Assignment Strategy

[Dubslow]

http://filesmelt.com/dl/screenshot321.png
Still trying to load as I type. (Note at bottom: Waiting for gpu.mersenne.info...)

Originally Posted by chalsall

Actually, that's only half true...

DC TF: 6.401 GHz Days Work Done, 24.797 GHz Days Work Saved.
LL TF: 12.675 GHz Days Work Done, 182.163 GHz Days Work Saved.

(Average per factor found. GHz Days Work Done is "Normalized".)

Either way, the Saved/Done ratio is still more than twice as big for LL as opposed to DC.

And on mersenne.info I get (at the top of course)

<title>W/elcome!!! :: GIMPS Visualisation</title>

(Really) minor bug, but hey :)

[diamonddave]

Originally Posted by Dubslow

Either way, the Saved/Done ratio is still more than twice as big for LL as opposed to DC.

Because all summer people worked to bring the DC range from 66 to 68 bit, so the easy factors are gone.

[Dubslow]

Indeed, I was one of them. 69 bits is rather high, which is why I stopped with DC-TF work. I'm still fairly sure that the CPU cores + GPU being put to DC TF would actually clear more exponents just doing straightforward TF. I could clear 2-3 a week if I quit mfaktc (doing LL-TF ATM).

[chalsall]

Originally Posted by Dubslow

http://filesmelt.com/dl/screenshot321.png
Still trying to load as I type. (Note at bottom: Waiting for gpu.mersenne.info...)

OK. I believe you. All I can tell you is that I, and many others, are currently having no problems. (I "tail -f" the HTTPD log files...)

Originally Posted by Dubslow

Either way, the Saved/Done ratio is still more than twice as big for LL as opposed to DC.

Agreed.

Originally Posted by Dubslow

And on mersenne.info I get (at the top of course)

<title>W/elcome!!! :: GIMPS Visualisation</title>

(Really) minor bug, but hey :)

Thanks for pointing that out. I appreciate hearing about every bug, even if only a spelling mistake or typo. (I'm an appallingly bad speller...)

[Dubslow]

Originally Posted by chalsall

OK. I believe you. All I can tell you is that I, and many others, are currently having no problems. (I "tail -f" the HTTPD log files...)



Agreed.



Thanks for pointing that out. I appreciate hearing about every bug, even if only a spelling mistake or typo. (I'm an appallingly bad speller...)

Funky.

(I'm in a weird mood this morning, perhaps because I'm awake and responsive and it's still morning. Maybe I should do this 'real bedtime' thing more often...)

[bcp19]

Originally Posted by chalsall

Actually, that's only half true...

DC TF: 6.401 GHz Days Work Done, 24.797 GHz Days Work Saved.
LL TF: 12.675 GHz Days Work Done, 182.163 GHz Days Work Saved.
(Average per factor found. GHz Days Work Done is "Normalized".)

Originally Posted by Dubslow

Either way, the Saved/Done ratio is still more than twice as big for LL as opposed to DC.

I'm a little confused here... 24.8 / 6.4 ~= 4 to 1 where 182 / 12.7 ~= 1.5 to 1. How is 1.5 twice as big as 4?

[chalsall]

Originally Posted by bcp19

I'm a little confused here... 24.8 / 6.4 ~= 4 to 1 where 182 / 12.7 ~= 1.5 to 1. How is 1.5 twice as big as 4?

182 / 12.7 ~= 14.3.

[bcp19]

Originally Posted by chalsall

182 / 12.7 ~= 14.3.

Wow, , I need my glasses lol.