Curious question: If no more WUs were issued...

[alanazar]

Newbie has a curious question here

If no more WUs were issued, then how long do you think it will take the current number of systems to finish the backlog of projects?

Put in another way, how many more TFLOPS would be needed to make this collaboration run faster than WUs being produced?

[7im]

I'm sure there is a TFLOPS answer to your question, but the answer would be short lived. FAH staff have said in the past that they have contingency plans in place to scale up the project to handle any additional amount of new computer power added to the project.

If you are asking if it's okay to donate processing time on a super computer, the answer is a solid YES. Pande Group will make more work units to back fill any shortage of work units.

[alanazar]

hehehe thanks for the answer, pretty clear....and i wish i had access to a super computer!

I was considering building my own farm similar to that 51xGPU beast...but my biggest problem today is finding the space for such a rig!

Anyways thanks!

[Zagen30]

If you're going for maximum points, a GPU cluster isn't the best option nowadays. A multi-socket server running bigadv will net you much better PPD at a much lower energy cost. It'd also take up a lot less space than a tower full of GPUs.

[alanazar]

wow thanks for the advice! I'll be very honest i'm not as much interested in racking up points, but more into doing my part in helping to crunch these WUs.

But you raise interesting points i'll look into 1) less space 2) more energy efficiency with the server motherboards...i'll have to do a bit of extra reading now