对FAH的一个疑问

20080030成电

看FAH的介绍，它说不把代码公开是防止有人伪造结果。但SETI的代码是开源，难道SETI就不怕伪造？
各种软件硬件都可能参加计算，所以FAH要有相应的版本。代码在不同的平台移植工作量大，如果是代码公开，好处是明显的。
一个计算的进程的工作，一般是1读文件，得到参数， 2计算， 3把结果写入文件。计算是最耗时的，也是最应该优化。1和3的代码不公开，只公开2的代码。仅仅是根据2的代码是不能伪造结果的。
只是我的猜测，我也不熟悉FAH和代码。

Julian_Yuen

听起来觉得挺有道理的。
坐沙发，等高人来解释

Youth

不太清楚fah的结果验证机制，我觉得如果任务包有冗余计算的话，应该就不需要担心作弊的情况，毕竟这应该是极少数用户的行为。

20080030成电

和验证机制无关。
比如 1从文件A读出一行字母ABC，2把字母反序成CBA，3，按自定义的格式写入文件B，把文件B上传给服务器。

如果只知道2，不知道3的格式，是不能伪造结果的。

Youth

就比如你说的2到3要传递一个计算结果，但这个总归也是文件，任何一个过程只要公开就有可能作假

zasdfgbnm

作假？
那他还算分布式干什么？
算分布式计算的目的是为了给科学事业做贡献的，而不是刷分的。

Julian_Yuen

不过遗憾的是，这种人貌似还真的存在，至少存在过....

foxhound

原帖由 gaoxiang5220977 于 2008-4-6 17:31 发表
作假？
那他还算分布式干什么？
算分布式计算的目的是为了给科学事业做贡献的，而不是刷分的。

以前貌似就在某个项目中出现过造假刷分的事件

zasdfgbnm

真是无聊，分又不能换钱，有刷分的必要吗？

Youth

volunteer computing 本质上是不可信赖的计算环境，所以要有冗余计算之类的

原帖由 gaoxiang5220977 于 2008-4-6 17:31 发表
作假？
那他还算分布式干什么？
算分布式计算的目的是为了给科学事业做贡献的，而不是刷分的。

cnchina

SETI CLASSIC吧？好像在哪个帖子里说过。。后来就不让合并帐号了。

PAGAN

最近刚买了部PS3，说实话最想“玩”的就是FAH@PS3哈哈～ 于是有机会比较了三个平台的FAH计算速度：

PS3 FAH的速度是0.15帧/秒，三个多小时完成一个工作单元；

E4300oc2.7GHz with X1950Pro GPU计算， CPU一个核心满载， 1%需要8分钟，13小时多才能完成1个工作单元！

QX9650oc3.68GHz 四路计算 每路命令行下1%需要35分钟！！！3500分钟完成4个工作单元，平均14个半小时才完成1个工作单元！！！

除了在PS3平台上看到斯坦福大学与SCE在Cell架构上有深度性能发掘之外，在X86及GPU上的计算效能真是太低了……

[ 本帖最后由 PAGAN 于 2008-4-15 17:02 编辑 ]

BiscuiT

HD 3850 有版友测试，12分钟完成450帧，这样也有0.625帧/秒
比PS3快4倍咯。。

而且 HD 3870 不到10分钟。。

RV770XT 5月要发布了，R600的暴力衍生物。。性能更加可观～

vmzy

官方在论坛里说了GPU是最快的客户端

宝康

3870的Log，大概可以看出将近2个小时一个wu吧，显卡和CPU没超频

[09:22:29] - Preparing to get new work unit...
[09:22:29] + Attempting to get work packet
[09:22:29] - Connecting to assignment server
[09:22:30] - Successful: assigned to (171.64.65.103).
[09:22:30] + News From Folding@Home: GPU folding beta
[09:22:30] Loaded queue successfully.
[09:22:31] + Closed connections
[09:22:31]
[09:22:31] + Processing work unit
[09:22:31] Core required: FahCore_11.exe
[09:22:31] Core found.
[09:22:31] Working on Unit 09 [April 15 09:22:31]
[09:22:31] + Working ...
[09:22:32]
[09:22:32] *------------------------------*
[09:22:32] Folding@Home GPU Core - Beta
[09:22:32] Version 1.00 (Fri Nov 2 13:40:51 PDT 2007)
[09:22:32]
[09:22:32] Compiler  : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.42 for 80x86
[09:22:32] Build host: Host
[09:22:32] Preparing to commence simulation
[09:22:32] - Looking at optimizations...
[09:22:32] - Created dyn
[09:22:32] - Files status OK
[09:22:32] - Expanded 45400 -> 246512 (decompressed 542.9 percent)
[09:22:32] Called DecompressByteArray: compressed_data_size=45400 data_size=246512, decompressed_data_size=246512 diff=0
[09:22:32] - Digital signature verified
[09:22:32]
[09:22:32] Project: 2799 (Run 0, Clone 262, Gen 11)
[09:22:32]
[09:22:32] Assembly optimizations on if available.
[09:22:32] Entering M.D.
[09:22:38] Working on 582 p2799_N68H_AM03
[09:22:38] Starting GUI Server
[09:23:55] Completed 1%
[09:25:04] Completed 2%
[09:26:12] Completed 3%
[09:27:21] Completed 4%
[09:28:29] Completed 5%
[09:29:38] Completed 6%
[09:30:46] Completed 7%
[09:31:55] Completed 8%
[09:33:03] Completed 9%
[09:34:12] Completed 10%
[09:35:21] Completed 11%
[09:36:29] Completed 12%
[09:37:38] Completed 13%
[09:38:46] Completed 14%
[09:39:55] Completed 15%
[09:41:04] Completed 16%
[09:42:12] Completed 17%
[09:43:20] Completed 18%
[09:44:30] Completed 19%
[09:45:38] Completed 20%
[09:46:47] Completed 21%
[09:47:56] Completed 22%
[09:49:04] Completed 23%
[09:50:13] Completed 24%
[09:51:21] Completed 25%
[09:52:30] Completed 26%
[09:53:38] Completed 27%
[09:54:47] Completed 28%
[09:55:55] Completed 29%
[09:57:04] Completed 30%
[09:58:13] Completed 31%
[09:59:21] Completed 32%
[10:00:29] Completed 33%
[10:01:37] Completed 34%
[10:02:46] Completed 35%
[10:03:55] Completed 36%
[10:05:04] Completed 37%
[10:06:12] Completed 38%
[10:07:20] Completed 39%
[10:08:29] Completed 40%
[10:09:37] Completed 41%
[10:10:45] Completed 42%
[10:11:53] Completed 43%
[10:13:02] Completed 44%
[10:14:10] Completed 45%
[10:15:19] Completed 46%
[10:16:27] Completed 47%
[10:17:35] Completed 48%
[10:18:43] Completed 49%
[10:19:52] Completed 50%
[10:21:00] Completed 51%
[10:22:08] Completed 52%
[10:23:17] Completed 53%
[10:24:25] Completed 54%
[10:25:33] Completed 55%
[10:26:41] Completed 56%
[10:27:50] Completed 57%
[10:28:58] Completed 58%
[10:30:06] Completed 59%
[10:31:15] Completed 60%
[10:32:23] Completed 61%
[10:33:31] Completed 62%
[10:34:40] Completed 63%
[10:35:49] Completed 64%
[10:36:58] Completed 65%
[10:38:07] Completed 66%
[10:39:15] Completed 67%
[10:40:24] Completed 68%
[10:41:32] Completed 69%
[10:42:41] Completed 70%
[10:43:50] Completed 71%
[10:44:58] Completed 72%
[10:46:07] Completed 73%
[10:47:15] Completed 74%
[10:48:24] Completed 75%
[10:49:32] Completed 76%
[10:50:41] Completed 77%
[10:51:49] Completed 78%
[10:52:58] Completed 79%
[10:54:07] Completed 80%
[10:55:16] Completed 81%
[10:56:25] Completed 82%
[10:57:33] Completed 83%
[10:58:42] Completed 84%
[10:59:50] Completed 85%
[11:00:59] Completed 86%
[11:02:07] Completed 87%
[11:03:16] Completed 88%
[11:04:24] Completed 89%
[11:05:33] Completed 90%
[11:06:42] Completed 91%
[11:07:50] Completed 92%
[11:08:59] Completed 93%
[11:10:07] Completed 94%
[11:11:16] Completed 95%
[11:12:24] Completed 96%
[11:13:33] Completed 97%
[11:14:41] Completed 98%
[11:15:50] Completed 99%
[11:16:59] Completed 100%
[11:17:59]
[11:17:59] Finished Work Unit:
[11:17:59] - Reading up to 1434888 from "work/wudata_09.trr": Read 1434888
[11:17:59] trr file hash check passed.
[11:17:59] - Reading up to 456628 from "work/wudata_09.xtc": Read 456628
[11:17:59] xtc file hash check passed.
[11:17:59] edr file hash check passed.
[11:17:59] logfile size: 92721
[11:17:59] Leaving Run
[11:18:02] - Writing 1985309 bytes of core data to disk...
[11:18:02] Done: 1984797 -> 1802003 (compressed to 90.7 percent)
[11:18:02]   ... Done.
[11:18:02] - Shutting down core
[11:18:02]
[11:18:02] Folding@home Core Shutdown: FINISHED_UNIT
[11:18:06] CoreStatus = 64 (100)
[11:18:06] Sending work to server