COMPLETE!!!! Thinking out loud about getting under 20M unfactored exponents

[chalsall]

Originally Posted by SethTro

I'm working in the 6.9M range doing P-1 (B1=4M, B2=110M). I have roughly a 2.5% chance of finding a factor per exp.

Thanks for the update. I have "released" that range from curation.

[petrw1]

Originally Posted by chalsall

12.0M or 12.5M might be of interest to you to work... Please let us know, and either reserve it through the GPU72 interface, or I can "release" it. Your choice.

12.1, 12.3 and 12.6 are most in need or more TF.

But up to firejuggler

[Prime95]

Originally Posted by SethTro

I'm working in the 6.9M range doing P-1

A recommendation for all users. Concentrate your P-1 efforts on larger exponents. 30.8 has some exciting P-1 developments coming. The smaller the exponent, the more P-1 will benefit from the new code.

I'll rush an emergency pre-beta to this group as soon as I can.

[petrw1]

Originally Posted by Prime95

A recommendation for all users. Concentrate your P-1 efforts on larger exponents. 30.8 has some exciting P-1 developments coming. The smaller the exponent, the more P-1 will benefit from the new code.

I'll rush an emergency pre-beta to this group as soon as I can.

How low is "low"?
With 1 exception we are down to 29.8M.
Most of the effort is in the 2x.xM ranges.

[Prime95]

Can you give me two similar exponents to P-1 please. I'll run one on 30.7 and play with the other in 30.8.

[petrw1]

Originally Posted by Prime95

Can you give me two similar exponents to P-1 please. I'll run one on 30.7 and play with the other in 30.8.

Not sure if you want B1/B2 as well. If not feel free to modify.

Wavefront:
Pminus1=N/A,1,2,26899799,-1,1000000,30000000
Pminus1=N/A,1,2,26899981,-1,1000000,30000000

Smaller:
Pminus1=N/A,1,2,9100033,-1,1000000,30000000
Pminus1=N/A,1,2,9100051,-1,1000000,30000000

[tuckerkao]

Originally Posted by Viliam Furik

This may be a bit of a shocker for you, but Uncwilly is Team_Inspector!

(I hope Uncwilly won't get mad at me for this...)

I just achieved 1 of my long term wishes, had Team_Inspector successfully certified 1 of my PRPs -> M2271301.

I believe this will bring me some future lucks.

[lycorn]

Originally Posted by chalsall

12.0M or 12.5M might be of interest to you to work...

Ok, so I will do P-1 on the 12.0M range. It will take a couple of weeks, I guess. Could you please "release" the range for me?
As for TF, I won´t be doing any in the twok ranges. My GPU is simply too slow. I´ve run several TF tests to 73 bits in the 12.9 M range, supposing no one was working there. Not a lot, and no factor found, but if Anon is reading this will probably wish to remove those work units from the worktodo file to avoid duplication.

[Uncwilly]

Originally Posted by tuckerkao

I just achieved 1 of my long term wishes, had Team_Inspector successfully certified 1 of my PRPs -> M2271301.

Certs are something that the server randomly assigns. Except for the ones that George posts in the Strategic DC thread, you can't self-assign certs. I have done over 700 certs in the last year. So, it is no big deal that I happened to get assigned yours. I haven't touched the machine that did for about a week. It runs all by itself.

[Zhangrc]

M108330041 has a factor: 104618158844229118808143
M108390157 has a factor: 86145124187394692093897
6 factors found out of 340 trials!
hurry on before PRP Cat 1 wavefront reaches 108.3M

[SethTro]

With James's help (thanks for everything). I rigged up a P-1 calculator in python and tried to optimize TF vs P-1. I use a similiar methodology to what Wayne does.

If anyone is curious my code is at twok_hard.py but it needs extra data and setup. It also makes a bunch of assumptions (all exponents have the same existing B1/B2, TF success of 1/ (bits + 5), ...) that I might refine. I think it should still be a reasonable estimate of how much work is left for each range

[17000000,17100000] 156 needed, current TF 2155 x 72

	Existing P-1 for interval greater than B1=100000, B2=2025000 => 2.0%
	Last P-1 2155 x 9.6% @ B1=20405627 B2=413213953
	0 GPU(72) + 74390.0 CPU | GHz-Days/factor  GPU: 0.0  CPU: 476.9 | 0.0x GPU/CPU
	Factors/Tests 0.0/0 TF + 156.0/2155 P-1 @ B1=20405627 B2=413213953

	Existing P-1 for interval greater than B1=100000, B2=2025000 => 1.8%
	Last P-1 2155 x 8.0% @ B1=11808812 B2=239128445
	60443 GPU(73) + 43049.8 CPU | GHz-Days/factor  GPU: 2159.7  CPU: 336.3 | 1.4x GPU/CPU
	Factors/Tests 28.0/2155 TF + 128.0/2155 P-1 @ B1=11808812 B2=239128445

	Existing P-1 for interval greater than B1=100000, B2=2025000 => 1.6%
	Last P-1 2155 x 6.3% @ B1=5694836 B2=115320430
	181329 GPU(74) + 20760.9 CPU | GHz-Days/factor  GPU: 3260.4  CPU: 206.8 | 8.7x GPU/CPU
	Factors/Tests 55.6/4310 TF + 100.4/2155 P-1 @ B1=5694836 B2=115320430

	Existing P-1 for interval greater than B1=100000, B2=2025000 => 1.4%
	Last P-1 2155 x 5.0% @ B1=3295622 B2=66736360
	423100 GPU(75) + 12014.4 CPU | GHz-Days/factor  GPU: 5104.1  CPU: 164.3 | 35.2x GPU/CPU
	Factors/Tests 82.9/6465 TF + 73.1/2155 P-1 @ B1=3295622 B2=66736360

	Existing P-1 for interval greater than B1=100000, B2=2025000 => 1.2%
	Last P-1 2155 x 3.5% @ B1=1324436 B2=26819846
	906644 GPU(76) + 4828.3 CPU | GHz-Days/factor  GPU: 8254.9  CPU: 104.6 | 187.8x GPU/CPU
	Factors/Tests 109.8/8620 TF + 46.2/2155 P-1 @ B1=1324436 B2=26819846

	Existing P-1 for interval greater than B1=100000, B2=2025000 => 1.1%
	Last P-1 2155 x 2.0% @ B1=369626 B2=7484926
	1873731 GPU(77) + 1347.5 CPU | GHz-Days/factor  GPU: 13733.4  CPU: 68.9 | 1390.5x GPU/CPU
	Factors/Tests 136.4/10775 TF + 19.6/2155 P-1 @ B1=369626 B2=7484926

	3314264 GPU(1605 x 78) + 0 CPU | 156/12380 TF

Each block stands for completing another bitlevel of TF. Starting with no additional TF and going up to completing TF 2^77

If we do no additional TF we'll need P-1 bounds around B1=20M, B2=413M to get 156 factors from 2155 P-1 tests

If we do one more bit level of TF it takes 60,400 GHz-Days (of GPU time) and we expect to find 28 factors BUT now we can do B1=12M, B2=239M with ~8% success to find the remaining 128 factors. This takes 43,000 GHz-Days (of CPU time).

At the extreme we can do 3 Million GHz-Days of TF (all 2134 exponents to 2^77 + ~1605 to 2^78)

It looks like the most reasonable strategy is going to be TF to 75 or 76 for this range. Depending on the actual P-1 data.

I think the worst range is 29.8M

[29800000,29900000] 91 needed, current TF 2090 x 75

	For M29,800,009 2.1% P-1:  B1=208981  B2=4179626
	Last P-1 2090 x 6.7% @ B1=8264681 B2=165293637
	0 GPU(75) + 53097.1 CPU | GHz-Days/factor  GPU: 0.0  CPU: 583.5 | 0.0x GPU/CPU
	Factors/Tests 0.0/0 TF + 91.0/2090 P-1 @ B1=8264681 B2=165293637

	For M29,800,009 2.1% P-1:  B1=268337  B2=5366746
	Last P-1 2090 x 5.3% @ B1=4264749 B2=85294990
	267864 GPU(76) + 27399.2 CPU | GHz-Days/factor  GPU: 10253.2  CPU: 422.3 | 9.8x GPU/CPU
	Factors/Tests 26.1/2090 TF + 64.9/2090 P-1 @ B1=4264749 B2=85294990

	For M29,800,009 2.1% P-1:  B1=344552  B2=6891038
	Last P-1 2090 x 4.0% @ B1=2200700 B2=44014007
	803592 GPU(77) + 14138.6 CPU | GHz-Days/factor  GPU: 15475.3  CPU: 361.9 | 56.8x GPU/CPU
	Factors/Tests 51.9/4180 TF + 39.1/2090 P-1 @ B1=2200700 B2=44014007

	For M29,800,009 2.1% P-1:  B1=442413  B2=8848268
	1875049 GPU(78) + 5466.6 CPU | GHz-Days/factor  GPU: 24220.7  CPU: 402.4 | 343.0x GPU/CPU
	Factors/Tests 77.4/6270 TF + 13.6/1305 P-1 @ B1=1362729.1842528547 B2=27254583.685057092

	1875049 GPU(2090 x 79) + 0 CPU | 91/6270 TF