2 machines working on same unit?!

Two rigs just downloaded the same unit half an hour apart - WTF?



What should I do?

leave them running, its fine. happen periodically but usually pretty rare. they send out multiple copies of the wu's for redundancy and validation of results. you just got lucky and got 2 of the same WU at the same time. you get the points for both WU's no matter what.

Righto - my first thought was this was something wacky from changing machine ID yesterday to escape that bloody dud 6900 that was stalking half the team... but machine IDs are different, and it should not matter anyway on different boxes.

Thanks for the quick reply. I shall relax and watch the ppd roll in then...

may i be worthy said:

I shall relax and watch the ppd roll in then...

Click to expand...

Reminds me of my bowling days...

Just double check in the top of the log files that the UserID number and/or the MachineID number are different.
If they are then it's a duplicate work-unit and you'll get points for folding it.
But if the UserID number + MachineID number are the same then its a copy work-unit and you'll get no points for turning it in.
But that only tends to happen if you clone a drive with folding on and you forget to remove your UserID number from it.

I tend to get only duplicate work units if I download 2 within 30 sec of each other.
Its rare, less than 1 in a 1,000, but it does happen.

Luck ............
[H]

Tigerbiten said:

Just double check in the top of the log files that the UserID number and/or the MachineID number are different.
If they are then it's a duplicate work-unit and you'll get points for folding it.
But if the UserID number + MachineID number are the same then its a copy work-unit and you'll get no points for turning it in.
But that only tends to happen if you clone a drive with folding on and you forget to remove your UserID number from it.
[H]

Click to expand...

Thanks for confirming this. A related topic appeared several months back and when I attempted to explain the distinction people were unsure about the info.

Thanks - I made sure the machine ID is different, although I had read that this was not needed, given that each machine has a unique User ID stored in the Windows registry.

The funny thing is, SR2#3 is a clone of SR2#2, but I deleted the registry entry on day 1 and it has been folding fine with a new one. But the machines with these identical units are SR2#1 ad SR2#2, which are not clones.

But happy to fold whatever Stanford sends me, as long as it is not a mistake!

you are correct. the unique userID is the only thing that matters. the machine ID is a local program thing that isnt part of the information relayed back to F@H.

Both cores on one of my 295's had exact work units yesterday. I didn't bother looking to see if I got points for both. The orange hi-lights are what made me notice it.

Fish

Happens to me once in a while. This is more prevalent on GPUs given that they cycle through several per day.

sirmonkey1985 said:

The Machine ID is a local program thing that isnt part of the information relayed back to F@H.

Click to expand...

Not quite correct.

The UserID sent to Stanford is made up of your UserID plus MachineID.
Which is why you only need to change your MachineID when you get stuck on a bad work-unit/server for Stanford to think it a new machine starting up.

I've max at 4 out of 13 GPU clients folding identical work-units in the past.
Luck ...........
[H]

Tiger is 100% correct.

The UserID is a 16-digit hexadecimal value which Stanford assigns to your client so that the servers can recognize your unique machine. It is unrelated to the UserName (which you assign yourself and) which is used to collect the points into an account. The CPUID (UserID + MachineID) identifying your client can be modified by setting the MachineID.

If the FAH client can not report a valid CPUID value then the FAH Assignment Server will give it an UserID value and this UserID value will get stored to the donator machine. But UserID itself is not a whole CPUID. To get the CPUID it has to add UserID and MachineID together. Now it has the proper CPUID to provide to the FAH servers.

MachineID identifies FAH client instances per computer. The donator MUST change the MachineID value if there is a need to run multiple copies of FAH client on THAT computer or the FAH servers will treat those multiple FAH clients as one and will assign the same WUs to all those clients.

Note: The same result (different CPUID value) can be achieved by changing UserID value, but that is not something the donator itself should ever do.

Finally, the FAH stats system is actually counting different CPUIDs, not active CPUs itself.