[QUOTE=Prime95;273540]These extra bits of TF are done manually on a rather ad hoc basis.[/QUOTE]An understatement IMAO.
I would replace "rather" with "extremely":smile:

David

[QUOTE=davieddy;273584]Does CUDALucas employ the "random shift" trick described at the
bottom of "The Math" page?
It is trivial to implement it.

David[/QUOTE]

dunno, I just ordered some books on number theory and expect to be buried there for awhile.

2 things:
A: When I manually get TF assignments, I am now getting 59M from 70 to 71 (which I increase to 72).

B: When I try and set an exponent range to get lower assignments, the page doesn't load/times out. I tried 0-55M, 0-56M, 0-57M, 0-58M, 0-59M, and 0-60M. In each instance the header for mersenne.org loads and nothing else; if I let it set for ten minutes it does the same thing (just the header), but says the page has finished loading. I check my assignments page, and nothing. I also know it's not lack of assignments, because the 0-60M range includes the 59M assignments I get when I don't specify a range. What's up?

Try asking for exponents with a lower bound of 50M or more. I was able to get 400 54Ms this way. Starting with zero may not be ideal depending on the way the Primenet server is coded.

I tried 0 after starting with 50M as the lower bound. I'll try again though.

Nope. Attached a screenshot. The little circle indicating the page was loading was there for about a minute before it went away. No more assignments appeared in my list.

[QUOTE=Dubslow;274760]2 things:
A: When I manually get TF assignments, I am now getting 59M from 70 to 71 (which I increase to 72).
[/QUOTE]
Good on you for going to 72.
I think/hope George intends to make 54M+ available for GPUs to TF
to 72 as soon as expos up to 60M are TFed to 71.
(LLs reaching 60M is ~2 years away).

[QUOTE=garo;274774]Try asking for exponents with a lower bound of 50M or more. I was able to get 400 54Ms this way. Starting with zero may not be ideal depending on the way the Primenet server is coded.[/QUOTE]

Some hard figures for 55M expo:
2.17 GHz days to TF 69 to 70
220 GHz days for 2 LLs.
If 70 bits is worth doing on a CPU, surely 72 bits on a GPU
is a conservative goal.
Note that "P-1 before last bit of TF" is pie in the sky these days:
A. Far too few takers
B. The CPU doing the LL can easily do P-1 where needed.
What it can't do (sensibly) is TF to 72 or more bits.

Finally a radical suggestion:
Since 80% of daily LL assignments are smaller than
54M, and by the nature of being half-chewed have not
been near a GPU in their life, why not make all the newly
rejected LLs available for a couple(*) of extra bits of prompt GPU TF
before assigning them to reliable CPUs for the Coup de Grace?

(*)Eric does 6 or 7 bits on my 45M assignments, Bless Him!

[QUOTE=davieddy;274788]

Some hard figures for 55M expo:
2.17 GHz days to TF 69 to 70
220 GHz days for 2 LLs.
If 70 bits is worth doing on a CPU, surely 72 bits on a GPU
is a conservative goal.
Note that "P-1 before last bit of TF" is pie in the sky these days:
A. Far too few takers
B. The CPU doing the LL can easily do P-1 where needed.
What it can't do (sensibly) is TF to 72 or more bits.
[/QUOTE]

70 bits isn't worth doing on CPU, 26.6 and the math page both list 69 as the max for 53M exponents. Also, there is no guarantee that CPUs doing LL have the memory required for a decent stage 2 P-1. I do, and that's why I might deliberately get P-1 assignments, do that, and then do the LL myself. Otherwise I do agree there's no sense in waiting with the last two bits of TF.

[QUOTE=Dubslow;274834]70 bits isn't worth doing on CPU, 26.6 and the math page both list 69 as the max for 53M exponents.[/QUOTE]

And 70 is the max for exponents >53M, e.g. 55M.

2.17 GHzdays to TF 69-70 has a 1/70 chance of finding a factor.
Say 1/100 if P-1 has been done. Even then it's worth it to try to
save 2 LLs.

BTW P95v26.6 = math page = GHz years forever = George
:smile:

David

Uhh... the math page says
Exponents up to: 58520000 [are] trial factored to 2^69

There is nothing stopping GPUs from taking 54Ms to 72 or 73. Just ask for LL tests and then change the assignment type.

I have started catching expiring LLs and using my GPU to TF them to 71 bits. I was doing 72 but I can't keep up without help.

[QUOTE=davieddy;274788]Good on you for going to 72.
I think/hope George intends to make 54M+ available for GPUs to TF
to 72 as soon as expos up to 60M are TFed to 71.
(LLs reaching 60M is ~2 years away).



Some hard figures for 55M expo:
2.17 GHz days to TF 69 to 70
220 GHz days for 2 LLs.
If 70 bits is worth doing on a CPU, surely 72 bits on a GPU
is a conservative goal.
Note that "P-1 before last bit of TF" is pie in the sky these days:
A. Far too few takers
B. The CPU doing the LL can easily do P-1 where needed.
What it can't do (sensibly) is TF to 72 or more bits.

Finally a radical suggestion:
Since 80% of daily LL assignments are smaller than
54M, and by the nature of being half-chewed have not
been near a GPU in their life, why not make all the newly
rejected LLs available for a couple(*) of extra bits of prompt GPU TF
before assigning them to reliable CPUs for the Coup de Grace?

(*)Eric does 6 or 7 bits on my 45M assignments, Bless Him![/QUOTE]