Anyone using GPUs to do DC, LL or P-1 work?

[sonjohan]

I can switch on Sunday evening from LL-TF to DC-TF.
I don't have access to the GPU-laptop right now.

[Mark Rose]

I am now throwing about 300 GHz-day/day at DC ;)

[kracker]

Originally Posted by LaurV

@kracker Man, this is priceless, hehe, thanks a billion, I don't know if you intended so, or you just forgot to change the user name for one of your machines when you switched all the cores from P-1 (for me) to DC (for you), but thanks a lot anyhow ... now honestly... you don't need to do DC for me right now I do this with Chris only, hehe

CPU Name    Exponent    Result Type    Received   age days    Result         GHz-Days
k-sigh       31324361          C      2014-01-29 13:15   20.5     5AFE83EF49C660__   37.0672 

Oops.

[chalsall]

Originally Posted by Mark Rose

I am now throwing about 300 GHz-day/day at DC ;)

Coolness, thanks.

And so everyone knows, observing the assignments coming out of Primenet for "slow and/or unknown machines" today, it appears that George has set the offset to be 34290000 or so.

So, for those who want to help build up a bit more of a buffer for this class of assignments, please set this in the "Low" field on the DC-TF assignment form.

We've got a good buffer above this already, but it would be nice to give these machines the lowest possible candidates to work on. And, also, once George has implemented the new recycling rules, there will be a brief surge for such assignments.

Thanks everyone.

[Mark Rose]

Originally Posted by chalsall

Coolness, thanks.

And so everyone knows, observing the assignments coming out of Primenet for "slow and/or unknown machines" today, it appears that George has set the offset to be 34290000 or so.

So, for those who want to help build up a bit more of a buffer for this class of assignments, please set this in the "Low" field on the DC-TF assignment form.

We've got a good buffer above this already, but it would be nice to give these machines the lowest possible candidates to work on. And, also, once George has implemented the new recycling rules, there will be a brief surge for such assignments.

Thanks everyone.

34290000 is conveniently above the 1608 DC assignments I have outstanding. I really shouldn't reserve giant batches hah.

[Uncwilly]

Originally Posted by chalsall

So, for those who want to help build up a bit more of a buffer for this class of assignments, please set this in the "Low" field on the DC-TF assignment form.

Do you think it is worth me pulling some of my CPU's off TF in the 332M range temporarily?

[chalsall]

Originally Posted by Uncwilly

Do you think it is worth me pulling some of my CPU's off TF in the 332M range temporarily?

Definitely not. Or, at least, not for TFing. IMO, a CPU should never be doing TFing any longer.

It /would/ be worth you finding someone to trade your CPU work for someone else's GPU TF'ing in the 332M range, on the other hand....

[chalsall]

Originally Posted by Mark Rose

34290000 is conveniently above the 1608 DC assignments I have outstanding. I really shouldn't reserve giant batches hah.

Not a problem. We only learnt today (by observation) what George had decided to be the transition point.

Your work will still be valuable and useful.

[tigreroars]

Just a question.

I've got a few cores open on a Core i5-760. Right now they're being used for P-1 pulled one at a time directly through the client and given 8gb of memory. Is there a better use for it? Should I switch it back to DC LL?

The GPU is currently doing TF in the DC range for half the day give or take.

Thanks in advance.

[garo]

Originally Posted by tigreroars

Just a question.

I've got a few cores open on a Core i5-760. Right now they're being used for P-1 pulled one at a time directly through the client and given 8gb of memory. Is there a better use for it? Should I switch it back to DC LL?

I think P-1 is fine for a CPU with so much memory. You could switch to DC LL if you wish as we have lots of P-1 firepower (for the moment).

[TheMawn]

Originally Posted by tigreroars

I've got a few cores open on a Core i5-760. Right now they're being used for P-1 pulled one at a time directly through the client and given 8gb of memory. Is there a better use for it? Should I switch it back to DC LL?

Have you got 8 GB total or 8 GB per core? If you have 8 GB total, you can leave one doing P-1 with as much memory as you feel comfortable with and put the rest to DC or LL. If it's 32 GB total, then I really don't know. You get diminishing returns as you commit more memory to P-1 so you could do a mix of LL/DC and P-1 as you desire.

Both work types are useful, though we do currently have a surplus of P-1...