mersenneforum.org Getting others to do the work on exponents I like (was: Trial Factoring Progress)
 Register FAQ Search Today's Posts Mark Forums Read

2021-11-10, 17:37   #375
axn

Jun 2003

122338 Posts

Quote:
 Originally Posted by Uncwilly Looking at the records I see that you did 22,900 GHz-days of factoring on this exponent beyond the recommended TF level. And the PRP on this exponent takes ~1,145 Ghz-days. So the extra TF work wasted the compute that could have gone into ~20 more PRP's
A 3080 does 50:1 TF:PRP GHz. That means by the time it produces 22k TF GHzDays it would had done maybe 500 PRP GHzDays. The calculus might be more favorable if it were a 6900 (8:1??).

So while I agree that doing too much TF is waste, this isn't the way to argue that point.

2021-11-10, 20:18   #376
tuckerkao

"Tucker Kao"
Jan 2020

67210 Posts

Quote:
 Originally Posted by axn A 3080 does 50:1 TF:PRP GHz. That means by the time it produces 22k TF GHzDays it would had done maybe 500 PRP GHzDays. The calculus might be more favorable if it were a 6900 (8:1??). So while I agree that doing too much TF is waste, this isn't the way to argue that point.
I was doing those extra levels using several of my friend's PCs to see how far the TF levels can go. The main purpose has been to help me deciding which new PC I want to buy. It's the best the keep the stats on 1 exponent instead of involving so many other M168,***,*23 exponents, so I can track it down the TF levels per time spent easier.

Once I get my new PC which I've already ordered, I think it'll be the best to do P-1 with B1 = 1,600,000 and B2 = 65,000,000 with the TF up to 2^79 on that.

TF from 2^78 to 2^79 of M168,***,*23 will take around 2 hours and 20 minutes on Geforce 3080.

Last fiddled with by tuckerkao on 2021-11-10 at 20:22

2021-11-11, 03:53   #377
tuckerkao

"Tucker Kao"
Jan 2020

10101000002 Posts

Quote:
 Originally Posted by Uncwilly Looking at the records I see that you did 22,900 GHz-days of factoring on this exponent beyond the recommended TF level. And the PRP on this exponent takes ~1,145 Ghz-days. So the extra TF work wasted the compute that could have gone into ~20 more PRP's.
If ViliamF was trial factoring the exponents, he was correct. If I did the similar thing I was wrong. The basic UncWilly feelings that could be observed by axn.

Nvidia Geforce 3080 is way more friendly in the video game field than AMD Radeon 6900 XT. I don't just acquire a video card for 1 single purpose, but I don't rule out the RDNA 3 Radeon 7900 XT in the future because AMD is catching up.

There are more users who are willing to work on the PRPs than TFs(especially the bit 1 higher than the recommended level) or P-1, so that makes sense for me to focus more on the TF efficiency than the PRP efficiency.

Geforce 3080 Ti cost \$665 more than Geforce 3080(is on sale) on the website I frequenly look, so there's no reason to buy something like Geforce 3090 because it costs more than double of that Geforce 3080 but not the TF rates.

Last fiddled with by tuckerkao on 2021-11-11 at 03:55

2021-11-11, 14:12   #378
Viliam Furik

"Viliam Furík"
Jul 2018
Martin, Slovakia

2·373 Posts

Quote:
 Originally Posted by tuckerkao If ViliamF was trial factoring the exponents, he was correct. If I did the similar thing I was wrong. The basic UncWilly feelings that could be observed by axn.
I do extreme TF on exponents that are on the FTC wavefront or far behind. The exceptions are 100M digit exponents and 1G digit exponents, but there are more people doing that, so it's nothing extraordinary.

You are TFing a not-as-interesting exponent in the 168M range to an extent that I have never done on exponents other than the 100M or 1G digit ones.

 2021-11-11, 14:58 #379 kriesel     "TF79LL86GIMPS96gpu17" Mar 2017 US midwest 10111111110112 Posts Appropriate TF levels and P-1 bounds are substantial functions of exponent and number of primality tests potentially saved. That is very basic information that has been publicly available for years or decades. https://www.mersenne.ca/exponent/168286523 TF: +6 bits above recommended levels; 23281 vs 364 (~64. TIMES what made sense) Even if that was on an RTX30xx, at its exceptionally high ~50:1 "exchange rate" TF/DP hourly productivity ratio, giving 22917/50 = 458 GHD DP, it could have accomplished 40.% of a PRP/GEC/proof, something useful instead, in that wasted time. P-1 134 GHD expended, vs ~29 recommended (and that assumes 2 tests saved); 134/29 = 4.62 TIMES recommended, reasonable, logical expenditure. Excess was 134-29 = 105 GHD DP, for another waste equivalent to another ~9.2% of a PRP/GEC/proof in that time. Waste of resources = 49.2% of PRP duration. Do it twice, and enough has been wasted to about fully PRP/GEC/proof-gen such a Mersenne number. At a more modest typical TF/DP "exchange rate", the waste is even larger. See also https://www.mersenne.ca/graphs/facto...M_20211111.png or the equivalent of https://www.mersenne.ca/cudalucas.php?model=706 for the GPU involved. (Access that by clicking on the GPU model name in https://www.mersenne.ca/cudalucas.php) Last fiddled with by kriesel on 2021-11-11 at 15:18
2021-11-11, 16:21   #380
slandrum

Jan 2021
California

2×32×17 Posts

Quote:
 Originally Posted by tuckerkao If ViliamF was trial factoring the exponents, he was correct. If I did the similar thing I was wrong. The basic UncWilly feelings that could be observed by axn.
You are free to do whatever you want with your hardware (or whatever hardware you are borrowing). But when you constantly talk about what you are doing, don't be surprised if people comment on it (sometimes negatively, more likely the more frequently you post). If you don't want the negative comments, don't post so much.

2021-11-11, 19:02   #381
mathwiz

Mar 2019

3358 Posts

Quote:
 Originally Posted by slandrum You are free to do whatever you want with your hardware (or whatever hardware you are borrowing). But when you constantly talk about what you are doing, don't be surprised if people comment on it (sometimes negatively, more likely the more frequently you post). If you don't want the negative comments, don't post so much.

 2021-11-11, 20:32 #382 RomanM   Jun 2021 2E16 Posts it's not meant to be, as say mrs. F!
2021-11-15, 22:21   #383
tuckerkao

"Tucker Kao"
Jan 2020

2A016 Posts

I finally found a factor that was greater than 2^100 -> M168184823, 1 of the objective that my father wants me to achieve starting next month. If I can apply the larger P-1 bounds which Kriesel not recommend using my next new PC, it's possible that I can find a factor larger than LaurV's gem which has been mentioned in 1 of the data threads.

Quote:
 Originally Posted by tuckerkao M168,*8*,*23 without known factors:
Both M168184123 and M168184823 revealed their factors, so my list from post #321 is getting significantly shorter.

Also there have 8 such exponents been factored up to 2^78 by Kriesel such as M168732503, but without the P-1, so my next new CPU should come in handy for that.

Last fiddled with by tuckerkao on 2021-11-15 at 22:42

2021-11-15, 23:10   #384
chalsall
If I May

"Chris Halsall"
Sep 2002

27DB16 Posts

Quote:
 Originally Posted by tuckerkao I finally found a factor that was greater than 2^100 -> M168184823, 1 of the objective that my father wants me to achieve starting next month.
If I may please share, I actually resonate with that.

My Father purchased for me the first Computer I could use at home. A TRS-80 Model 1. I had to buy (with hard cash) my own Assembler, of course, to get down to the Assembly Language (about a hundred bucks at the local Radio Shack).

A few years later my Father helped me buy my next computer. A Commodore 64. But, this time, I had to pay for it myself. Father helped with the "bridging", but this was *my* purchase. I worked *hard* with physical labor to pay off that debt.

I miss my Father. He was a difficult man. But the discipline helped me learn how to stand alone.

2021-11-15, 23:50   #385
Viliam Furik

"Viliam Furík"
Jul 2018
Martin, Slovakia

2·373 Posts

Quote:
 Originally Posted by RomanM it's not meant to be, as say mrs. F!
Who is Mrs. F?

 Similar Threads Thread Thread Starter Forum Replies Last Post LaurV GPU to 72 81 2020-12-02 05:17 jfamestad PrimeNet 3 2016-11-06 20:32 lidocorc PrimeNet 4 2008-11-06 18:48 jocelynl 15k Search 0 2003-07-11 14:23 ThomRuley Software 5 2003-05-30 20:34

All times are UTC. The time now is 20:03.

Fri Jan 28 20:03:29 UTC 2022 up 189 days, 14:32, 1 user, load averages: 1.30, 1.50, 1.47