[QUOTE=tha;394355]I made a graph of the 53.000+ factors in the 12M range. I made a distinction between the lowest factor of an exponent and successive factors of each exponent found. I'd like to hear comments on the drop from 55 bits to 57 bits, the blue bars in the graph.

[The number of bits are rounded up, so ²log(exponent)=33.1 is counted as 34 bits][/QUOTE]

The blue lines are _all factors_ as opposed to first factors. When recalculated using actual first factors, the chart looks like this.

Here's the numbers:
[CODE]25 2349
27 3038 107
28 1301 97
29 2048 271
30 1642 305
31 1674 396
32 1479 389
33 1426 452
34 1421 458
35 1305 502
36 1194 514
37 1143 525
38 1049 532
39 986 564
40 916 576
41 941 573
42 877 576
43 854 608
44 741 602
45 771 629
46 708 579
47 661 611
48 684 613
49 644 613
50 639 615
51 634 574
52 605 571
53 533 593
54 536 600
55 533 612
56 497 365
57 466 4
58 463 7
59 453 5
60 399 14
61 402 8
62 403 6
63 404 5
64 359 7
65 357 2
66 353 2
67 90
68 71
69 68
70 65 1
71 50 3
72 34 3
73 38
74 41
75 28 1
76 38 2
77 29 2
78 27 1
79 21
80 22 1
81 19 2
82 16
83 14 1
84 9 1
85 16
86 10
87 7
88 6
89 4
90 3
91 8
92 7
93 5
94 2
95 2
96 3
97 2
100 3
101 1
102 2
105 1
110 1
112 1
115 1
[/CODE]

[QUOTE=axn;394401]The blue lines are _all factors_ as opposed to first factors. [/QUOTE]

Thanks, it was an error in copying cells.

I will re-run some if not full 12M range to 61bits, so expect some hundreds of factors coming soon. However, what I am curious is how you determine which is the first factor. Is it by date-found or the smallest of many? It would be nice to see if we have some misses in any range by date-found (smaller factor to be found on a later date). This will tell us something on the TF trustfulness in the range. Could any of you check if we have misses (smaller found later) in 12M range, since you already have the data.

[QUOTE=bloodIce;394409]I will re-run some if not full 12M range to 61bits, so expect some hundreds of factors coming soon. However, what I am curious is how you determine which is the first factor. Is it by date-found or the smallest of many? It would be nice to see if we have some misses in any range by date-found (smaller factor to be found on a later date). This will tell us something on the TF trustfulness in the range. Could any of you check if we have misses (smaller found later) in 12M range, since you already have the data.[/QUOTE]

I just ran a couple of hundred exponents from 55 to 59 bits this night and all were OK. I define first exponent as the smallest here, since I did not use any data with dates in it.

I am currently using one GTX580 to do the range from 66 to 68 bits and found 12 factors in the first 2% of the range.

[QUOTE=LaurV;394395]Few thousands already (new). Unfortunately, nothing "first" yet. Still going.[/QUOTE]
Finished Unc's range, nothing "first". Few interesting cases, like [URL="http://www.mersenne.org/report_exponent/?exp_lo=332860091&full=1"]this one[/URL], for example, with four factors in a close range. Nothing important for the project itself, wasted time.

Since we started this discussion I did one million expos from 100M and another million from 332M, finding over 7300 "new" factors, for about 5500 expos, but nothing "first", i.e. all the expos had already smaller known factors (well, their associated mersenne, I mean). This activity is loss, unless we have some "known intervals with less than expected factors", or we have some "user with laptop problems" as already happened. Taking random expos from 0 to 6x bits (to dublecheck former TF work) is no gain for the project. Therefore I will stop this, and going back to classical LL/DCTF from gpu72 for now.

Never Odd Or Even from 2014-03-08
 

But, we do have a user with a faulty laptop:
[URL]http://ftp.mersenneforum.org/showthread.php?t=19184[/URL]

Everything on the date 2014-03-08 from Never Odd Or Even, should be double-checked :smile:. I am doing extensive ECM on these, but as I said a year ago, any GPU power will help a lot. Back then I PMed George without a result and I was instructed to write a spider on a toilet paper with brown crayons. Well, I did that, I have the spider, conceived as recommended. Any GPU power to spare anyone?

After reading that thread it looked like he was finding most of the factors. It would not be an effective use of time to recheck everything, unless George can provide us with a list of suspect results,

[QUOTE=Mark Rose;394428]After reading that thread it looked like he was finding most of the factors.[/QUOTE]
What do you mean by that? NOOE missed the factors, others found them. It is only one date in question, where the misses were detected. It is not all his work (the rest I hope is genuine). It is one, very faulty report of ~230 expos mysteriously TFed to 70 for a less than a week. There were proofs that in this work he (or some imposer) missed factors.

[QUOTE=bloodIce;394465]What do you mean by that? NOOE missed the factors, others found them. It is only one date in question, where the misses were detected. It is not all his work (the rest I hope is genuine). It is one, very faulty report of ~230 expos mysteriously TFed to 70 for a less than a week. There were proofs that in this work he (or some imposer) missed factors.[/QUOTE]

So were those 230 expos doublechecked?

No, they were not checked. I am sorry if I do not express myself clear. There are ~230 suspicious exponents reported on one exact date by one exact user. At least two misses in these are detected by me (the third is in a previous work). Nothing was done to correct the error (only jokes on my expense). That is why I do systematic ECM work on these to clear them (well I did find 6 factors + 2 missed factors in total of ~ 36 ECMed expos, so there is a point). However, a lot of the rest are probably erroneously reported to be factored to 70bits, although that is probably not the case. That is the story.

I can redo them if someone provides me with a list.