Dpc Responds

[mad mikee]

Just some balancing stuff and NO I AM NOT defending our future dinner

1. I don't know about Intel stuff, but ALL my nodes have docking times under 9 mins and a couple are around 8:05-8:10 so don't think that is a good criteria.

2. Various people here have used the technique of creating an install that includes a node.prp so that they aren't tied to requirement of having the registration server up (since it is ALWAYS available )

3. During My conversation w/ Charles, we were discussing about uploading multiple resultsQueue.dat files from a single machine, and while I thought I remember that it somehow 'knows' the node where the file was created, he remembered it as being attached @ Upload time. Even If I'm right on this, 2) would explain them all coming from one node.

Just had to point these particulars out, not speaking on any other practices ATM, good or bad.

[RocKer]

What in the hell is here going on,am a few days away from xs-forum,open the xs-link on desktop to see this tread here(and others),my god ,and the stats are all over the place,and there are some people missing,new one's to,i only read in this tread did'nt read any other forum's,yet,but i will allso take a look at dpc forum.

So at this moment i don't have any idea what really is happend(need to read),yes somebody is cheating,but who or what sub-team

I know one thing,i dont cheat,i find that if somebody cheats,the only lie against themselfs.

If the wont pics,the just have to ask,simpel as that,i have 5 rigs running at the moment,1 x DX1.6ghz @ 3.2ghz,1x DX2.0ghz @ 2.4ghz,1 x A643200+ @ 2.4ghz,1 x XPM2600+ @ 2.5ghz and 1 x P2 200mhz,there will come 2 more,1 x 2500+Barton @ 2.3ghz and 1 x DX1.6ghz @ 3.2ghz.

[Movieman]

Bloody_Sorcerer: In your particular situation with ghosting machines weekly I can understand your creating the shortcuts or you'd go mad getting it all done.
Back to the cows:
Ok, so now we should beleive that the entire output of the DPC_ICA_COWS for the last 7 days(690,500 cands) came from one location in one huge business or school, or factory that was controlled by one person who'd imaged the app so as to save the 5 mins tops of the install because there were so many machines to put it on.
If that were true, I'd say ok, makes sense, the guy needs to make this into assembly line work just as Bloody_Sorcerer has to. Pure common sense in those scenarios.
But this is different, this is an entire team putting out an average of 98,643 a day over one weeks timeframe, No one on the whole team had a different node? No one on this team ever had to do a format and grab one off the D2OL page? ever? from anyone? This node has upped as of this am 1,833,306 cands. This is one amazing node, it goes to every machine in the group and no one ever had the need to use another. Tell me that this sub team is one guy with 300PC's in a huge business and I'll apologise, anything short of that and I find it very difficult to beleive that any group, separated by any distance with all the factors that effect PC's from harddrive failure to electrical surges to virus's and no one ever needed another node? C'mon guys, do we all look like we just got off the turnip truck yesterday?
Can I prove it :NO! But if it looks like, tastes like and smells like, it probably is.
and to end my point, we've yet to hear from the Cows anything definitive to explain it.
One last point that is very relevant: By using this one node are you not defeating the purpose of the project? Let me play devils advocate: Isn't the point to explore all the possibilities and by using this one node you are hindering that? This point I may well be wrong about as my understanding of the particular node usage isn't total and if I'm wrong on this I apologise on this point.

[[XC] Duc]

The answers from DPC:

"They are not cheating" (which is indeed possible)

"We have no right to accuse them" ( not even if we point out what we are suspicious about... freedom of speech seems to be irrelevant)

"Until we can actually prove they are cheating they refuse to use another method to buffer there candidats... it works for them, so why change it" (even if they would get the same results with the non-cheating method)

"They are reluctant to answer any questions about their computers or output" (because people could lose jobs over it and it's none of our bussiness how and where they get the results)

"We should talk to the project leaders (Charles) if we think there's something wrong and they (Charles) have to prove what's wrong by showing serverstats, uploadstats or whatever... IF there's something wrong with the results they (Charles) should prevent people from using the 'result.dat-copy-method' by changing the program, method to up cands or whatever." (until then, they will NOT stop using this method which is know to be 'the' method where cheating is possible, even if you can buffer in a non-cheating way)

This is a personal translation of the opinion I can detect in the answers I get... (just stating the "personal" part before CowArt says I'm not translating exact sentences )

[Entity_Razer]

what I must add

690050 canidates, at an average of 10minutes per canidate (which is short, very short in my opinion


thats

4 792,01 DAYS of docking needed to do those. Now take into account that some anthrax's or others can take up to 15, if not 20 minuts. But lets stick to 10 minuts per canidate, thats still 4 792 DAYS of docking needed

Now you please tell me how many computers they have to do THAT in say..... 3 months time, lets take 3 months time, = 90 days

I find 53 computers, something that is NOT that hard to find, 53 computers, but now take into account that we took a 3 month period for this , I do not know for sure how many members ICA counts, maybe someone of the team itsself would like to enlighten me on that, or even how many comps they count.

But a normal office computer is about a P4 2.5Ghz with its hyperthreading disabled thanks to a cheap motherboard by dell or whatever.

Now also, a lot of canidates require a LOT more then 10 minutes to complete.

Take that into account and you're beginning to see my point. Not to mention you can only start counting from 12 09 2005 since that was you latest big flush so lets say THE ENTIRE team went into buffering then? Its possible but to MANUALLY copy all those result files and copy them back.... I have to admit it seems HIGHLY unlikly, if it was just nodes being added, hey, thats VERY doable but how things are done now.... you'd need a small army to upload on all those comps

[lv_dicedealer]

Here is a thread over at FDC that I responded to... In it are some remarks from a DPC member.

LINK

[[XC] moddolicous]

Lost prophet posted this:

ICA has 0 CANDS LEFT (source DPC forum, they say it theirselves so I bet they won't try to dump again), if we all stayed in XS team we could have made it to nr1 before new year. Guess we lose

If we can come back as a team, #1 can be ours! Lets go, come one guys. BTW, whose xs_payback?

[[XC]thewildblue]

Thats hardass from XR I think.

If you check the thread in the place not to be said I said the same thing about us rebanding......

[STEvil]

even my "best candidate'" is newer..

You guys (the cows) dont have to list your work or post incriminating pictures, but say listing the machines exactly as DDTUNG did and maybe saying the machines are @work in (country) would help.

Anyways, I hope this gets resolved.

edit

fixed typo

[DDTUNG]

Just a few more comments.

One node vs multiple nodes. Besides ease of installation, using one node makes it easy to scroll down the list of nodes when you are checking node stats in the D2OL stats pages. One obvious disadvantage is that when you want to check on the status of any one particular machine, you have to physically access it, or use remote access or monitoring software which consumes resources. And as has been pointed out, it can invite abuse.

Online uploading vs manual uploading. Manual uploading is time consuming. It is also dangerous in that a set of results can become corrupted due to instability or accidents such a power failure. I know because in December I lost close to 5000 cands when I was unable to upload due to the NAS outage, and at the same time our building had its power cut for minor repairs. Our team members prefer to upload every 6 hours to an 'offshore' node and add the node when desired.

Docking times. These do matter because it is common knowledge that AMD rigs have lower docking times than Intel rigs, and the average docking times for two identical machines should be comparable. Which brings me to my next point.

We have asked for a list of the machines that ICA_COWS use because from that list it is possible to estimate their normal output. It is also possible to estimate the correct average docking time. We can understand your difficulty about the possibility of people losing jobs, that's why we are only asking for brief specs, not pictures or locations. That is also why we always advise our members to ask for permission to run D2OL on rigs that they don't own. How difficult can that be.

In life we often come across situations where things don't seem to add up. In this case, to me at least, too many things didn't add up. Some of these things, such as the string of submissions in multiples of 10, are almost statistically mind boggling.

It is perfectly understandable that you take offense to these suggestions of cheating, and the resultant scrutiny. I myself went through that with F@H some 2 years ago, and ended up quitting the project after giving my side of the story. But the innocent should have nothing to hide, and the guilty will be caught and punished in the end.

DDTUNG

[DDTUNG]

Today being a holiday I had alot of time to sit down, gather some facts and figures, and do some calculations.

FACTS: ICA_COWS have -an average docking time of 12 minutes
-dumped roughly 700,000 cands in the last 7 days
-last dumped on Oct 11, 2005, so buffered 75 days

Average docking time of 12 minutes means the average machine they used produced 120 cands per day. 700,000 cands in 75 days gives an average of approximately 9,000 cands per day. So they used on average 75 machines on a given day. Each of those 75 machines would have completed 9000 cands, or 4.5 sets of the 2,000 cands buffers they claimed they uploaded last week.
That means they had around 350 sets of these 2,000 cands buffers to upload last week, which they did, manually. From my experience with manual uploads, it takes about 20 minutes to manually upload 2,000 cands from 1 resultQueue.dat. Using one machine around the clock, that would take around 5 days. But why would anyone with 75 machines use just one of them to upload? I wouldn't torture myself that way. So let's say they used 5 machines to upload these cands, and it would have taken 1 day round the clock, or around 3.5 hours per day spread over 7 days. But the actual upload periods per day were much longer than that. So one could say they tried to keep XS guessing by slowing the upload rate by using fewer machines, meaning they would have to work longer hours uploading. Fine. But there were at least a couple of stretches of uploads that took place throughout the night, so they must be better men(sorry I mean cows ) than I am by enduring these all-nighters.

As I said, it all seems like simple arithmetic, but somehow doesn't add up.

DDTUNG

[Movieman]

As I said, it all seems like simple arithmetic, but somehow doesn't add up.

DDTUNG
--------------------------------------------------------------------------
After pulling an all nighter trying to find out what the heck was going on here Thursday I finally passed out about 1pm yesterday. This kept bothering me.
Was I accusing some totally innocent group of guys from across the Atlantic that I've never met?
It's possible, as much as I hate to admit it, it is possible that they did this legitimately. But to do so the resources and time involved is absolutely huge!
So lets stop and say to ourselves that they are as pure as the driven snow and I will prostrate myself towards the east and ask forgiveness for all I've said. I'd like something in return that will stop this from ever happenning again. It's a very simple thing and keeps EVERYONE honest beyond reproach.
NO duplicating nodes. You want to carry the installer on a cd to save the time of downloading it, fine. BUT install it, register it and get on to the next one. If this means that because of time constraints you can only install it on 6 machines in a day instad of 26 well thats a problem that you'll have to live with.Life just isn't perfect but to continue this as it is now is plain and utter BS!
I have no problem losing to someone with better resources than i do, but I refuse to play in a game where by doing things in a "certain" way you have the ability to cheat and the others involved are not on a level playing field.
Before this incident started, I wrote a post where I said " When this is all over whether we win or lose I think we should congratulate the COWS on a superb effort."
Another issue that bothers me is the lack of wanting to disclose what you are running for equipment. WHY? How could that harm you in any way?
No one is asking you to say that John Jones at IBM has the app on 300 mainframes, just what they are, not what business they are attached to.
What is the problem with disclosing this info?
The ball is now in your court, lets do this fair and square or lets not do it at all.
If you reply in the negative, you might as well admit what most of us already beleive. If you reply in the positive, tomorrow will start as a new day and I can go back to thinking of more important things like how to pay my electric bill this month with 2450 watt of PS running 24/7
Happy New Year to all!

[Entity_Razer]

ARKONXX SPEAKS (ICA COWS)
** This is a translation from the DPC forums. For the DPC reading this, I will translate as unbiased as possible, if you have problems with my translations, I suggest you translate it yourself and PM it to me so I can edit this post or you just post it yourself**


Dear DPC D2OL members,

Due to the big disturbance that has arrised we will try to shed some light on our operations/case.

We started of with 150 computers, the majority being P4 3ghz. Over take we where able to expand to rougly 750 PC's. Off course these computers are not all P4 3Ghz machines, we do not want to exclude the slower computers. We think every bit count dont you agree?

The timeframe?
Everyone can see in what timeframe this took place and posting a picture is something we personally find a bit stupid as they are not all in one room.

Previusly Team_Gaol was accused by XS of cheating. Was it proven that they where guilty of such a thing? NO. We find that XS judges people way to fast. Even the slightest resistance from being overtaken and they say it's cheating. How can THEY produce that kind of output daily? Starting 1-1-2005 up to 29-12-2005 they went from About 1 Mil to 15 mil. A 14 MIL. difference. Do we pose question at that? No. Then why do they start asking question when we post about 700 000

We have always seen XS as a threat but to keep ahead of them till the end of the year was a secundary goal. We have as a goal to make a big as possible flush in one go. en to become nr 1 as DPC. But seeing recent development we let these targets go away and we found that keeping XS from the nr1 spot was best suited as our primary target. I warned people for XS in previus post already that they where comming.

Again, we are not cheaters! Wij did our best for the good of the D2OL project. Everyone withing DPC knew of what we where doing, that we where buffering. We hope we still have the trust of the DPC memvers en that Xs will leave it at this. Xs, or Free DC right now, will eventually overtake us. Thats a fact, they have more active people at the project with bigger capacities. I think we where able to prevent that DPC closes 2005 as nr 2 and that we have completed our objective. Again, to all a happy newyear!

Members of DPC_ICA_COWS

[SoF]

time for us to go back to team xs or what do you think?

[Entity_Razer]

To respond to DPC:

Please question our output. We have about 460 members here on D2OL, a top producer who has 32 dual Xeons and a lot of other computers, jeff has about... what was it in his top days? about....10 dual xeons. Windforce has a lot of P4's at his disposale, Conrad had a lot of comps at his command as a system operator at the office before he got promoted), .....

hell LVDICEDEALER once had a Quad Itanium 2 ....

Please do question our output if you feel that will do anything. fact is we just have a lot of active members.

[Movieman]

ARKONXX SPEAKS (ICA COWS)
** This is a translation from the DPC forums. For the DPC reading this, I will translate as unbiased as possible, if you have problems with my translations, I suggest you translate it yourself and PM it to me so I can edit this post or you just post it yourself**


Dear DPC D2OL members,

Due to the big disturbance that has arrised we will try to shed some light on our operations/case.

We started of with 150 computers, the majority being P4 3ghz. Over take we where able to expand to rougly 750 PC's. Off course these computers are not all P4 3Ghz machines, we do not want to exclude the slower computers. We think every bit count dont you agree?

The timeframe?
Everyone can see in what timeframe this took place and posting a picture is something we personally find a bit stupid as they are not all in one room.

Previusly Team_Gaol was accused by XS of cheating. Was it proven that they where guilty of such a thing? NO. We find that XS judges people way to fast. Even the slightest resistance from being overtaken and they say it's cheating. How can THEY produce that kind of output daily? Starting 1-1-2005 up to 29-12-2005 they went from About 1 Mil to 15 mil. A 14 MIL. difference. Do we pose question at that? No. Then why do they start asking question when we post about 700 000

We have always seen XS as a threat but to keep ahead of them till the end of the year was a secundary goal. We have as a goal to make a big as possible flush in one go. en to become nr 1 as DPC. But seeing recent development we let these targets go away and we found that keeping XS from the nr1 spot was best suited as our primary target. I warned people for XS in previus post already that they where comming.

Again, we are not cheaters! Wij did our best for the good of the D2OL project. Everyone withing DPC knew of what we where doing, that we where buffering. We hope we still have the trust of the DPC memvers en that Xs will leave it at this. Xs, or Free DC right now, will eventually overtake us. Thats a fact, they have more active people at the project with bigger capacities. I think we where able to prevent that DPC closes 2005 as nr 2 and that we have completed our objective. Again, to all a happy newyear!

Members of DPC_ICA_COWS

Interesting reading, answer the questions that you choose to and ignore the ones you don't want to answer.
I still have the question to you sir of why and how all of your output(90+%) for the last week was on one node? This has been asked more than once with no reply. On at least 1 day your entire output was on that one node.
Do you expect me to beleive that no one on your team uploaded ANY cands that day besides that one node? Do you control the uploading of every menber with such an iron hand that they do not allow their PC's to send the finished cands without your approval.
Am I the only one who went to the D2OL web page and actually followed the instructions and installed it that way?
I have 3 dual xeons, 2 with HT, so 4 instances each on the 2 machines with HT and 2 instances on the machine without HT, a total of 10 instances and thats what I have, 10 nodes.
I'd also asked about a one node per install usage and that you've also ignored. The odd's of all those cands being done legitimately on that one node are worse than the lottery!
To me, it's sort of cut and dried, once again, the numbers speak for themselves.

[[XC] moddolicous]

To respond to DPC:

Please question our output. We have about 460 members here on D2OL, a top producer who has 32 dual Xeons and a lot of other computers, jeff has about... what was it in his top days? about....10 dual xeons. Windforce has a lot of P4's at his disposale, Conrad had a lot of comps at his command as a system operator at the office before he got promoted), .....

hell LVDICEDEALER once had a Quad Itanium 2 ....

Please do question our output if you feel that will do anything. fact is we just have a lot of active members.

That was FreeCableGuy with the itanium I THINK. I know he's getting a dual opty rig (2x285's). As for the ICA's output thing, I have one question: If you could have gotten in trouble installing on all those comps, how did u have time to install them without being caught? I can think of only 1 quick way of doing it, but I'll see if you figure it out first.

[Soepkip]

Well hello, i'm from DPC too ^^

Anyways, the same node explanation. I'm just thinking, but what if they put a client on an USB stick and just ran off all the computers?

And for the way they bufferd, maybe they had to flush it all from their homeadress, because the 750 computers they had couldn't connect to D2OL thanks to a firewall. Maybe that is a reason only 1 node was used for flushing?

[[XC] moddolicous]

You mean buy like a 64mb ram stick and leave it in the comp? Or do you mean copied that D2ol folder to each comp from the same usb stick? That firewall thing wouldnt make sense because each dump ended in a zero (except for the last 2 or so).

[RoyalFlusher]

Yes, you just have to install D2OL once on a computer. You also register it. You edit your prefs. You make a copy of the D2OL installation on a CDRom or a USB-stick. You copy the map on the systems you wish to 'infect' and make a shortcut in your startupmap. You manually update the buffers and take it after that offline. With the right parameter controller:file you let D2OL do invisible its work. As simple as that it is!

[[XC] moddolicous]

Alright, that was the response I was waiting for. Thats the only effient way to use 1 node. But doesnt coping that same work (from the original folder) just give the same work to the same comps?? But since one of you said that they cant connect through the firewall, how are they getting access to the results quene file? If you can explain how they get theresultsquene to their home comp, then maybe this whole thing can be resolved.

[RoyalFlusher]

Soepkip was 'supposing' and there were some flaws in his reasoning.

I suppose they managed it on the way I described.
After registering the client they stopped the client BEFORE it downloaded its workbuffer. Then they copied it. Once copied on the computers where they mirrored it, they gave it access to the internet to download the workbuffer. After they ran out of work on those machines they copied the resulQueue.dat files to USB-Stick and deleted the original ones. Then they dowloaded again a workbuffer. And so on...

[DDTUNG]

A P4 3.0 without HT crunches an average of 95 cands per day, or 4 cands per hour, giving average docking time of 15 minutes. A P4 3.0 with HT can do up to 150 cands per day running two instances of the client, giving 75 cands per day per client, or 3 cands per hour per client, resulting in average docking time of 20 minutes. Since most of your 150 machines were P4 3.0, you should have average docking time of between 15 and 20 minutes, depending on what proportion had HT capabilities. Since you said that the 600 other machines you used for the finish were mostly slower, you would have an even higher average docking time. But the D2OL stats page show your average docking time as 12 mins 03 secs. How is that possible?

With 150 P4 3.0 rigs your daily production should be between 14,000 and 22,500 cands, again depending on the proportion with HT. Since your last dump was 75 days before Dec. 25, how is it that you only had 700,000 cands instead of 1,050,000 to 3,375,000 cands, without even including the other 600 machines that you used later?

And how do you explain that with so many machines of varying speeds, that you were able to have dumps that ended with 0 not only on consecutive days, but every hour, and actually every 15 minutes? Did the slower machines all finish the 2,000 cands?

I will stop here for now because I think there is enough for everyone to form a conclusion.


DDTUNG

[Bloody_Sorcerer]

docking time is definitely not a decent way to judge anything; i have an average docking time of 7 minutes 58 seconds, but only average 26 candies a day. Yesterday, my machine was on all day doing pretty much nothing but crunching (a bit of background use, such as surfing the net, and a few VMs running) but I only did 59 candies, which would imply an average docking time of 25 minutes. not all is as it appears.

[shmaa]

A P4 3.0 without HT crunches an average of 95 cands per day, or 4 cands per hour, giving average docking time of 15 minutes. A P4 3.0 with HT can do up to 150 cands per day running two instances of the client, giving 75 cands per day per client, or 3 cands per hour per client, resulting in average docking time of 20 minutes. Since most of your 150 machines were P4 3.0, you should have average docking time of between 15 and 20 minutes, depending on what proportion had HT capabilities. Since you said that the 600 other machines you used for the finish were mostly slower, you would have an even higher average docking time. But the D2OL stats page show your average docking time as 12 mins 03 secs. How is that possible?

With 150 P4 3.0 rigs your daily production should be between 14,000 and 22,500 cands, again depending on the proportion with HT. Since your last dump was 75 days before Dec. 25, how is it that you only had 700,000 cands instead of 1,050,000 to 3,375,000 cands, without even including the other 600 machines that you used later?

And how do you explain that with so many machines of varying speeds, that you were able to have dumps that ended with 0 not only on consecutive days, but every hour, and actually every 15 minutes? Did the slower machines all finish the 2,000 cands?

I will stop here for now because I think there is enough for everyone to form a conclusion.


DDTUNG

Wow.. nicely played, DD.