Posts by William

log in
1) Message boards : Number crunching : Download errors? [ Not resolved! And at a loss... ] (Message 1732020)
Posted 1 day ago by Profile WilliamProject donor
If two machines are affected and they don't have an identical setup, then it rather points to a transmission problem than a problem with the computer.
2) Message boards : Number crunching : benchmark stock vs. optimized -- problem (Message 1731068)
Posted 4 days ago by Profile WilliamProject donor
The [Cuda60 experimental] test build on my site has been getting the best unsolicited feedback of all the Linux builds I personally made so far, so worth comparing. As Arkayn mentions work is underway, though very much at a snails pace with work pressures.

'my site' being

If the tortoise gets there, I don't see any reason for the snail not to get there.
3) Message boards : Number crunching : Download errors? [ Not resolved! And at a loss... ] (Message 1730377)
Posted 5 days ago by Profile WilliamProject donor
I'm no expert on MD5 but as has been pointed out very early on, if BOINC says 'MD5' error it means that the MD5 checksum of the file as received by BOINC did not match the MD5 checksum of this file as supplied by the project.

IOW the file you got is different from the file that was sent out.

So, something is interfering with the files you get. might be AV might be something else, might just be something writing something into the header of the file - you need to figure out what might be tampering with the files, or if anything else might be resulting in receiving corrupted files.
4) Message boards : Number crunching : Alternative Instead Of BOINC v7.6.9? (Message 1730312)
Posted 6 days ago by Profile WilliamProject donor
And as previously discussed ad nauseam, the OSs no longer work sequentially but on different threads and different priorities which to some extent are not under the control of the user/applications .
Actually I think what is needed is a firm handover tag IN THE FILE. Something like a 'complete' tag.

But that would require changes to both applications and client - the applications to write the tag, the client to read it, and client to deal with 'old app, doesn't write such a tag'. [I think here the app could put an 'I'm giong to tag when this file is complte' tag at the beginning'

And now we reached level of complexity that is simply not going to happen.

We can get stuff done to the client, but we can't tell the projects what to do.
5) Message boards : Cafe SETI : Raccoon Update XXII - All Are Welcome In The Critter Cafe (Message 1728491)
Posted 12 days ago by Profile WilliamProject donor
They are supposed to have a thriving colony here, but so far I've failed to spot any in the wild.
Maybe I'll be more lucky when we have our own garden.

I think it has been mentioned you can attract them with cat food?

I seriously hope they won't go after my bee hives though!
6) Message boards : Cafe SETI : Raccoon Update XXII - All Are Welcome In The Critter Cafe (Message 1727823)
Posted 14 days ago by Profile WilliamProject donor
I am pretty sure the average kid knows much nastier words than I do. And a lot more of them.
Must be sheltered upbringing...

[hello all ;)]
7) Message boards : Cafe SETI : Missing Persons Bureau (a.k.a. The Mother Hens Thread) (Message 1727821)
Posted 14 days ago by Profile WilliamProject donor
Woohoo, Soft^Spirit is back.


She appears to be missing again :(
8) Message boards : Number crunching : Problem with new update (Message 1726179)
Posted 20 days ago by Profile WilliamProject donor
Another update Monday!?


Okay. What the difference between 7.6.9 and 7.6.10?

(Edit) I am running a AMD Windows 7 desktop.

7.6.10 is a Mac-only update - and at the moment, I can't find what it was that was changed.

We are doomed!
9) Message boards : Number crunching : Lost all my points after update to BOINC's last version (Message 1726175)
Posted 20 days ago by Profile WilliamProject donor
Thousands separators were part of the latest client updates, yes.
Apparently David quite forgot that not everyone uses ',' for thousands and '.' for decimals. (German does it the other way round and the Indians have a completely different system)

There was some discussion on the boinc alpha mail list, but I don't know if proper localisation of that feature (i.e. increase readability of large numbers by adding separators) was introduced.
10) Message boards : Number crunching : Oldie Newbie question about work done, credits, missing units, stats etc. (Message 1723815)
Posted 27 days ago by Profile WilliamProject donor
That's a lot of questions in one go :D
I'll try answer a few, starting with the basics and we'll gradually fill you in.
I'm sure you'll end up with even more questions :)

Ok, so first basic BOINC functionality:

According to your cache settings, BOINC will request work, crunch it and return it.
The returned work then needs to wait for the wingmate to validate. If both results are in and agree, the workunit (WU) is validated and the results (signals found) are added to the science database (we don't get to see that one).
24 h after validation the validated tasks are purged (deleted) from the database, i.e. they disappear from your task list.
Older tasks remaining are either still waiting for the second result 'waiting for validation' or the other result disagreed and they become 'inconclusive'. Inconclusives get send out again until a consensus is reached or the unit is deemed faulty. Inconclusives usually turn into valids. You may see the occasional error. Unless you get a lot of errors or invalids it's nothing to worry about. If in doubt about something, ask.

BOINC has a set of policies how it uses your computer. These can be set over the web interface or locally. NB local settings override web ones. if you are trying to change something over the net and it doesn't appear to work, check if you have local preferences set.

You may want to uncheck 'suspend when in use' and 'suspend GPU when in use' - else BOINC will only run if you are not using the computer at all. Unless you have really heavy usage of you machine you won't notice BOINC running in the background. IF you have GPU intense applications, special exceptions can be made locally (e.g. boinc gets suspended only when you play game XY).

'Suspend when non-boinc CPU usage is above x%' can be used - but 'use at most X % of CPU time' is best left alone - if you set that to e.g. 50% boinc will run one second, stop one second, run one second... should you need to throttle boinc (e.g. for temperature control on a laptop) there are far better 3rd party programs around.
'use at most x % of CPUs' can be used to 'leave a core free' (or several) some of the applications benefit if you don't devote everything to boinc and some CPU is left free to 'feed the GPU(s)'.

on the seti preference page you can adjust which applications you would like to run and whether or not you want to use the CPU or the GPU(s) at all. some people use only their GPUs on SETI and crunch other projects on the CPU.

Lunatics offer optimised applications. If you have Windows, this is done via a handy installer. (everythig is available seprately too, should ypu want to update an app before a new installer is out).

ATM the GPU apps offered by Lunatics are the same as those you get directly via SETI (commonly referred to as 'stock'). That's because all current GPU apps are developed by volunteers and then implemented into stock. That may take a while, so with stock you tend to lag behind the newest developments. If you want easy, stay on stock.

CPU apps are a different matter. There you will most likely get something more efficient if you use Lunatics. The downside is that you have to keep an eye on the boards, because you won't automatically get new applications, but have to install those yourself.

I hope that answered a few questions :)
11) Message boards : Number crunching : Panic Mode On (100) Server Problems? (Message 1723811)
Posted 27 days ago by Profile WilliamProject donor
lol. or we are down in the history books as the group that found ET :D
12) Message boards : Number crunching : Why even do this? (Message 1723137)
Posted 29 days ago by Profile WilliamProject donor
He has a valid question. How much he contributes should not make any difference to anything.

And I regard 'red hot' as symbolic for 'using power without any reason'

So, to me, the OP's question reads along the lines of 'The scientists are sitting on a heap of data. Are they going to analyse that data ay some point?'

Because why generate even more data, if nobody is ever going to make any use of it? That would be just pouring energy down the drain. Which, looking at the greater picture would be a very silly thing to do. If you contribute to climate change, you should at least have a good reason for doing so.

So, as a scientist I would say, yes that data is going to be analysed. No good scientist would produce data for data's sake. _when_ that data is going to be analysed is an entirely diffenret question. it may be years or decades from now. but still some day somebody is going to sift through that heap and maybe some unit crunched last year, last dacade or in '99 has something in there that points to something.

yes, looking for ET is something of a question of belief. but the OP would not be here if he had not, at some point, shared that belief.

However, asking whether the data you are generating (however much or little of it) is going to be looked at sooner or later remains a valid question.

IMO the answer is yes.

Please stop accusing people of trolling that ask valid questions.
13) Message boards : Number crunching : Can someone think of an explanation for (Message 1722023)
Posted 4 Sep 2015 by Profile WilliamProject donor
'Yes, we can!'

'The team' got onto it and we think we are looking at a massive haunting of ghosts caused by a local client loop. [Which cleared around the 13th of August]

We think what happened is the following:

The client made a request. For reasons unknown (slow line? corporate firewall? AV? ) the _large_ reply with the allocated tasks was not received by the client (i.e. the client had stopped listening for the reply by the time the reply arrived). The server however thinks those tasks were allocated. With 'resend_lost_tasks' off this causes what we call ghosts.

Now, on the client 'no reply' triggers a small 'RPC backoff' of about 1 minute.
After that minute the client repeats the request. The server says 'OI, you just asked 1 minute ago, you have to wait 5:03 minutes before you can ask again!'
THAT causes the client to patiently wait 5:03 minutes before it makes the next request. Which gets allocated tasks. Where the reply isn't received in time.
Which makes the client ask again after about a minute. Server says 'go away, come back in 5'

So, the client has effectively tied himself up in a loop, because the 'small and fast' 'go away' replies arrive and the 'large and slow' replies with the tasks don't.

The exponential backoff you get for 'didn;t get tasks' is not triggered in this scenario.

As it seems to have cleared itself up we are reluctant to have anything done about it. The cure might be worse than the (very localised) problem. been there done that, ate the hat.
14) Message boards : Number crunching : download to flash drive (Message 1721601)
Posted 3 Sep 2015 by Profile WilliamProject donor
Remember, use small words. :)

But not this small please
Tailoring your level to your audience tends to be a challenge.
If you don't really know the level of the audience you almost always get it wrong.
And then you can stubbornly start too high each time even after 5 years of comms. ;)
15) Message boards : Number crunching : download to flash drive (Message 1720878)
Posted 1 Sep 2015 by Profile WilliamProject donor
Did the original problem get sorted?
16) Message boards : Number crunching : Panic Mode On (98) Server Problems? (Message 1703690)
Posted 21 Jul 2015 by Profile WilliamProject donor
Credits? what credits?

You not only screw up credits, you also mess with a few server side held personal and global variables that are in turn used for runtime estimates.
Messed up credit is just a secondary effect.

These days a few people moving work across devices will not influence the bigger picture much - the impact is on your own APR and on credit (for you and your wingmate) of course.
The latter makes it an ethical decision ;)

I would discourage from rescheduling in general - you don't want to rock the boat more than you absolutely have to.

I live by the principle 'there are no rules, only guidelines'
That makes you responsible for your actions.
If you can stand before the angry mob and say 'yes I did it, I had my reasons' - well. But don't be amazed if you get lynched before you managed to explain your reasons.
17) Message boards : Number crunching : Suddenly BOINC Decides to Abandon 71 APs...WTH? (Message 1699813)
Posted 9 Jul 2015 by Profile WilliamProject donor
So, after a nice email we've had the fix applied.

I had a quick test and it doesn't look like it has been deployed yet though.
18) Message boards : Number crunching : Bad News on BOINC funding (Message 1698936)
Posted 6 Jul 2015 by Profile WilliamProject donor
I'll let Richard explain why he thinks branching is bad.
19) Message boards : Number crunching : Bad News on BOINC funding (Message 1698934)
Posted 6 Jul 2015 by Profile WilliamProject donor
Excuse me for being ascerbic, but this is a disaster and some of us are trying to limit the damage.
20) Message boards : Number crunching : Bad News on BOINC funding (Message 1698933)
Posted 6 Jul 2015 by Profile WilliamProject donor
What all this panic about?

AFAIK SETI@home functions w/o US government funding 3, 4, 5+ years already. So what? The single sign of past funding is the line on the bottom of SETI beta site "AstroPulse is funded in part by the NSF through grant AST-0307956" that should not be there quite long time ago. That funded part was changed even in algorithm heavely enough, not speaking of about all GPU versions that did not fund by NSF ever. Rather RFBR granting mention would be more appropriate :P ;D ;D ;D

Yes dear, this is not about SETI funding, it is about BOINC funding.

BOINC being the platform SETI is running on, as you might be aware, since you occassionally post to boinc mailing lists.

If you don't want to worry about the bigger picture, fair enough. Just don't complain if the gallery closes you wanted to hang your picture in.

Next 20

Copyright © 2015 University of California