[QUOTE=Dubslow;304821]Don't release it manually. GPU 2 72 can only pick up exponents that are expired at midnight by PrimeNet's old assignment cleanup program (AFAIK).[/QUOTE]

Wrong.

[QUOTE=chalsall;304835]Unfortunately Dubslow spoke without fully understanding how things work...

GPU72 hasn't kept this assignment because I don't like having too many held at any one time. Since PrimeNet won't be handing out 58Ms for LL testing for quite some time (unless you explicitly ask for one via the manual assignment page) we have lots of time to bring them back in to take to 73 bits in the near future.[/QUOTE]

That is reasonable. Dubslow's statement, while describing what I would consider a beneficial methodology, didn't match observed behavior.

[QUOTE=chalsall;304835]Unfortunately Dubslow spoke without fully understanding how things work...

GPU72 hasn't kept this assignment because I don't like having too many held at any one time. Since PrimeNet won't be handing out 58Ms for LL testing for quite some time (unless you explicitly ask for one via the manual assignment page) we have lots of time to bring them back in to take to 73 bits in the near future.[/QUOTE]

[QUOTE=chalsall;304836]Wrong.[/QUOTE]

[QUOTE=sdbardwick;304840]That is reasonable. Dubslow's statement, while describing what I would consider a beneficial methodology, didn't match observed behavior.[/QUOTE]

:max:

:razz:



[SUP][SUP][SUP][SUP][SUP][SUP]Is it okay to spam the spam thread? :P[/SUP][/SUP][/SUP][/SUP][/SUP][/SUP]

Whose thread is this anyway?

Fermi

PS You might call it spam.
I call it "posts which pass over Chris's head".

[QUOTE=davieddy;304813]Most of them <72,

Why don't you focus on it then?

D

By "focus" I mean 59M to 60M to 73 in one go. (200 per day).
NOT 45M to 70+M one bit at a time.
Creaming off the low-hanging factors merely diminishes the
incentive to do a proper TF job before an exponent is EVER assigned
to a CPU for LL (P-1 if needed).

Forget the spilt milk.[/QUOTE]

I've been doing what you say for months now. Hurrying as fast as possible.

[QUOTE=Chuck;304857]I've been doing what you say for months now. Hurrying as fast as possible.[/QUOTE]
Great stuff! How many a day?
The sensible target is 200 (several GPUs obviously).

Keep up the good work!
David

Contentious?
 

MOI ???

Many (most?) teachers dread awkward questions from clever pupils.

I love them.
95% of the the time I answer them and further enlighten the enquirer.
5% of the time I learn something new!

Not unlike here:smile:

D

[QUOTE=davieddy;304871]Great stuff! How many a day?
The sensible target is 200 (several GPUs obviously).

Keep up the good work!
David[/QUOTE]

Depending on the lower limit, about 8 per day.

[QUOTE=Chuck;304887]Depending on the lower limit, about 8 per day.[/QUOTE]
THX
A concrete case study will help us all!

One GPU? What type?

My core2 2GHz does an unsurprizing 2 GHz days/day on LL.
~60 days per test.
Your effort alone is keeping 480 similar LL workers supplied
with properly TFed expos.

Re your "lower limit" comment, even if no TF had been done,
the time taken would be at most double that for 72 to 73,
and you would be rewarded by finding loads of factors.
As there is a systematic creaming off of long-hanging factors afoot,
your reward is the satisfaction of making a most valuable contribution
to the GIMPS project, and if any of your expos should be prime,
you should receive an acknowledgement!

All I am looking for is 25 others like you to reach 200/day.
George? Chris? GET IT TOGETHER!

D

[QUOTE=davieddy;304917]As there is a systematic creaming off of long-hanging factors afoot, your reward is the satisfaction of making a most valuable contribution to the GIMPS project, and if any of your expos should be prime, you should receive an acknowledgement![/QUOTE]

You are forgetting human nature David.

The truth is most people who are doing TFing are doing so in order to find the most factors possible. Until I started setting a lower limit to the pledge level, many (not all, but many) were only pledging one additional bit level. This is why we've ended up generally doing a "breadth first" search.

As I've said before, there are exceptions. Chuck (and bcp19/Pete in the DC range) is an exceptional exception -- he volunteered to do the less "rewarding" final step from very early on in the project. He's the reason there's an option for "Highest TF level" on the assignment page.

[QUOTE=davieddy;304917]All I am looking for is 25 others like you to reach 200/day.[/QUOTE]

Keep an eye on the [URL="http://www.gpu72.com/reports/overall/graph/month/"]Trial Factoring Depth per Day in the LL Range[/URL] graph.

[QUOTE=davieddy;304917]George? Chris? GET IT TOGETHER![/QUOTE]

We're now ahead of the LL wave (oh, and more than 500 days ahead of the DC wave). We've managed to have most (if not all) assignments from PrimeNet appropriately TFed and P-1'ed for several months now.

I would argue that we [B][I][U]do[/U][/I][/B] have it together.

[QUOTE=chalsall;304921]You are forgetting human nature David.
[/QUOTE]
I blame it on senility