Questions and Answers :
Unix/Linux :
Very slow
Message board moderation
Author | Message |
---|---|
Bergelin Send message Joined: 5 May 06 Posts: 12 Credit: 39,730 RAC: 0 |
I have used seti at home on a windows xp platform for a while. I changed to Linux Suse 10.1 for a week ago, and Boinc is working perfectly. My only problem is that it is taking many hours longer to complete a work compared to when i used it in Windows. Is this how it is suposed to be, or is something wrong? Sorry for my bad english, I live in Sweden. Many thanks! /Markus |
Dotsch Send message Joined: 9 Jun 99 Posts: 2422 Credit: 919,393 RAC: 0 |
Since SETI Enhanced the computing times depending on the angle rate on which the WU was recorded. The computingtimes differs by a factor of 10. So your systems are only compareable with WUs with the same angle rate. |
Bergelin Send message Joined: 5 May 06 Posts: 12 Credit: 39,730 RAC: 0 |
Since SETI Enhanced the computing times depending on the angle rate on which the WU was recorded. The computingtimes differs by a factor of 10. How do I se what angle rate my WU have? |
Dotsch Send message Joined: 9 Jun 99 Posts: 2422 Credit: 919,393 RAC: 0 |
Since SETI Enhanced the computing times depending on the angle rate on which the WU was recorded. The computingtimes differs by a factor of 10. It is included in the WU header, you can "grep true_angle_range" in your WUs in the projects/setiathome.../ directory. |
Bergelin Send message Joined: 5 May 06 Posts: 12 Credit: 39,730 RAC: 0 |
Since SETI Enhanced the computing times depending on the angle rate on which the WU was recorded. The computingtimes differs by a factor of 10. Thank you, now I see it. |
NH7S Send message Joined: 20 Mar 06 Posts: 1 Credit: 3,931,758 RAC: 0 |
I'm using Fedora Core 5. MFLOPS was showing about 770 under FC5. With Windows XP Home MFLOPS was about 1400. Same machine a PENT D 930. Although MFLOPS was reporting about 1/2 under FC5 VS XP, the progress being done on a work unit appears to be the same (subjective, no objective mesurement was made). So if your work unit completion has increased by hours, I'd say you do have a problem. |
Dotsch Send message Joined: 9 Jun 99 Posts: 2422 Credit: 919,393 RAC: 0 |
I'm using Fedora Core 5. MFLOPS was showing about 770 under FC5. With Windows XP Home MFLOPS was about 1400. Same machine a PENT D 930. Although MFLOPS was reporting about 1/2 under FC5 VS XP, the progress being done on a work unit appears to be the same (subjective, no objective mesurement was made). No, he has not a problem. It is a known issue, that the boinc benchmarks under linux are slower than the benchs under Windows. Also, you can only compare WU processing times with the same angle rate. There are a lot of different angle rates send out with SETI Enhanced. And the computing time depends on the angle rate. |
Bergelin Send message Joined: 5 May 06 Posts: 12 Credit: 39,730 RAC: 0 |
Actually, after comparing my systems, I would say that the time it takes to analyse a WU is quite the same. |
Bergelin Send message Joined: 5 May 06 Posts: 12 Credit: 39,730 RAC: 0 |
Is the optimized clients much better then the usual BOINC? What do you use? |
Dotsch Send message Joined: 9 Jun 99 Posts: 2422 Credit: 919,393 RAC: 0 |
Is the optimized clients much better then the usual BOINC? What do you use? The optimisations and higher benchmarks of the boinc client did not mather, because the SETI Enhacned are measuring the comuted FLOPS. I am not shure, if there will be a optimized SETI Enhanced application for Linux and Windows is available. |
Bergelin Send message Joined: 5 May 06 Posts: 12 Credit: 39,730 RAC: 0 |
It exists optimized clients for Linux and WindowsIs the optimized clients much better then the usual BOINC? What do you use? Will an optimized client not run faster? |
Justin La Sotten Send message Joined: 26 Jul 99 Posts: 11 Credit: 29,855,985 RAC: 0 |
I'm having the same problem. Under windows xp i was averaging around 400-500 credits per day on a P4 3.0 HT proc with an average time of about an hour and a half. Under Ubuntu 6.06 with SMP, i'm averaging around 200-300 with an average time of about 3 hours. This is a pretty significant decrease. I understand the angle rate effects processing times, however, i'm looking at the data over a period of time, the linux machine is in every case significantly slower in every single case, there hasn't been a single time where the processing is faster on the linux side than it is on the windows side. Could there be any other reason for this? My kernel is setup to understand SMP with HT. Any help would be appreciated. |
Dotsch Send message Joined: 9 Jun 99 Posts: 2422 Credit: 919,393 RAC: 0 |
I'm having the same problem. No, I think you could not only compare the RAC of the two OSes. Over a long period of time you are not shure that you get WUs with the same ARs. The only way to compare is to take a bunch of WUs (best with different ARs) and compute it (with normal priority) with the app in standalone mode on both systems and compare the runtimes of the WUs. |
©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.