RIP DCTF. - Page 3

chalsall · 2014-12-24, 17:14

Quote:

Originally Posted by petrw1

My vote it TF to "optimal ranges" in both and use the "escape hatch" when necessary.
That way many/most/all (if we're lucky) will get optimal TF.

OK, I've updated the Estimated Completion report as per VictordeHolland suggestion.

EC jumps from ~200 to ~400 days.

I'll update the LLTF section of the report later today, going to 75 (it's going to be a BIG delta!).

Quote:

Originally Posted by petrw1

BTW I have a premonition there will be a few more GPUs under the Christmas Tree this year.

Ssssshhh!!!! Don't jinx it!

Prime95 · 2014-12-24, 17:20

Quote:

Originally Posted by chalsall

in fact we should be going to 75 for many of the LL ranges.

What do people think? Should we aim to go "all the way" in both the DC and LL ranges, and see how we fair? We could always release at "only" 74 if there was a crunch.

Primenet takes this into account and goes to 75 or higher if that is the most "profitable" assignment available. Hope you get better so that we can work together in converting to a new system.

chalsall · 2014-12-24, 18:03

Quote:

Originally Posted by chalsall

I'll update the LLTF section of the report later today, going to 75 (it's going to be a BIG delta!).

OK, this section of the report has been updated. If we try to go to 75 starting at 60M, the estimated completion for LLTF'ing to 80M jumps from 603 days to 1,965! (Eeeep!)

This means that we (just) couldn't keep up with Primenet's LL'ing, which will probably only take ~1,466 days. However, the 1,965 day estimate is based on LLTF's 30 day rolling average, which has be rather seriously redirected to DCTF'ing; some firepower could certainly be brought back to LLTF'ing if needed.

Thoughts?

chalsall · 2014-12-24, 18:04

Quote:

Originally Posted by Prime95

Hope you get better so that we can work together in converting to a new system.

Getting there. Under new drugs; my doctor says this is one of the worst cases she's ever seen.

Have PM'ed you and Madpoo to start the discussion on implementation.

Mark Rose · 2014-12-24, 18:15

Quote:

Originally Posted by petrw1

PS: Does anyone have personal experience to tell me if a 500W PS is enough for a i5-2500 and a GTX-970 GPU? Neither OC'd. Nothing but GIMPS on it.

That will work perfectly fine as long as the power supply has the power connectors for the GTX-970.

Mark Rose · 2014-12-24, 19:11

Quote:

Originally Posted by chalsall

This actually brings up a thought... It would take a while to implement (it's not hard, but not trivial either), but I could add another pulldown menu in the "Get Assignments" form which let's the requester inform GPU72 what class of GPU they have. E.g.: "GeForce GTX 580", etc al.

Thoughts, comments, alternatives?

I would classify the cards by "compute version" and the AMD equivalent. There's a plethora of models. For instance, would my GT 430's be listed? Probably not. But the two of them are giving me a combined 100 GHz-d/d. It might confuse new people. At the least there should be a "Don't know/don't care" option selected by default.

chalsall · 2014-12-24, 19:37

Quote:

Originally Posted by Mark Rose

I would classify the cards by "compute version" and the AMD equivalent.

Yeah, that was my thinking. Class rather than model in the calculation for the cross-over points.

Quote:

Originally Posted by Mark Rose

At the least there should be a "Don't know/don't care" option selected by default.

Yeah. If the (default) "Don't know/don't care" option is selected the WMS and LGPU72D options will assign work on the "worst case senario" where it is still profitable.

Mark Rose · 2014-12-24, 20:40

Quote:

Originally Posted by chalsall

OK, I've updated the Estimated Completion report as per VictordeHolland suggestion.

EC jumps from ~200 to ~400 days.

I've noticed that estimated_completion and estimated_completion/primenet have differing end goals for 40+M. Also, some of the cell backgrounds have the wrong colour. It's not critical, but I might as well point it out before someone else does ;)

Mark Rose · 2014-12-24, 20:47

Quote:

Originally Posted by chalsall

Getting there. Under new drugs; my doctor says this is one of the worst cases she's ever seen.

Hopefully they work! I'm sorry to hear it's taking a while to get well.

TheMawn · 2014-12-24, 21:03

Quote:

Originally Posted by petrw1

PS: Does anyone have personal experience to tell me if a 500W PS is enough for a i5-2500 and a GTX-970 GPU? Neither OC'd. Nothing but GIMPS on it.

Oh heavens yes. Modern hardware is so energy efficient at stock clocks. Depending on your ambition, you could underclock and undervolt if possible.

I also think we could consider pushing the optimal levels. Releasing at 74 won't be the end of the world either. Same for 71/72/73 down in the DCTF range, which has way more firepower than it needs considering the "rush" to finish it all.

Hopefully you get better soon, Chris.

Xyzzy · 2014-12-24, 22:17

We have too many video cards and have run out of slots to put them in.

http://www.evga.com/Products/Product...02G-P4-3753-KR

This particular card does not require the extra power hookups that most cards need. Although it is labeled as part of the 700 series it uses the same ultra-low-power architecture that the 970 and 980 uses.

If someone wants it we would part with it for $100 plus shipping.

It has the optional metal backplate which is a neat addition.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Getting unneed DCTF work	Mark Rose	GPU to 72	4	2018-01-01 06:14

2014-12-24, 22:17	#33
Xyzzy "Mike" Aug 2002 20037₈ Posts	We have too many video cards and have run out of slots to put them in. http://www.evga.com/Products/Product...02G-P4-3753-KR This particular card does not require the extra power hookups that most cards need. Although it is labeled as part of the 700 series it uses the same ultra-low-power architecture that the 970 and 980 uses. If someone wants it we would part with it for $100 plus shipping. It has the optional metal backplate which is a neat addition.