Welcome to MilkyWay@home

Nbody WU Flush

Message boards : News : Nbody WU Flush
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 11 · 12 · 13 · 14 · 15 · 16 · 17 . . . 22 · Next

AuthorMessage
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,946,411
RAC: 22,500
Message 73421 - Posted: 9 May 2022, 10:49:29 UTC - in response to Message 73420.  

Prime Grid does it that way, any task gotten by you before the Challenge starts will not be counted towards your challenge credits, also any task received after the Challenge ends does not count either, that prevents people from cranking up their cache and returning units over the next 10 days to 4 months after the Challenge ends.


This is news to me. Don't Primegrid just declare my credits as I return the tasks? It's the pentathlon that decides what to include.


YES in the Pentathlon, but not in the Prime Grid run ones as it's different, and they have more than half a dozen per year each lasting from a few days to a few weeks.
ID: 73421 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Mr P Hucker
Avatar

Send message
Joined: 5 Jul 11
Posts: 990
Credit: 376,142,956
RAC: 2
Message 73423 - Posted: 9 May 2022, 10:52:29 UTC - in response to Message 73415.  

Yep. I'm in violent agreement to rethink the limit, which was probably legit at onetime. Boinc is 20+ years old, and there are a lot more and a lot faster home computers now. And even 10 year old GPUs are screaming fast. Shucks, my R9 280x with a busted fan does 1 separation every 30 seconds. Or 2880 tasks in 24 hours. Sadly Boinc does not support that burn rate here. As I write this, I am waiting for GPU separation tasks.

The capacity in the MW community is definitely there, even more so when Unixchick and her WCG friends drop by for a month or so. It is high time to rethink the server side.
How do you get a 280X to do that? I only get one every 38 seconds, running a handful at once.

What makes you think Boinc is the limit? It's the settings Tom has chosen that limit how much we can buffer.
ID: 73423 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Mr P Hucker
Avatar

Send message
Joined: 5 Jul 11
Posts: 990
Credit: 376,142,956
RAC: 2
Message 73424 - Posted: 9 May 2022, 10:54:03 UTC - in response to Message 73416.  

What's stopping you getting 100 at a time and maxing out your CPU with no settings changes?

100 at a time does not max out my settings, which is set to 3 days worth of work across all task types. The issue is, my computer runs dry every few hours or so, and I would like to keep it busy, but MW does not seem to support that. I think my computer limit is something like 550 tasks all together.
But Nbody doesn't have the "can't send and receive tasks at once" bug like seperation does, so surely a connection with the server up to every 1.5 minutes is enough to keep your CPU busy?
ID: 73424 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Mr P Hucker
Avatar

Send message
Joined: 5 Jul 11
Posts: 990
Credit: 376,142,956
RAC: 2
Message 73425 - Posted: 9 May 2022, 10:57:36 UTC - in response to Message 73417.  
Last modified: 9 May 2022, 10:58:14 UTC

That's a slow GPU at 44Gflops, my Tahitis at 1000Gflops only get 2 or 3 hours work, then an annoying 10 minute delay because of Boinc thinking the server has no work available, when in reality it was just refusing to hand it out on the same contact as receiving the last ones.

Yep. that's why I recently got an R9 280x, at your recommendation. Pleased as punch with it, but like you, can't seem to keep it fully employed, at MW anyway. Mikey(and I don't ever poke him!) has suggested 4 other space based DC projects but would like to stick here and make it work. I see really good things coming out of this research.
:-) Glad to see someone listened to me! (I've got 6 of them now!) Plus a Fury and a Baffin which aren't DP-based.

When the server isn't messing around, I get 2 to 3 hour's work followed by a ten minute delay. Not the end of the world. There are ways around it but they involve making your own Boinc Client. I believe Joseph Stateson does it. Linux only though, nobody's reprogrammed it for Windows. You can also schedule your computer to nudge Boinc to update with MW every x minutes using Windows task scheduler, but that means it's doing it unnecessarily during the 2-3 hours you don't need to, so wearing out the server.

And I too like this research.
ID: 73425 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Mr P Hucker
Avatar

Send message
Joined: 5 Jul 11
Posts: 990
Credit: 376,142,956
RAC: 2
Message 73426 - Posted: 9 May 2022, 10:58:45 UTC - in response to Message 73421.  

Prime Grid does it that way, any task gotten by you before the Challenge starts will not be counted towards your challenge credits, also any task received after the Challenge ends does not count either, that prevents people from cranking up their cache and returning units over the next 10 days to 4 months after the Challenge ends.


This is news to me. Don't Primegrid just declare my credits as I return the tasks? It's the pentathlon that decides what to include.


YES in the Pentathlon, but not in the Prime Grid run ones as it's different, and they have more than half a dozen per year each lasting from a few days to a few weeks.
Ah ok, I never do those. For badges I assume?
ID: 73426 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HRFMguy

Send message
Joined: 12 Nov 21
Posts: 236
Credit: 575,029,129
RAC: 37,181
Message 73427 - Posted: 9 May 2022, 14:32:07 UTC - in response to Message 73423.  

How do you get a 280X to do that? I only get one every 38 seconds, running a handful at once.

What makes you think Boinc is the limit? It's the settings Tom has chosen that limit how much we can buffer.

I'm running a Sapphire TOXIC R9 280X OC, which is one of the faster versions of the 280x series. Plus this in the app config file:
<app_version>
<app_name>milkyway</app_name>
<plan_class>opencl_ati_101</plan_class>
<avg_ncpus>0.866</avg_ncpus>
<ngpus>0.333</ngpus>
</app_version>

so 3 at a whack.
I guess I don't know why I think boinc is the limiting factor. If Tom has in fact set the limit, then why not open it up? There are volunteers here who are ready, willing and able to run more tasks, so why slow them down?
ID: 73427 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HRFMguy

Send message
Joined: 12 Nov 21
Posts: 236
Credit: 575,029,129
RAC: 37,181
Message 73428 - Posted: 9 May 2022, 14:39:02 UTC - in response to Message 73424.  

But Nbody doesn't have the "can't send and receive tasks at once" bug like seperation does, so surely a connection with the server up to every 1.5 minutes is enough to keep your CPU busy?

You wold think, but that is not the case.
ID: 73428 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
alanb1951

Send message
Joined: 16 Mar 10
Posts: 208
Credit: 105,445,808
RAC: 36,628
Message 73429 - Posted: 9 May 2022, 15:18:54 UTC - in response to Message 73428.  

But Nbody doesn't have the "can't send and receive tasks at once" bug like seperation does, so surely a connection with the server up to every 1.5 minutes is enough to keep your CPU busy?

You wold think, but that is not the case.

I suspect it depends on whether N-Body is your only selection (or, perhaps, whether you select for GPU work or not...)

I was doing GPU only until recently when I switched to CPU, N-Body only, and I don't see the delay any longer. I suspect that if I tried for CPU and GPU at the same time I'd see the bug re-surface, but I don't know whether it would re-surface if I stuck with CPU only but allowed Separation tasks, and as my reason for "N-Body only" is to offer admittedly limited help to clearing out the backlog I have no intention of testing that!

Cheers - Al.

P.S. At present I don't end up getting work every time my systems ask, but there's no explicit correlation between returning work and failure to get new work... Sometimes failure to get work (if you don't have 100 tasks already) might just be down to more people asking than there's work available, or an unbalanced mix in the feeder buffer! It seems to be shipping out over 32,000 Separation tasks an hour, so requests are plentiful :-)
ID: 73429 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Mr P Hucker
Avatar

Send message
Joined: 5 Jul 11
Posts: 990
Credit: 376,142,956
RAC: 2
Message 73430 - Posted: 9 May 2022, 16:34:16 UTC - in response to Message 73427.  

I guess I don't know why I think boinc is the limiting factor. If Tom has in fact set the limit, then why not open it up? There are volunteers here who are ready, willing and able to run more tasks, so why slow them down?
He recently told me he'd increased the limit, but it didn't take effect. He said it might when he rebooted but it didn't. Some kind of problem with the total in the buffer too, as people would start taking a substantial amount of the buffer at once. And there was mention of recompiling something? As with everything in Boinc, you change one thing, everything else breaks.
ID: 73430 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Mr P Hucker
Avatar

Send message
Joined: 5 Jul 11
Posts: 990
Credit: 376,142,956
RAC: 2
Message 73431 - Posted: 9 May 2022, 16:37:22 UTC - in response to Message 73429.  

I suspect that if I tried for CPU and GPU at the same time I'd see the bug re-surface, but I don't know whether it would re-surface if I stuck with CPU only but allowed Separation tasks, and as my reason for "N-Body only" is to offer admittedly limited help to clearing out the backlog I have no intention of testing that!
I've often had CPU and GPU at once. I had no problem getting loads of CPU work, even at the same time as getting or reporting GPU work, or reporting CPU work. The only problem I had was getting loads of seperation on the CPU when I wanted Nbody. There's no combination of settings to stop that.
ID: 73431 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HRFMguy

Send message
Joined: 12 Nov 21
Posts: 236
Credit: 575,029,129
RAC: 37,181
Message 73435 - Posted: 9 May 2022, 23:39:57 UTC - in response to Message 73431.  

The only problem I had was getting loads of seperation on the CPU when I wanted Nbody. There's no combination of settings to stop that.

I see that here and it annoys me to no end. That is a hindrance to clearing out the n body back log. Perhaps I can run an n body only client on the same PC?

btw, the back log seems to be growing a little bit right now.....
ID: 73435 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,946,411
RAC: 22,500
Message 73438 - Posted: 10 May 2022, 10:40:07 UTC - in response to Message 73426.  

Prime Grid does it that way, any task gotten by you before the Challenge starts will not be counted towards your challenge credits, also any task received after the Challenge ends does not count either, that prevents people from cranking up their cache and returning units over the next 10 days to 4 months after the Challenge ends.


This is news to me. Don't Primegrid just declare my credits as I return the tasks? It's the pentathlon that decides what to include.


YES in the Pentathlon, but not in the Prime Grid run ones as it's different, and they have more than half a dozen per year each lasting from a few days to a few weeks.


Ah ok, I never do those. For badges I assume?


Yes
ID: 73438 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,946,411
RAC: 22,500
Message 73439 - Posted: 10 May 2022, 10:45:58 UTC - in response to Message 73427.  

How do you get a 280X to do that? I only get one every 38 seconds, running a handful at once.

What makes you think Boinc is the limit? It's the settings Tom has chosen that limit how much we can buffer.

I'm running a Sapphire TOXIC R9 280X OC, which is one of the faster versions of the 280x series. Plus this in the app config file:
<app_version>
<app_name>milkyway</app_name>
<plan_class>opencl_ati_101</plan_class>
<avg_ncpus>0.866</avg_ncpus>
<ngpus>0.333</ngpus>
</app_version>

so 3 at a whack.
I guess I don't know why I think boinc is the limiting factor. If Tom has in fact set the limit, then why not open it up? There are volunteers here who are ready, willing and able to run more tasks, so why slow them down?


The idea it happened in the first place was because someone came in with half a dozen state of the art gpu's and hogged all the units so none of the rest of us got any tasks at all, if Tom didn't keep some kind of limit then someone could do that again but this time with the 3080Ti, 3090 series or some $5k gpu that can put all of those to shame. At that point your 280x would be sitting idle because they just can't make tasks fast enough for everyone. The idea of Boinc is to let everyone help, not just one or two people that end up doing all the work, that also helps the Scientists because a diverse set of crunchers proves that the theory and execution of the tasks is valid.
ID: 73439 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Mr P Hucker
Avatar

Send message
Joined: 5 Jul 11
Posts: 990
Credit: 376,142,956
RAC: 2
Message 73441 - Posted: 10 May 2022, 12:29:17 UTC - in response to Message 73435.  

The only problem I had was getting loads of seperation on the CPU when I wanted Nbody. There's no combination of settings to stop that.
I see that here and it annoys me to no end. That is a hindrance to clearing out the n body back log. Perhaps I can run an n body only client on the same PC?

btw, the back log seems to be growing a little bit right now.....
You can run two clients on the same PC, I just can't be bothered. It's complicated enough keeping 9 computers running. I'm constantly tweaking, moving hardware around, replacing worn out power connectors, etc.

What idiot invented the GPU power connector? They just don't last!! Way too much current in a tiny flimsy thing. If anyone knows where I can buy a quality one that doesn't tarnish in a few month's use and drop enough volts to make it crash, please let me know!
ID: 73441 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Mr P Hucker
Avatar

Send message
Joined: 5 Jul 11
Posts: 990
Credit: 376,142,956
RAC: 2
Message 73442 - Posted: 10 May 2022, 12:30:24 UTC - in response to Message 73438.  

YES in the Pentathlon, but not in the Prime Grid run ones as it's different, and they have more than half a dozen per year each lasting from a few days to a few weeks.
Ah ok, I never do those. For badges I assume?
Yes
Have you printed off all your badges and framed them? You could make some kinda certificate looking thing.
ID: 73442 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Mr P Hucker
Avatar

Send message
Joined: 5 Jul 11
Posts: 990
Credit: 376,142,956
RAC: 2
Message 73443 - Posted: 10 May 2022, 12:31:33 UTC - in response to Message 73439.  

The idea it happened in the first place was because someone came in with half a dozen state of the art gpu's and hogged all the units so none of the rest of us got any tasks at all, if Tom didn't keep some kind of limit then someone could do that again but this time with the 3080Ti, 3090 series or some $5k gpu that can put all of those to shame. At that point your 280x would be sitting idle because they just can't make tasks fast enough for everyone. The idea of Boinc is to let everyone help, not just one or two people that end up doing all the work, that also helps the Scientists because a diverse set of crunchers proves that the theory and execution of the tasks is valid.
Fair enough but I'm sure they could double it. Having only 2.5 hours work for a reasonable GPU is silly.
ID: 73443 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,946,411
RAC: 22,500
Message 73444 - Posted: 10 May 2022, 14:13:19 UTC - in response to Message 73442.  

YES in the Pentathlon, but not in the Prime Grid run ones as it's different, and they have more than half a dozen per year each lasting from a few days to a few weeks.
Ah ok, I never do those. For badges I assume?
Yes


Have you printed off all your badges and framed them? You could make some kinda certificate looking thing.


No I just let them show every once in awhile when I post...like now :-))
ID: 73444 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,946,411
RAC: 22,500
Message 73445 - Posted: 10 May 2022, 14:14:14 UTC - in response to Message 73443.  

The idea it happened in the first place was because someone came in with half a dozen state of the art gpu's and hogged all the units so none of the rest of us got any tasks at all, if Tom didn't keep some kind of limit then someone could do that again but this time with the 3080Ti, 3090 series or some $5k gpu that can put all of those to shame. At that point your 280x would be sitting idle because they just can't make tasks fast enough for everyone. The idea of Boinc is to let everyone help, not just one or two people that end up doing all the work, that also helps the Scientists because a diverse set of crunchers proves that the theory and execution of the tasks is valid.


Fair enough but I'm sure they could double it. Having only 2.5 hours work for a reasonable GPU is silly.


I don't know of anyone that disagrees with that
ID: 73445 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Mr P Hucker
Avatar

Send message
Joined: 5 Jul 11
Posts: 990
Credit: 376,142,956
RAC: 2
Message 73446 - Posted: 10 May 2022, 14:33:31 UTC - in response to Message 73444.  

YES in the Pentathlon, but not in the Prime Grid run ones as it's different, and they have more than half a dozen per year each lasting from a few days to a few weeks.
Ah ok, I never do those. For badges I assume?
Yes


Have you printed off all your badges and framed them? You could make some kinda certificate looking thing.


No I just let them show every once in awhile when I post...like now :-))
That image is ambiguous, it doesn't state if that's just you or if it includes your ears.
ID: 73446 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile HRFMguy

Send message
Joined: 12 Nov 21
Posts: 236
Credit: 575,029,129
RAC: 37,181
Message 73447 - Posted: 10 May 2022, 14:45:32 UTC

5/10/2022 8:16:07 AM | Milkyway@Home | update requested by user
5/10/2022 8:16:08 AM | Milkyway@Home | Sending scheduler request: Requested by user.
5/10/2022 8:16:08 AM | Milkyway@Home | Not requesting tasks: don't need ()
5/10/2022 8:16:10 AM | Milkyway@Home | Scheduler request completed


How do I clear this don't need () so I can get more work?
ID: 73447 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Previous · 1 . . . 11 · 12 · 13 · 14 · 15 · 16 · 17 . . . 22 · Next

Message boards : News : Nbody WU Flush

©2024 Astroinformatics Group