Message boards :
Number crunching :
overclocking, seti stability and companies that rape you
Message board moderation
Previous · 1 · 2
Author | Message |
---|---|
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
Even so, the race car has to be reliable: if it breaks before the final lap, it won't win even if it's the fastest car out there. Haven't read or followed Mr. Chapman, but it's IOTTMCO. The car that self destructs just past the finish line did in fact finish. If it self-destructs just a tiny bit before the finish line is guaranteed to have lost. Those who don't have an engineering background may not really be thinking about reliability and margins. Back in the day (late-70's) I worked on mainframe design. One machine I worked on had a 60ns clock. It had to be shown running perfectly at 55ns before it could leave the factory (at 60, not 55). Overclockers are trading a reduced margin for extra speed. You can do that if your system has lots of extra margin, but you can't reduce the margin to 0 (or less) and have a reliable machine. |
Jim Send message Joined: 28 Jan 00 Posts: 614 Credit: 2,031,206 RAC: 0 |
I can't speak to the mindset or philosophies of others, but for myself if my rig performs flaky in any way in the long term (days, weeks) it's not a successful overclock. I'll admit again that I sent in a day's worth of 0 credit results on one of my machines. It was stable under every test I have. However, I am using Crunch3r's optimized apps and clients and, just as I was fairly warned, I sent out crud. I backed off about 3 or 5 MHz on the fsb and it's all sweetness and light now. I feel no guilt for the bad results. With my faster-than-OEM rigs I can do more science! It's win-win in any way I can see it. Without love, breath is just a clock ... ticking. Equilibrium |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
I can't speak to the mindset or philosophies of others, but for myself if my rig performs flaky in any way in the long term (days, weeks) it's not a successful overclock. I'll admit again that I sent in a day's worth of 0 credit results on one of my machines. It was stable under every test I have. However, I am using Crunch3r's optimized apps and clients and, just as I was fairly warned, I sent out crud. I backed off about 3 or 5 MHz on the fsb and it's all sweetness and light now. DOH..... there you go confusing the issue with logic and common sense! :-) That's one of the really cool things I like about BOINC, pushing your HW successfully over a variety of the projects is one of the toughest real world test environments around. Best of all, you get the opportunity to accomplish something else useful during the testing as well, instead of just heating up the room. ;-) Alinator |
Steve Cressman Send message Joined: 6 Jun 02 Posts: 583 Credit: 65,644 RAC: 0 |
Been running a 17% overclock on this system for a couple years on air cooling. Never a problem even with Crunch3r's app. But I from time to time do run test to make sure of stability. If it is not stable it is not a succesful overclock. This system is busy doing more than just Boinc so stability is very important to me for more than one reason :) This system can go much higher with better cooling. Borrowed from a friend his cryo cooling setup and was able to go from stock 1.83 to just over 3.4Ghz. Wish I had a 1000 bucks lying around to get one for myself. He runs a 2.2 at 4.2Ghz with this cooling unit. Too Cool. 98SE XP2500+ @ 2.1 GHz Boinc v5.8.8 And God said"Let there be light."But then the program crashed because he was trying to access the 'light' property of a NULL universe pointer. |
©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.