tasks left. 'priming the pump' when you are empty can be a bit difficult. I suggest you periodically hit the update button until you get a few.

And yes, if there are few tasks to be had for NV because of lots VLAR you are competing for a very limited resource and must get lucky on your request.
15) Message boards : Number crunching : Panic Mode On (102) Server Problems? (Message 1783139)
Posted 28 days ago by Profile WilliamProject Donor
yes what happened is suddenly earlyer the whole project reset with out of pointer errors and asked me to attach a project it killed everything and that was on lunatics then i added seti again and it cleaned out the the project folder completely so i had to reinstall lunatics

sounds to me like you got yourself a new datadir. that would explain some of the stuff.

BOINC stores the whereabouts of its program and data dirs in a registry entry.
If you change the data dir away from the default, but for some reason lose the registry entry, reinstalling boinc will (unless you point it to the previously used dir [which incidentally is how you move data dirs across computers]) go back to the default. the default then of course is completely empty - no projects no lunatics no nothing.
you reattach to the project - you get stock. boinc will also realise that the call comes from a fresh install, it will look up your computers, find the old tasks, find you don't have them on board (cause they are in a different dir) and mark them as abandoned. it will then proceed to give you stock on the original host id wth the original APRs.
when you then reinstall lunatics, the registry is checked for boinc and stuff put in the right place, upon which you go back to anonymous, happily crunching whatever tasks you had.

the part of the code supposed to deal with errors and how many tasks you should be getting if you have lots of errors is about as capricious as the code that selects the fastest app.
sometime it works, sometimes it doesn't.

Life goes on. In the big picture it doesn't matter if one dot is the right colour. I forgot which % of dots needs to be worng for you to notice the picture is getting distorted.
16) Message boards : Number crunching : Lunatics Help (Message 1782895)
Posted 29 days ago by Profile WilliamProject Donor
Richard is out, so as one of the other persons involved in the installer:

Hi

I just found this useful thread so can someone answer some questions for me?

1) What benefits does this optimised app offer, compared to the standard SETI apps? (I used to use "crunch3r's" optimised app, ages ago, so I'm aware of what benefits that app gave - is this one any better ?)

we wouldn't call it optimised if it wasn't any better, now would we?

The picture isn't quite that simple though, so I'll break it down by device types.
CPU: stock has a 'lowest denominator' app and an AXV app. if your CPU falls in between, we definitely have a better version. I'm not sure about AVX - the optimised app might still be a little faster than stock.

GPU If you have NV then the installer _currently_ has the same app as stock. That's because for several years Jason as an independent volunteer developer has been providing the NV app to the project. Since stock tends to move a little more slowly, Lunatics installer can be ahead of stock for a while.
ATI same picture. Raistmer along with Urs provide the OpenCL app. Again, the installer may have a newer version (but not right now, we just pushed all we had at the v8 transition.)

installer offers the additional advantage to select exactly what apps to run (like AP on the CPU only) and to be independent of BOINC's erm capricious 'fastest app' selection.

2) I set this up last night with the installer (win_32 v0.44) and I know that I have a Fermi type GPU (GTX580)...but the installer "offered" the cuda32 GPU app...whereas recent tasks I've run (before using the optimised app) were running (mostly) cuda50 tasks on the GPU. What's the difference ?


There's guidance in the readme - the cuda32 flavour is preselected as it's the 'will work in any case' choice. For a Fermi type you indeed want cuda50.

The difference is in the cuda runtime libraries used. newer cards are faster with higher cuda versions. older cards are faster with older cuda versions.
Jason could explain in detail why...

So, do I over-ride the installer and select the cuda50 app ?

Thanks in advance :)

Don't need to override - just run it again and change your selection.
17) Message boards : Number crunching : Say goodbye to SETI@home v7. (Message 1781303)
Posted 22 Apr 2016 by Profile WilliamProject Donor
at that rate, you are going to run out of size and colour before you run out of tasks :D

you check once a day? care to make a graph? I'd like to make a quick extrapolation.
18) Message boards : Number crunching : Average Credit Decreasing? (Message 1781255)
Posted 22 Apr 2016 by Profile WilliamProject Donor
ok, so what we really need is somebody with a bucketful of cash, who says 'here have that - but only if you make sure credit works' ?
19) Message boards : Number crunching : Average Credit Decreasing? (Message 1781229)
Posted 22 Apr 2016 by Profile WilliamProject Donor
No, now BOINC has created the credit monster, I think it has a responsibility to calculate it with the same rigour as the projects calculate their science. And we have a responsibility to report the problems as fully and accurately as we can.

No objections, your honour.

But credit should have a sound mathematical basis and a faultless computational algorithm implementation.
And CreditNew hasn't. Not the latter and I'm still not convinced about the former.

Looking at credit in this detail is like looking at a house that is falling down, has gaps all over the planks and a leaking roof, when the architect used a twig for a ruler and the builders used cardboard for wood...
20) Message boards : Number crunching : Average Credit Decreasing? (Message 1781221)
Posted 22 Apr 2016 by Profile WilliamProject Donor
[Edit:] @Richard: noticed you have a 970 in with a 750ti in two machines ? both those are likely to claim on the high side for 750ti tasks... (peak_flops_from_wrong_device X elapsed)

I did declare the 970s in my original statement of evidence, m'lud...

But I hadn't picked up from the previous discussion of code that mis-matched GPUs would have that effect (perhaps because I never ran mis-matched GPUs until January). If that's the case, there's no way of applying rigorous mathematics to credit until the server has proper, individual, records of each separate GPU.

On another tack, I'me seeing a remarkably low proportion of shorties in the current Arecibo mix - that's a big plus for my 750s.

please don't mention maths and CreditNew in the same sentence...
Next time I need a RNG I'm just going to use credit.


Next 20

Copyright © 2016 University of California