• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

benz4140 is 979 worldwide.

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Well I can answer a couple of questions. First of all you only get credit for a work unit once. You can't retransmit the same unit and get credit for it more than once. Secondly the supercomputer wouldn't do anything for seti unless the client had been hacked and the cluster was able to process the work unit in multiple threads. As it stands the seti client is not a multithreaded application. Also it isn't possible to manually edit a completed work unit and change the reported completion time. Their server uses an algorithm to check the entire result file against the reported cpu time. A similar method is being used to check the results of our benchmark submissions.
 
unfortunately TC is right
seti is not multithreaded and as such cant be run on a Beowulf cluster (without a HACKED client)

the results seem bogus and the fact that the guy hasnt answered any emails or anything only leads me to the same conclusion
 
lonewolf1983 said:
unfortunately TC is right
seti is not multithreaded and as such cant be run on a Beowulf cluster (without a HACKED client)

the results seem bogus and the fact that the guy hasnt answered any emails or anything only leads me to the same conclusion

If seti was Multithreaded I would be having some serious fun! Work units would not even last 2 minutes here :)
 
TC
Correct you can't get credit for the same wu name. Just tested. There still may be a way to assemble the files with completed results utilizing the new wu. Just too tired to try right now.
 
racecar12 said:

Racecar- That's a great find. Blows my supercomputer theory.

Can you find this profile too.

edit: As far as hacking goes, computers can do things very quickly if you automate the process.
The "love virus" generated 10's of millions of eMails in a days time.

Creating 25k result.sah files could be done very quickly once the correct algorithm is determined.

Too bad the orginal eMail to Berkeley didn't have the profile linked then maybe we wouldn't be spending all this time discussing benz4140.
 
Last edited:
hallen said:


Racecar- That's a great find. Blows my supercomputer theory.

Can you find this profile too.

edit: As far as hacking goes, computers can do things very quickly if you automate the process.
The "love virus" generated 10's of millions of eMails in a days time.

Creating 25k result.sah files could be done very quickly once the correct algorithm is determined.

Too bad the orginal eMail to Berkeley didn't have the profile linked then maybe we wouldn't be spending all this time discussing benz4140.
I searched JerryFel and had no luck. benz's profile was right under our nose but kinda missed by all. See the "little head" next to benz's name? That is his profile:D
 
hallen

LOL I found that one also. Really not sure. Can you get info on which isp he uses? That could narrow things down.
 
Hey guys (and gals?),

Just dropping in from the pond to let you in on a little info you apparently do not know. Even in the 3.03 client it is possible to successfully manipulate the apparent time required to complete a work unit. It is entirely possible to show a WU completion time ranging anywhere from a couple of minutes to near infinity.

I found this quirk on accident and got tagged for about 21 days on a WU that actually ran in about 16 hours (P3 under a heavy load). Soon found that it works both ways and you can artificially alter the time reported for completion.

Important note: The actual clock time is unchanged. If your machine requires four hours, then you can do six WU/day, each with "CPU time" = whatever you wish.

It is possible that this could be scripted without much effort.

Personally I still think benz is not on the up-and-up, and am grateful that your team has taken steps to verify his methods. Also grateful for the attitude of tossing cheaters. Nonetheless, I wanted to let ya know that you can cheat, at least the reported WU times. JD (Dorrellco) also knows about this and discovered it much the same as I did and will vouch for the authenticity of my report.

If anyone really cares how it's done and wants to know, I posted about it a long, long time ago in the pond. Good luck finding the post, but it *can* be found.

Happy crunching.
 
# 885. Huge numbers this week.

496 + 421 + 522 = 1439

vs.

Top Crunchers

1.) 273.209 WUs – Benz4140 (03/16/02)

7 * 273.209 = 1912.463
 
Do we know he's real? Do we know he isn't one of us? Watching quietly from within as we discuss the matter. Makes you think....
 
for my experience, i know it could be easily "extend" the WU completed time.

i've put together a new cruncher (just crunching). installed OS, patches, eveything. and it started to crunch. then later, i found out that the system clock (BIOS) wasn't setted correctly, it was still the original, non-adjusted time e.g. 1/1/01. so, i reboot back to BIOS, adjusted the time, then back to windows, and crunch on... and when it finished that WU, i've check the log (from Spy), wow!! it spent almost a year of CPU time to crunch that WU!!!! then i figured out that since SETI recorded the start of that WU crunching time (1/1/01), and the finished time (11/xx/01). so it turns out that my crunch spent 11months to just crunch that single WU!! and i saw my ave. CPU time jump up BIG TIME!!

well, that's my experience to how to "extend" a WU time. "shorten" it?? dunno. and won't try!!
 
I know SETI was contacted before the last major crisis but they have not gotten back to us on this. Maybe Benz is a small fish to them but he is a big fish to us...is it not about time to jog their memory on this issue? :(

Cy
 
# 865

460 + 496 + 421 + 522 = 1899.000

vs.

Top Crunchers

1.) 273.209 WUs – Benz4140 (03/16/02)

7 * 273.209 = 1912.463
 
I know>Just wears my mind out wondering what he is doing to get such results. I wish he would help lead us to the wealth:D
 
Back