Davieddy's World

[davieddy]

Originally Posted by c10ck3r

Area of a square= L²

Great to see you picking up on our high-powered discussion.
But where's the snide remark?

[davieddy]

Originally Posted by chalsall

That is his unsupported claim. And what he would like everyone to believe.

It's not true.

The "unsupported claim" is that LL allocation is "chomping at the heels" of TF.
It is what everyone including you believes.
1) Your tables show clearly that TF would drop below LL completion rate were we to TF everything >63M to 74 bits.
2) Your proposal is to go to 74 bits for a couple of months (squandering the unnecessarily hard-earned lead) and then say "Oh dear. Looks like we'll have to go back to 73 bits for some expos".

BTW what accounts for the small number of expos<63M TFed to <73 bits in your table compared with the thousands of TFs still allocated on the primenet summary page?

We shouldn't. Although that is what he would like us to do.

That was "divert resources to LL rather than TF above the optimal level".

You are nuts. What do you imagine defines the optimal TF bitlevel?

The LL completion rate needs to increase by 10% per year, which requires a 33% increase in computing power, to maintain our expectancy of 4 years for a new prime. Some of this can be expected from increased CPU speed, and some from increased number of LLs in progress.
IMO we could reasonably hope to improve on 33%, and reduce the expected time further. After all, it was languishing at 6 years until quite recently.

David

[kladner]

Originally Posted by davieddy

Great to see you picking up on our high-powered discussion.
But where's the snide remark?

It seems that you have provided that part, and can be counted upon to do so when you are not crying foul on someone else.

I do congratulate you on your skill at keeping so many people on the line indefinitely with variations on the same arguments.

[c10ck3r]

Originally Posted by davieddy

Great to see you picking up on our high-powered discussion.
But where's the snide remark?

That was the snide remark...

[c10ck3r]

Originally Posted by davieddy

1) You are nuts. What do you imagine defines the optimal TF bitlevel?

2) The LL completion rate needs to increase by 10% per year, which requires a 33% increase in computing power, to maintain our expectancy of 4 years for a new prime. Some of this can be expected from increased CPU speed, and some from increased number of LLs in progress.
IMO we could reasonably hope to improve on 33%, and reduce the expected time further. After all, it was languishing at 6 years until quite recently.

David

1) The optimal TF bitlevel, IMHO, is that at which the probability of finding a factor per GHzDay spent is equal to the rate of completing 2 LL tests. I'm sure this could be better written, but...well, whatcha gonna do about it? :P
An argument could be made for going further, but that seems to be more math-y than I feel like going into atm.
2) The LL completion rate doesn't "need" to do anything. It could be equal to zero and we could still progress "efficiently". If you look at any other prime search, you'll notice that sieving to a proper depth is done before any tests are begun. They sieve the entire range, not just the first ~20M candidates. Just my 2cents.

[Aramis Wyler]

That was a much shorter break than I had anticipated.

[chalsall]

Originally Posted by c10ck3r

The optimal TF bitlevel, IMHO, is that at which the probability of finding a factor per GHzDay spent is equal to the rate of completing 2 LL tests. I'm sure this could be better written, but...well, whatcha gonna do about it? :P
An argument could be made for going further, but that seems to be more math-y than I feel like going into atm.

Just for clarity, please remember that James has done a wonderful analysis based on empirical data showing where the "optimal TFing level" actually is.

We should have been going to 75 from 57M onward; we simply didn't have the firepower to do so.

[chalsall]

Originally Posted by davieddy

3) Setting the default to 73 in no way precludes folk going to 74 if the mood takes them. The point is to avoid allocating anything <73 for LL.

For (what feels like) the thousandth time, we haven't, and WON'T!!!

If you turn out to be correct and I'm wrong, we simply release a few already at 73 if and when needed.

Frankly, this is obviously futile. You're now on my "ignore" list.

[Uncwilly]

Originally Posted by chalsall

If you turn out to be correct and I'm wrong, we simply release a few already at 73 if and when needed.

Reminds me of this:
http://c2.com/cgi/wiki?IdontHaveToBe...tFasterThanYou

[swl551]

Originally Posted by kladner

It seems that you have provided that part, and can be counted upon to do so when you are not crying foul on someone else.

I do congratulate you on your skill at keeping so many people on the line indefinitely with variations on the same arguments.

Daviddy, you have the skills and the tenacity of a very successful car salesman. However, here, your style isn't making the sell. Do you really expect to win the minds any longer? I would've thrown in the towel by now and basically ditched these "fools". I think you enjoy the fight too much for your own entertainment. Thanks for keeping me entertained though.

[davieddy]

Originally Posted by chalsall

We should have been going to 75 from 57M onward; we simply didn't have the firepower to do so.

For the 50th time in two days, we don't quite have the firepower to take all expos >63M to 74 bits, but that doesn't seem to deter a man wearing rose-coloured spectacles such as yourself.

x

PS Note how clearly James' graphs follow my simple rule that the bitlevel should increment when the exponent increases by 1.26 (time on TF remains proportional to time for an LL test).