留言板 :
Number crunching :
Seti Enhanced vs Seti Enhanced Optimized
留言板合理
| 作者 | 消息 |
|---|---|
|
Odysseus 发送消息 已加入:26 Jul 99 贴子:1808 积分:6,701,347 近期平均积分:6
|
The bit I refer to is "must have an effect on correct results". How do we know that the changes to the original SETI app that non-SETI people are making are correct, and that they are not introducing bugs when they tweak the code to run faster? That’s what the validator is for. No result is granted credit until it’s been compared with at least two others (forming a “quorumâ€) and found to to match. So if someone were to release a defective app, its users would soon realize that something was wrong: they’d be earning little credit and many of their results would be highlighted in pink on the website. The same applies to users who overclock their systems beyond the point of stability. Even the stock apps on a healthy system will occasionally produce invalid results, perhaps where the math functions built into different types of CPU round off differently in the sixteenth decimal place or whatever, causing a discrepancy between systems, but these errors are quite rare, less than 1% I should think. |
|
Paul Russell 发送消息 已加入:18 Aug 99 贴子:6 积分:136,513 近期平均积分:0
|
Stargate's post contained something that I've wondered about ever since I first read stuff about the optimised apps: I think people have forgotten what we are doing here is crunching WU's as we get them, any patches,fixes or other for the best result (come on guys) it must have an effect on correct results?...It's more like a game of "Battle Ship" Here... The bit I refer to is "must have an effect on correct results". How do we know that the changes to the original SETI app that non-SETI people are making are correct, and that they are not introducing bugs when they tweak the code to run faster? SETI have a pool of testers who crunched loads of workunits for ages in the Beta programme before Enhanced released, and we still see serious bugs creeping in that require a new version (.15) to be released quickly after the Enhanced app was let loose. Seems to me that I can have a stab at "optimising" the app, screw it up totally, release it as "Paul's Optimised App" having 'tested' it for two days had then feed rubbish results back until a problem is spotted by SETI. If a problem is spotted. Or am I missing something obvious? I've never bothered installing an optimised app, by the way. Regards, Paul. |
|
n7rfa 发送消息 已加入:13 Apr 04 贴子:370 积分:9,058,599 近期平均积分:0
|
Is there going to be a Enhanced version where people will not have to worry about looking for the next Patch (Cheat) to crunch WU's faster? 'SORRY' Crunch3r's Optimized Apps already provide improved computing time. Here's the 4 results returned on a WU: Computer FP Speed BOINC App CPU Time Claimed P4 3.00GHz 1316.7 5.2.13 Std App 24864 62.45 PD 2.80GHz 1388.62 5.2.13 Std App 21388 62.45 AMD3200+ 1055.33 5.2.13 Std App(?) 21861 62.45 PD 3.00GHz 1497.11 5.4.8 SSE3 Opt 12158 62.81 As you can see there is about a 40+% reduction in CPU time between the 2 fastest results. Of course some of the improvement could be because of the HW/OS.
|
|
1mp0£173 发送消息 已加入:3 Apr 99 贴子:8423 积分:356,897 近期平均积分:0
|
Is there going to be a Enhanced version where people will not have to worry about looking for the next Patch (Cheat) to crunch WU's faster? 'SORRY' It took a while before some pretty talented folks found a way to make the old SETI/BOINC apps faster -- and many of those optimizations are already in 5.x. Some of those tricks are pretty impressive. Others come from tuning the complier a little closer to the specific CPU type and are good for a few percent. The big gains are going to take folks a while to find, so I wouldn't expect big gains in the next week or two. I could be wrong, but waiting is. |
Jim-R. 发送消息 已加入:7 Feb 06 贴子:1494 积分:194,148 近期平均积分:0
|
Hi Jim R my email is jrosstheman at bellsouth.net Jim Some people plan their life out and look back at the wealth they've had. Others live life day by day and look back at the wealth of experiences and enjoyment they've had. |
Stargate 发送消息 已加入:19 May 02 贴子:60 积分:53,358 近期平均积分:0
|
Hi Jim R Have you got a contact # Mate |
Jim-R. 发送消息 已加入:7 Feb 06 贴子:1494 积分:194,148 近期平均积分:0
|
Ok!! It looks like with 5.15 is the going thing, I ask the same question again is there a need for the Optimized Enhance, I'm only asking this as I would like to keep upto date, on options... There is still a need for optimized enhanced programs if you want to get the most crunching speed possible, however there is not nearly as much speed improvement as with the optimized 4.18 crunchers. The 4.18 "standard seti" application was not optimized, or just basically optimized by removing debugging info, etc. from the executable. The new "enhanced seti" programs are already optimized for speed as much as possible and still be useable on any and all computers regardless of cpu's, etc. The only "optimizations" that need to be done are for the various special instruction sets available on the various cpu's, such as mmx, sse, sse2, sse3 etc. And since the main application is already "optimized" for crunching speed, these optimizations do not result in nearly as much improvement in speed. So if you want to squeeze the most out of your system, yes, go for the latest version of the optimized applications you can find. However if you don't want the hassle of keeping up to date with the app_info.xml file changes needed to keep up with all the latest version changes, I would suggest with just going for the standard application until the versions get settled down into the "final release version". Then if you want to squeeze every last drop of performance out of your computer then you can get an optimized cruncher for what I've heard is about a 20% improvement in speed. This is in huge contrast to the 300% or more improvements in the standard vs optimized 4.18 seti cruncher, but it is still a respectable improvement. Jim Some people plan their life out and look back at the wealth they've had. Others live life day by day and look back at the wealth of experiences and enjoyment they've had. |
Stargate 发送消息 已加入:19 May 02 贴子:60 积分:53,358 近期平均积分:0
|
Ok!! It looks like with 5.15 is the going thing, I ask the same question again is there a need for the Optimized Enhance, I'm only asking this as I would like to keep upto date, on options... Is there going to be a Enhanced version where people will not have to worry about looking for the next Patch (Cheat) to crunch WU's faster? 'SORRY' It's strange as it happened with CLASSIC, We forgot the meaning of the project... I think people have forgotten what we are doing here is crunching WU's as we get them, any patches,fixes or other for the best result (come on guys) it must have an effect on correct results?...It's more like a game of "Battle Ship" Here... I'm sick of seeing "How" do I get my comp working till it's white hot, then when she breaks down it's Boinc's fault, run the WU as is... If you don't want the graphic side of things and there was a fair way of doing it without patches, it would be in the Add-On's "it's not" so do it the correct way ... I'm Sorry for speaking my mind, If I'm an arsehole so be it... I tried asking this Question earlier when I started this post, but got bugger all from it, Apart from Mike, Thankyou Mike... I hope massive changes come about or this project is doomed before it gets full steam... Sorry for what I've said, BUT as they say the truth can hurt, at times... Stargate :-| |
Mike 发送消息 已加入:17 Feb 01 贴子:32210 积分:79,922,639 近期平均积分:80
|
So far as i know its only optimized in speed. No graphics for sure. Mike With each crime and every kindness we birth our future. |
Stargate 发送消息 已加入:19 May 02 贴子:60 积分:53,358 近期平均积分:0
|
Hi, would like to know if there is any major difference between the standed Enhanced compaired to the Optimized versions like the(Crunching Speed)of a WU?? As I was under the impression that when the Enhanced version came along, that Optimizing would not be needed is true or false?? Any views on this would be great Regards Stargate |
©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.