PDA

View Full Version : Eric contacted about "benz4140"


Morpheus
01-18-02, 06:38 PM
At Tim's request, I have sent an email off to Dr. Korpella @ SETI... an excerpt of that email is as follows:

After a long time of crunching for the team his output has risen quickly to a suspiciously fast level—currently producing WUs at a rate of about 230 units a day. Further, Benz is not responding to our email (nor does he participate in our team forum).

The OC.com SETI Team would consider it a real favor if someone on your end could verify the integrity of the WUs (and/or SETI client) in question. As always, our team values the integrity of the science behind the project and would not want to be associated with individuals devaluing that effort.

We look forward to hearing from you and appreciate your assistance,

Kris Kurrus

TC
01-18-02, 06:42 PM
Thanks

chawken
01-18-02, 06:44 PM
Nicely stated Kris.

Cy
01-18-02, 07:36 PM
Very nice Kris. All we want as a team are dedicated members crunching as much as possible at all costs except for cheating. If his units are legit then we are exceptionally lucky to have his speed. If they are not, then we need to uphold the integrity of our team, it's members and the science represented by the SETI@Home project.

Cy

rogerdugans
01-18-02, 08:07 PM
Nicely put.
It is a lousy situation to be in, but its definitely the right thing to do.

LandShark
01-19-02, 12:41 AM
now i wondering what will Eric @ SETI say 'bout this...???

davidsnot
01-19-02, 11:19 AM
I may be mistaken, but it looks like benz has either quit dumping, or is getting rejected. No more wu's for a couple of days now. Maybe he's fixing the problem. Hate to lose a heavy hitter.

LandShark
01-19-02, 11:43 AM
well, Benz usually dump his WU every 2-3 days.

and yeah, i hate to see losing a heavy (heaviest, actually) cruncher too. but, that don't mean by using patch client tho...!! if so, our team could be into the TOP in no time if all our members are using patched client, right?!!

davidsnot
01-19-02, 11:46 AM
Right. Ain't no point in doing it if the numbers aren't right.

hallen
01-19-02, 01:46 PM
XP 1500+'s heading under $100. A7M266's $69. at newegg.

Everyone dumps Pentium rigs on eBay and upgrades would make up any loss of benz4140 wu's.

hallen
01-20-02, 01:34 PM
If benz4140 numbers prove legit everone around here is going to feel lousy. I can see that there are equipment configurations capable of hitting those numbers. People already talking about 2 minutes/wu with the current crop of mass market cpu's.

There can be commercial value in becoming the world's leading cruncher so I can see people willing to eat the cost of a RISC system to crunch big.

Also think what it means if benz4140 keeps crunching big number and the rest of you dump the Intel stuff and get going.

JMHO.

Harvey

Wookie8662
01-20-02, 02:59 PM
Heya,

Please remember people, this is all about doing it legit.
It doesnt matter how many WU units a person has , if they are done with a non-legit client.

Its great that Benz can churn out all those massive work packets.
If I could do it, I would too. But, the numbers do not mean a thing, if its not done with a proper client.

If he is doing it with a proper client, then great for him! Keep cranking out those very impressive amounts of WU!

If he is using a "patched" client, then the appropriate steps should be taken.

I will not "pass judgement", until the facts are in.

TC
01-20-02, 03:06 PM
Originally posted by hallen
If benz4140 numbers prove legit everone around here is going to feel lousy. I can see that there are equipment configurations capable of hitting those numbers. People already talking about 2 minutes/wu with the current crop of mass market cpu's.

There can be commercial value in becoming the world's leading cruncher so I can see people willing to eat the cost of a RISC system to crunch big.

Also think what it means if benz4140 keeps crunching big number and the rest of you dump the Intel stuff and get going.

JMHO.

Harvey Harvey - with all due respect, there is absolutely nothing available that can come close to crunching Seti units in 2 minutes. Only a handful of the fastest processors you can buy are able to do them consistently under 3 hours. Less than 2 hours is highly unlikely. The only way to get times that fast is with a patched client, period. A super computer could if the client was complied for such a system, but the Berkeley folks have not done so. If anything is being run on such a system it is not an official client and Berkeley does not condone, nor will they accept results from anything other than their official client. The central issue here is his time per unit, not necessarily how many units he's producing per day. There are plenty of other users out there that produce much more than he does, but they're doing it legitimately with a large number of average computers. The reason we take this seriously is we had this happen before with another member. His computer was either patched or was overclocked so much that it would not operate properly. He was completing units about every 10 minutes. Berkeley examined his returned results and found them to be worthless. He was contacted about the problem and refused to do anything, so they froze his account. They have issued warnings about using patched clients and overclocked machines that are producing corrupt results. All of us enjoy the competition of ranks and best time per unit, but some resort to illigitimate means because they care about nothing but the "score." They're willing to sacrifice the science of the project just to get big numbers. It's not unlike using a patched game to cheat. We do not wish to see our team dishonored by such a person. That is why we're looking into this matter.

zouo
01-20-02, 04:03 PM
Originally posted by TC
Harvey - with all due respect, there is absolutely nothing available that can come close to crunching Seti units in 2 minutes.

Damn. I have Benz's times nailed ;) http://www.mrpibbs.com/misc/quicktimes.jpg

Joking aside, although I think Benz's times are dubious I think it would have been more diplomatic
to have contacted S@H about Benz's times without bringing it to everyone's attention until S@H had given a verdict.

hallen
01-21-02, 08:57 AM
TC- I have to agree with you and withdraw MHO. The economics of dedicating an expensive RISC computer just don't make sense.

I calculate $200. to build a 7.5 wu/day system so 17 of them or $3,400. would replace benz4140's output.

I don't think $3,400 buys much in the Supercomputer store, or even Dell for that matter.

One of these days, something is going to come in out of the blue and knock everyone's number apart. I would love to get my hands on an engineering sample of AMD's Clawhammer myself.

So keep crunching and good luck with the benz4140 situation.

Harvey

Basher
01-21-02, 09:08 AM
Originally posted by hallen
TC- I have to agree with you and withdraw MHO. The economics of dedicating an expensive RISC computer just don't make sense.

I calculate $200. to build a 7.5 wu/day system so 17 of them or $3,400. would replace benz4140's output.

I don't think $3,400 buys much in the Supercomputer store, or even Dell for that matter.

One of these days, something is going to come in out of the blue and knock everyone's number apart. I would love to get my hands on an engineering sample of AMD's Clawhammer myself.

So keep crunching and good luck with the benz4140 situation.

Harvey



Nice job on top rank jumper this week :) Actually a time record by a mile.

hallen
01-21-02, 09:22 AM
Thanks Basher- I've been doing computers a long time and SETI is a wonderful experience. It's the best measurement of hardware and software performance I've encountered.

It's also gives me a idea of the direction our kind is going. Night before last I finished my dumps and took my dog out for our late night (early morning) walk. An Alberta Clipper high pressure had just come through and the sky was brilliant. I could almost talk to the stars.

Keep crunching - you bet.

Harvey

Sir-Epix
01-21-02, 11:34 AM
Originally posted by hallen
Thanks Basher- I've been doing computers a long time and SETI is a wonderful experience. It's the best measurement of hardware and software performance I've encountered.

It's also gives me a idea of the direction our kind is going. Night before last I finished my dumps and took my dog out for our late night (early morning) walk. An Alberta Clipper high pressure had just come through and the sky was brilliant. I could almost talk to the stars.

Keep crunching - you bet.

Harvey

I don't know where in Cleveland you live, but here (north olmsted) we have a lot of light pollution.

Morpheus
01-21-02, 01:28 PM
Originally posted by zouo
Joking aside, although I think Benz's times are dubious I think it would have been more diplomatic to have contacted S@H about Benz's times without bringing it to everyone's attention until S@H had given a verdict.

Yes, that is true... however, when I sent the email off to Berekely, the dubious nature of benz's times were already being discussed in at least 2 threads...

IMHO, rumor has the potential to be damaging to the whole team...

As well, if one reads the content of the email above... the question to SETI is to verify the integrity of the WUS, rather than to verify any type of wrongdoing is taking place... this is a request for verification, rather than an accusation...

Hope that helps...

hallen
01-21-02, 03:24 PM
Originally posted by Sir-Epix


I don't know where in Cleveland you live, but here (north olmsted) we have a lot of light pollution.

Willoughby. We don't usually get that clear either but timed it right.

Maybe Alaska is the answer.

PhoenixMDM
01-21-02, 03:43 PM
Originally posted by hallen


Willoughby. We don't usually get that clear either but timed it right.

Maybe Alaska is the answer.

Willoughby! Willoughby! Next stop is Willoughby!

Sorry, couldn't help myself... Too much Twilight Zone...