mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   GPU to 72 (https://www.mersenneforum.org/forumdisplay.php?f=95)
-   -   RIP DCTF. (https://www.mersenneforum.org/showthread.php?t=19897)

chalsall 2015-01-19 18:37

[QUOTE=petrw1;392861]Is LL getting close to not keeping up?[/QUOTE]

To optimal levels? Yes.

[QUOTE=petrw1;392861]Just to be clear on "breadth first" doesn't make sense. Are you implying it is better to:
1. Take all DC to 69, then 70 then 71 then 72
2. Take each lowest DC all the way to 72 ... this is where I currently am
3. Stop wasting time on DC and do LL instead[/QUOTE]

Option 3 first. Option 2 second. Option 1 if it amuses you.

Mark Rose 2015-01-19 19:10

[QUOTE=chalsall;392849]Further, it should be noted that doing the "breadth first" search many are doing may be fun, but it doesn't really make all the much sense.[/QUOTE]

Neither does having a 680 watt space heater under my desk consuming 1.8 gigajoules per month dividing numbers. ;)

That's about 25 kg of CO[sub]2[/sub] per month into in the air based on the local energy mix.

petrw1 2015-01-19 19:12

[QUOTE=chalsall;392870]To optimal levels? Yes.[/QUOTE]

So when I see about 45,000 Cat 3 (Rounding 65-70M)
and about 25,000 Cat 4 (70-75M)

.... are these NOT all to optimal level; in other words were they released at a lower bit level but with available resources should be deeper?

Never mind, I can look it up ... VERY FEW exponents in the 6x - 7xM range are factored to the optimal 75 bits.

[url]http://www.mersenne.info/trial_factored_tabular_data/2/60000000/[/url]
[url]http://www.mersenne.info/trial_factored_tabular_data/2/70000000/[/url]

petrw1 2015-01-19 19:28

[QUOTE=chalsall;392870]
Option 3 first. [/QUOTE]

I could move my 500GhzDays/Day there.

While I'd like to see DCTF die I'll go where most needed.

chalsall 2015-01-19 19:35

[QUOTE=petrw1;392883]While I'd like to see DCTF die I'll go where most needed.[/QUOTE]

LLTF is most needed.

Mark Rose 2015-01-19 19:42

Y'all guilted me into option 2.

LaurV 2015-01-20 10:54

[QUOTE=Xyzzy;392863]What is the optimal LLTF strategy?[/QUOTE]
1. Buy 1000 GPUs (possibly Titans, or newer, possible in boxes with dual Xeonx...)
2. Install misfit
3. Profit

retina 2015-01-20 11:10

[QUOTE=LaurV;392937]1. Buy 1000 GPUs (possibly Titans, or newer, possible in boxes with dual Xeonx...)
2. Install misfit
3. Profit[/QUOTE]I think you left out a step:

1.5 Install a new 4000A 3-phase incoming line

axn 2015-01-20 11:17

[QUOTE=LaurV;392937]possibly Titans[/QUOTE]

A Titan costs $1000, burns 250W and produces 400 GHz-day/day

A 980 costs $600, burns 165W, and produces 520 GHz-day/day

A R9 290X costs $400, burns 250W, and produces 600 GHz-day/day


Don't buy Titans for TF.

petrw1 2015-01-20 15:39

[QUOTE=axn;392941]A Titan costs $1000, burns 250W and produces 400 GHz-day/day

A 980 costs $600, burns 165W, and produces 520 GHz-day/day

A R9 290X costs $400, burns 250W, and produces 600 GHz-day/day


Don't buy Titans for TF.[/QUOTE]

My 970 was $450 (Canadian even) ... not sure what it burns ... produces 515 /day if I keep the room cold. :) It heat throttles itself.

ET_ 2015-01-20 15:52

[QUOTE=petrw1;392955]My 970 was $450 (Canadian even) ... not sure what it burns ... produces 515 /day if I keep the room cold. :) It heat throttles itself.[/QUOTE]

a cold room in Canada means -20°C... :razz:


All times are UTC. The time now is 08:00.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.