Another nice ECM finding:

[I]ECM found a factor in curve #407, stage #2
Sigma=361548692013119, B1=250000, B2=25000000.
UID: lycorn/asteroid, M507827 has a factor: 1011161162890642691485323460063 (ECM curve 407, B1=250000, B2=25000000)[/I]

k= 995576409772070696797653 =33 × 883 × 41759003807393594933

Totally out of reach for TF/P-1...

Somebody reported a TF factor for an exponent in the First LL range, just above the current recommended factoring limits:

M65840041 has a factor q=31805465644706384183591; log2(q)=74.752...
k = 5 * 47 * 1027812292817 = 241535888811995; log2(k)=47.779...
1027812292817/74.752 = 13749629345.26...

GPU to 72 hands out TF assignments to 2[SUP]75[/SUP].

[QUOTE=TheJudger;392355]GPU to 72 hands out TF assignments to 2[SUP]75[/SUP].[/QUOTE]

And some people "choose" to go many bits higher even for much lower assignments

[QUOTE=TheJudger;392355]GPU to 72 hands out TF assignments to 2[SUP]75[/SUP].[/QUOTE]

Ok, then, I thought it was 74, as that is all I see. Thank you for the correction.

mersenne.ca reports that GPU72 Trial Factoring Limit for M65,840,041 is 73. Probably outdated info...

Mersenne.ca is the "correct" limit.

[rant]
(Un)/Fortunately GPU72 and TGTB have seen the boosting on the factoring power and decided to raise some limits, even if the factoring to those limits is not very efficient (and it should be faster to directly LL, especially for the expos with P-1 done, but hey, finding factors is cool :razz:). Fortunately there are only (very) few exponents in this situation and we will get rid of them fast. More painful will be DCTF to 72/73 for the lower ranges (where there are LOTS of exponents, and far lower efficiency, maybe my calculus is in the weeds, but even[URL="http://www.gpu72.com/reports/estimated_completion/primenet/"] these tables[/URL] show that it will be faster to DC them directly.... for example at 60M, we would need 5 days to LL/DC everything, but we would need 9 days to take all to 75 bits...)
[/rant]

[QUOTE=TObject;392359]mersenne.ca reports that GPU72 Trial Factoring Limit for M65,840,041 is 73. Probably outdated info...[/QUOTE]

Depends on the manufacturer and card... A Nvidia [URL="http://www.mersenne.ca/cudalucas.php?model=9"]"compute v2.0"[/URL] card, for example, is "profitable" going to 75.

[QUOTE=LaurV;392373]...maybe my calculus is in the weeds, but even[URL="http://www.gpu72.com/reports/estimated_completion/primenet/"] these tables[/URL] show that it will be faster to DC them directly.... for example at 60M, we would need 5 days to LL/DC everything, but we would need 9 days to take all to 75 bits...)[/QUOTE]

You are comparing apples and oranges... The table you linked to shows the estimated completion if all resources were dedicated to a particular range based on the throughput for TF'ing and LL'ing.

James' analysis, on the other hand, shows where the optimal cross-over point is between using a GPU to do TF'ing vs. LL'ing.

[QUOTE=chalsall;392433]You are comparing apples and oranges... The table you linked to shows the estimated completion[U] if all resources were dedicated to a particular range[/U] based on the throughput for TF'ing and LL'ing.[/QUOTE]
Like in "if all users of gpu72 would TF the 60M to 75 bits it would take 9 days, but if all users of gpu72 would LL the same 60M, they will finish in only 5 days" apples and oranges?
:razz:

[QUOTE=LaurV;392373]but hey, finding factors is cool :razz:)[/QUOTE]

That's all I do :razz: