Possible way to cheat (don't do this)

作者	消息
Fuzzy Hollynoodles 志愿者测试人员发送消息已加入：3 Apr 99 贴子：9659 积分：251,998 近期平均积分：0	消息 309895 - 发表于：19 May 2006, 0:50:07 UTC - 回复消息 308175. Fuzzy Yes please, my problem here at moment is house alterations, so I can't promise to do any analysis on figures until the kitchen is re-done and that cannot be done until other jobs are complete. Andy Ok, I'll mail the result to Pappa, even he's pretty busy himself. You guys are busy, aren't you?! :-D I'll save the result site anyway, so you can have it when you want. :-) And the WU is really slow, only about 7.6% in more than an hour. And I can't still make Crunch3r's client work, so this is the raw score! Done. :-) "I'm trying to maintain a shred of dignity in this world." - Me ID: 309895 ·

Fuzzy Hollynoodles 志愿者测试人员发送消息已加入：3 Apr 99 贴子：9659 积分：251,998 近期平均积分：0	消息 308175 - 发表于：17 May 2006, 3:06:29 UTC - 回复消息 308157. Fuzzy Yes please, my problem here at moment is house alterations, so I can't promise to do any analysis on figures until the kitchen is re-done and that cannot be done until other jobs are complete. Andy Ok, I'll mail the result to Pappa, even he's pretty busy himself. You guys are busy, aren't you?! :-D I'll save the result site anyway, so you can have it when you want. :-) And the WU is really slow, only about 7.6% in more than an hour. And I can't still make Crunch3r's client work, so this is the raw score! "I'm trying to maintain a shred of dignity in this world." - Me ID: 308175 ·

W-K 666 志愿者测试人员发送消息已加入：18 May 99 贴子：13931 积分：40,757,560 近期平均积分：67	消息 308157 - 发表于：17 May 2006, 2:52:53 UTC - 回复消息 308078. Hans, One of the reasons I kept up the credits questions on Beta, was because I had already seen my credits/results with respect with AR was not in line with the credits awarded, and I already saw the opposite affect at VLAR and posted on Beta accordingly but I don't really have enough data at AR's lower than 0.422. And at moment I have workers in doing house alterations and don't really have time or opportunity at the moment to do more analysis. Andy Andy, I have a 5.12 WU from here at the moment with AR = 0.0080106188031524. Preempted at the moment with 3.38% at 30 minutes with 10 hours to go, according to the manager. If you want the result and creditclaim, when it's finished, please let me know, I can mail it to Pappa, so he can forward it to you, if you want it. Fuzzy Yes please, my problem here at moment is house alterations, so I can't promise to do any analysis on figures until the kitchen is re-done and that cannot be done until other jobs are complete. Andy ID: 308157 ·

Fuzzy Hollynoodles 志愿者测试人员发送消息已加入：3 Apr 99 贴子：9659 积分：251,998 近期平均积分：0	消息 308078 - 发表于：17 May 2006, 1:02:17 UTC - 回复消息 306965. Hans, One of the reasons I kept up the credits questions on Beta, was because I had already seen my credits/results with respect with AR was not in line with the credits awarded, and I already saw the opposite affect at VLAR and posted on Beta accordingly but I don't really have enough data at AR's lower than 0.422. And at moment I have workers in doing house alterations and don't really have time or opportunity at the moment to do more analysis. Andy Andy, I have a 5.12 WU from here at the moment with AR = 0.0080106188031524. Preempted at the moment with 3.38% at 30 minutes with 10 hours to go, according to the manager. If you want the result and creditclaim, when it's finished, please let me know, I can mail it to Pappa, so he can forward it to you, if you want it. "I'm trying to maintain a shred of dignity in this world." - Me ID: 308078 ·

Hans Dorn 志愿者开发人员志愿者测试人员发送消息已加入：3 Apr 99 贴子：2262 积分：26,448,570 近期平均积分：0	消息 306971 - 发表于：16 May 2006, 3:45:09 UTC - 回复消息 306965. And at moment I have workers in doing house alterations and don't really have time or opportunity at the moment to do more analysis. Andy I have some data points coming up for you: 12hrs CPU time and still counting, on a P4 3.2 :o) Regards Hans ID: 306971 ·

W-K 666 志愿者测试人员发送消息已加入：18 May 99 贴子：13931 积分：40,757,560 近期平均积分：67	消息 306965 - 发表于：16 May 2006, 3:39:25 UTC - 回复消息 306746. EDIT: possible lesson: don't mind credits so serious!! Agreed. If further optimization is possible (I hope so), then the whole image will change again. Regards Hans Hans, One of the reasons I kept up the credits questions on Beta, was because I had already seen my credits/results with respect with AR was not in line with the credits awarded, and I already saw the opposite affect at VLAR and posted on Beta accordingly but I don't really have enough data at AR's lower than 0.422. And at moment I have workers in doing house alterations and don't really have time or opportunity at the moment to do more analysis. Andy ID: 306965 ·

Hans Dorn 志愿者开发人员志愿者测试人员发送消息已加入：3 Apr 99 贴子：2262 积分：26,448,570 近期平均积分：0	消息 306746 - 发表于：15 May 2006, 22:04:17 UTC - 回复消息 306687. EDIT: possible lesson: don't mind credits so serious!! Agreed. If further optimization is possible (I hope so), then the whole image will change again. Regards Hans ID: 306746 ·

Tetsuji Maverick Rai 志愿者测试人员发送消息已加入：25 Apr 99 贴子：518 积分：90,863 近期平均积分：0	消息 306687 - 发表于：15 May 2006, 21:19:46 UTC 最近的修改日期：15 May 2006, 21:40:28 UTC I checked Eric's Sun/SPARC and found the credits per time is opposite from Intel processors; large AR has smaller credits / time than smaller AR!! see this post. It's clearly difference between architectures. Then we should go back to benchmark/cobblestone? I don't like it....I'm at a loss.. Hans may be right in "it's even worse"... Anyway I raised this problem in a different aspect! I'm going to bed!! EDIT: possible lesson: don't mind credits so serious!! ID: 306687 ·

Tetsuji Maverick Rai 志愿者测试人员发送消息已加入：25 Apr 99 贴子：518 积分：90,863 近期平均积分：0	消息 306628 - 发表于：15 May 2006, 20:22:51 UTC I suggested this unfairness in beta project message board. The difference is much more than twice!! //TMR Luckiest in the world. WMD = Weapon of Mass Distraction. Click this table. ID: 306628 ·

Tetsuji Maverick Rai 志愿者测试人员发送消息已加入：25 Apr 99 贴子：518 积分：90,863 近期平均积分：0	消息 306567 - 发表于：15 May 2006, 19:02:52 UTC 最近的修改日期：15 May 2006, 19:13:13 UTC Actually I agree in that current fpops estimate is not accurate (in my impression larger AR gets about twice credits per time, and I like to keep many large AR WU's in my cache...in beta project! Here in public project, I keep my RAC 13; ie one of my lucky numbers). But if it's corrected, chance of making cheating "optimized" clients is very slim or negligible. But there's still chance if one uses core clients prior to 5.2.6. Luckiest in the world. WMD = Weapon of Mass Distraction. Click this table. ID: 306567 ·

Hans Dorn 志愿者开发人员志愿者测试人员发送消息已加入：3 Apr 99 贴子：2262 积分：26,448,570 近期平均积分：0	消息 306500 - 发表于：15 May 2006, 17:29:08 UTC - 回复消息 306495. 最近的修改日期：15 May 2006, 17:38:15 UTC Snipped some elaborate cheating scheme... So I think we need to take such cheating into consideration. I couldn't say this before... It gets even worse. When you look at the credits/hr you get for different angle rates, you'll find huge discrepancies. It would be easy to automatically cancel "bad" WUs and increase your RAC. Since calculation times are so long, even a drop in your daily WU quota wouldn't matter much. At the moment at least, the flop counting doesn't agree very well with the actual crunching times per WU. Regards Hans (sarcasm) It's at least as broken as the benchmarking for the standard app is. (/sarcasm) ID: 306500 ·

Tetsuji Maverick Rai 志愿者测试人员发送消息已加入：25 Apr 99 贴子：518 积分：90,863 近期平均积分：0	消息 306495 - 发表于：15 May 2006, 17:24:29 UTC 最近的修改日期：15 May 2006, 17:26:15 UTC Now we are running _enhanced, getting credits based on fpops, I think I can say something about this one. Without fpops, this cheating is easy to do, and difficult to find. Since the validator compares results we return and gives credits if half or more of the signals detected are in accordance with the canonical result, one can cheat by skipping some detecting code. Say, if one makes a cruncher which skips several signal checking codes (for example, skip 1/3 of chirp/fft pairs, or skip gaussian search, etc...), the results will be likely validated as "weakly similar", and it's more than 30% -40% faster. But now we count credits according to fpops, we can find such an application (if any) when a cruncher claims clearly less, but is validated. One may still be able to make such a cheating application but it's more difficult than before. Fortunately now SETI@home team claims to make the source open to public if one makes optmized clients and distributes. I couldn't say this before... Luckiest in the world. WMD = Weapon of Mass Distraction. Click this table. ID: 306495 ·

©2020 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.