Posts by Joel

1) Questions and Answers : GPU applications : Lots of inconclusive tasks (Message 2038306)
Posted 16 Mar 2020 by Profile Joel
Post:
Thanks, I'll keep an eye on it
2) Questions and Answers : GPU applications : Lots of inconclusive tasks (Message 2038190)
Posted 16 Mar 2020 by Profile Joel
Post:
Before the project shutdown, I thought I'd contribute a little more using a new computer I haven't used before. It seems to be producing a lot of inconclusive tasks, mostly on the GPUs (one integrated in the CPU, one discrete). They don't seem to all have wingmen in common, so I'm wondering if the problem is on my end. Is this likely to be from overheating?

https://setiathome.berkeley.edu/results.php?hostid=8913588&offset=0&show_names=0&state=3&appid=
3) Questions and Answers : Macintosh : Haswell GPU (Message 1760986)
Posted 31 Jan 2016 by Profile Joel
Post:
I don't think there's any way to get updated Intel graphics drivers on OS X other than what Apple provides, sadly. Is that driver version known to cause problems on Windows?
4) Questions and Answers : Macintosh : Haswell GPU (Message 1760897)
Posted 30 Jan 2016 by Profile Joel
Post:
I just got a new computer so decided to start crunching again after not running for a while. This is the machine:
http://setiathome.berkeley.edu/show_host_detail.php?hostid=7919890


I'm seeing a lot more inconclusive results than I would expect so far. When they get sent out again, we'll see what the third wingman has to say, but I'm concerned that something is going wrong on my end with this many. The CPU is running pretty hot, though I've restricted it to use only 75% of the cores and 75% of the CPU time (so with HT this results in 6 threads running, plus the OpenCL app). All of the inconclusive WUs were on the OpenCL GPU. Anything to be concerned about?
http://setiathome.berkeley.edu/results.php?userid=8972474&offset=0&show_names=0&state=3&appid=
5) Questions and Answers : GPU applications : Fermi and pre-Fermi mix? (Message 1212583)
Posted 31 Mar 2012 by Profile Joel
Post:
Ah, thanks Vyper/Gundolf. I see that it is possible to use both cards in the same machine, but one should use different BOINC directories for each. That's a little bit of trouble, but at least it can be done. Thanks!
6) Questions and Answers : GPU applications : Fermi and pre-Fermi mix? (Message 1211710)
Posted 29 Mar 2012 by Profile Joel
Post:
I haven't been crunching for a while due to the high electric bill, but as my situation changes I am thinking of restarting it. I had two machines, one with a Fermi card (GTX 460) and one with 2x pre-Fermi cards (GTS 250). If I get back into it, I'll probably just use one machine. Last I checked, it was not recommended to mix a Fermi card and a non-Fermi card in one machine using the same BOINC client. I was using Lunatics optimized apps. Is it possible or recommended by now to mix this hardware together and have the right WUs go to the right apps?
7) Message boards : SETI@home Science : SETI institute asking for money (Message 1120221)
Posted 22 Jun 2011 by Profile Joel
Post:
I imagine most of us here heard about the Allen Telescope Array's hibernation a few months ago. Now the SETI Institute (no relation to SETI@home) is asking for public donations to help reopen it.

https://setistars.org/

Of course if you plan to help out, don't forget that SETI@home needs monetary support as well.
8) Message boards : Number crunching : 38g / 275.33 = Much better GPU loading (Message 1119449)
Posted 20 Jun 2011 by Profile Joel
Post:
I've got a GTX460 too, so that is interesting to me. I guess this would reduce or eliminate the need to run Fred's Priority tool, then? Does this benefit only apply to Fermi cards or is this a general improvement for all CUDA cards? (I haven't updated yet)
9) Questions and Answers : Windows : help astropulse (Message 1119447)
Posted 20 Jun 2011 by Profile Joel
Post:
Note that Astropulse units are not always available. Even when things are set up correctly, sometimes you can get them, sometimes you can't.
10) Questions and Answers : Macintosh : boinc / Xgrid (Message 1117485)
Posted 15 Jun 2011 by Profile Joel
Post:
(1) is fine, but I don't necessarily agree with (2). With modern CPUs, there's a much fuzzier definition of "waste cycles". Power management features often throttle down CPU performance when it's not being used heavily. A lot of people run SETI or other BOINC projects on dedicated computers. I'd say the goal of BOINC at this point is to get as much volunteer computing power for scientific projects as possible, whether the cycles are "spare" or not. But the overall point that Xgrid takes special configuration is true. Xgrid is sort of replicating BOINC on smaller, more private scales. It kind of allows you to run an internal BOINC-like project on all the computers your organization owns.
11) Questions and Answers : Macintosh : Kernel Panic Mac Mini (Message 1117484)
Posted 15 Jun 2011 by Profile Joel
Post:
You might want to explain how to reliably repeat it if you are trying to see if anyone else has had the same problem.
12) Questions and Answers : GPU applications : Wish Astropulse in CUDA (Message 1109753)
Posted 25 May 2011 by Profile Joel
Post:
It's worth noting that Lunatics is indeed working on GPU applications for Astropulse, but they're of course volunteers, so be patient.
13) Message boards : Number crunching : CUDA on AGP - is possible? (Message 1109119)
Posted 23 May 2011 by Profile Joel
Post:
I can collect price data, but maybe someone could supply (at least rough) possible performance data? I understand that hardly anyone here running AGP and PCI cards, but maybe relative performance for their PCI-E counterparts at least?


I have used several PCI-E 9500GT cards for MB, and I was getting about 1000 RAC out of each of them using optimized apps, if I remember correctly. I think they did a normal WU in about an hour and a half, maybe a little less.
14) Questions and Answers : Windows : I want to help SETI, not learn software. (Message 1107743)
Posted 19 May 2011 by Profile Joel
Post:
I agree that you are using as few processes as necessary. The only one that will use any appreciable amount of resources is setiathome_6.03_windows_intelx86.exe, which is the one that really really matters, because it's doing the science. If your computer is running slower than you'd like, you can either adjust your SETI preferences to use fewer resources (eg, run only when you are not using the computer, or use only a certain amount of your CPU, etc) or find out what other things might be using resources (perhaps viruses, or even the virus protection software itself...)
15) Message boards : Number crunching : News on Green Bank (Message 1106859)
Posted 16 May 2011 by Profile Joel
Post:
If we're going to get a year's worth of data out of this, that means everyone will get a crack at it. When I first heard that SETI@home would be getting a limited amount of observing time, I figured we'd crunch through it really quickly, but there's much more being recorded at Green Bank than at Arecibo, so even the people turning in just a few WUs here and there will get a chance to find something in this targeted data. And of course those of us with GPUs will have even greater odds of stumbling upon ET than we do now (maybe that's not saying much though!). Very nice news.
16) Questions and Answers : Macintosh : OS update (Message 1106853)
Posted 16 May 2011 by Profile Joel
Post:
It should almost certainly remain the same host after an OS update, so Skildude's suggestion about merging hosts probably won't be necessary, but there's always that option if anything goes weird.
17) Message boards : Number crunching : Total credit for the entire project (Message 1104722)
Posted 9 May 2011 by Profile Joel
Post:
Exactly what I wanted, thanks fellas
18) Message boards : Number crunching : Total credit for the entire project (Message 1103961)
Posted 6 May 2011 by Profile Joel
Post:
I've been looking through the project stats recently, and I have been wondering what is the total amount of credit granted to all S@H users since the move to BOINC? One could write a script to estimate this by scraping it from 500 pages of stats, but I wonder if this is already available somewhere in its entirety. Does this exist, or has anyone done like I said to estimate it by summing up the top 10000 users?
19) Message boards : Number crunching : How much computing power does seti@home need (Message 1103329)
Posted 4 May 2011 by Profile Joel
Post:
FWIW, if S@H's 552 teraflops were all contained in one machine, this would be the 12th most powerful supercomputer in the world.

http://www.top500.org/list/2010/11/100
20) Message boards : Number crunching : AVX Extensions - Ongoing development? (Message 1089729)
Posted 23 Mar 2011 by Profile Joel
Post:
Jason, thanks for the roadmap. Great to know what you guys are up to. I'd help if I could, but it'll be a few years before I manage to squeeze in enough study of computer science to be of much use. My coding skills are too narrowly focused on statistics in my field to help in this arena.

We know what green, yellow and orange are on the chart, but could you maybe elucidate a little bit what the grey bracket pointing to Urs indicates (Urs, feel free to jump in if you are reading ;)) ? Is the AVX/v7/CUDA optimization for Linux/Mac a "longer term" goal? Dare I even ask about CUDA on Mac?

Thanks, and good luck


Next 20


 
©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.