Message boards : Number crunching : sam25
Message board moderation
Author | Message |
---|---|
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
My first 4 from batch 742 passed the zip 12 + restart files point late yesterday. The restarts are a little over 111 Megs. And my old HP computer which suffered a hd failure, now has a nice new hd. (HPs are really great to work with, with all of the hinged cages for getting at things.) Because I didn't have a disk with Windows 7 on it, I couldn't get the data back, so it's now running Mint 18.3 + Wine + BOINC (Windows). And it has 7 batch 742, 5 of them with 1 failure, and 2 with 2 failures. 3 of the 1s and one of the 2s are running, and all have passed the zip 2 point. So I don't know why so many computers can't run them. |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
The first 4 have now completed. "out" files are about 315-120 Kb. About 8.5 days to complete. |
Send message Joined: 15 Jan 06 Posts: 637 Credit: 26,751,529 RAC: 653 |
The first 4 have now completed. Are you running on full cores? They are taking a little over 9 days on my i7-4771 (Win7 64-bit), even though I have limited it to 4 virtual cores, with nothing but desktop use on the other cores. (And I am not overclocking.) I have been wondering how useful hyperthreading is on CPDN. |
Send message Joined: 5 Aug 04 Posts: 1120 Credit: 17,202,915 RAC: 2,154 |
I have been wondering how useful hyperthreading is on CPDN. I used to have a 32-bit computer with two 3.06 GHz hyperthreaded Xeon processor chips, 8 GBytes RAM, etc. Running Linux. So it thought I had four processors. I ran somewhat faster than when I turned on hyperthreading, but by no means twice as fast. I have a newer 64-bit machine now with one 1.8 GHz Xeon 4-core processor, and it seems to run work units of other projects just as fast as before. I do not know if the architecture is responsible, that 64-bits is better, or that the Linux OS is better. And I cannot tell about Climate Prediction, since there are no Linux tasks being offered. |
Send message Joined: 15 Jan 06 Posts: 637 Credit: 26,751,529 RAC: 653 |
So it thought I had four processors. I ran somewhat faster than when I turned on hyperthreading, but by no means twice as fast. Hyperthreading allows you to run more threads than the real cores you have, usually with an increase in total output. But I am limiting CPDN to four virtual cores, which is the same as the number of real cores. So my issue is whether I would just do better by turning off HT. I have had suspicions in the past that might help, especially since CPDN is based on older code that might not benefit from HT much. (I think I will just turn it off for the next batch, and see.) |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
When the nam50's showed up, I received 8 of them, and, not having run that many at once for years, I let all of them run. Which slowed down the computer badly, and took 4.5 days to complete. So for the next batch, I did what I've done for years, and Suspended 4 of them while they were still downloading. The 4 that were left took 2 days to run, so same as I've always found it to be: only run 4 at a time. |
Send message Joined: 15 Jan 06 Posts: 637 Credit: 26,751,529 RAC: 653 |
OK, thanks. I now have a second Windows machine (an i-4790) on it, and the SAM25's seem to be doing fine with hyperthreading, so I will leave it on for now. |
Send message Joined: 18 Jul 13 Posts: 438 Credit: 25,620,508 RAC: 4,981 |
I have 4 WUs from batch 742 running on i5-2520M CPU @ 2.50GHz (2 real cores) and so far these WUs are at 26 days completed below 90%. I expect them to run at 5-7 days more. Avg sec/TS at around 17.2 when all 4 are running. First zips at 14.8 and climbing. Haven't tried it on the i7-4790 with WINE yet. |
Send message Joined: 15 May 09 Posts: 4538 Credit: 19,004,017 RAC: 21,574 |
(HPs are really great to work with, with all of the hinged cages for getting at things.) They might be now but I remember a time when they were a real pain with lots of proprietary bits so it was difficult to find things to fit. One I worked on had non standard motherboard fittings so I couldn't change it without drilling some new holes to fit risers into. |
Send message Joined: 18 Jul 13 Posts: 438 Credit: 25,620,508 RAC: 4,981 |
I figured out why it takes 30 days to complete these 742s on i5-2520M. My laptop overheats and reduces the frequency to 797 MHz for few seconds so it seems half of time runs at this frequency. Even with 2 WUs it is too hot. Need to do some clean up and than will see |
Send message Joined: 7 Aug 04 Posts: 2187 Credit: 64,822,615 RAC: 5,275 |
I figured out why it takes 30 days to complete these 742s on i5-2520M. My laptop overheats and reduces the frequency to 797 MHz for few seconds so it seems half of time runs at this frequency. Even with 2 WUs it is too hot. Need to do some clean up and than will see Besides cleaning out the vents, a cooling pad would help if that's something you're willing to deal with. Even a platform that tilts/raises the bottom of it off the desk so more air can move around may be somewhat helpful. |
Send message Joined: 15 May 09 Posts: 4538 Credit: 19,004,017 RAC: 21,574 |
Yes, mine runs noticeably cooler if I put something under it to raise it a bit, though it only slowed down without doing that when the temperature was up around 30C this summer. |
Send message Joined: 18 Jul 13 Posts: 438 Credit: 25,620,508 RAC: 4,981 |
All my laptops are lifted for better air circulation. I cleaned the fans and put new thermal paste on two of them. The T61 runs 10 degrees cooler 74-75 deg, the T420/i5-2520M runs 20 degree cooler though at 2 WUs for the moment. |
©2024 cpdn.org