I come with the news many have expected to happen:
This run has crashed.
However! I've since switched to a virtual machine running a throwaway installation of Linux Mint. With this new layer of stability, I can crash-proof any future runs from now on.
Now, for the current run;
With the aid of the new 128 spec + 1024 combine, this gem leeches 44.25% more mana at two less hit levels of the old managem at G102e.
Wait, that can't be right. Well, have a look here;
This is a gem built in 4096, with a 64c operation to bring the cost to G103e. Compared to a standard 64c gem, this one leeches 50% more mana at G103e. Quite the improvement!
With all the improvements to the farm, it should counteract any OS-emulation lag and keep on pace with a native OS run. See you at Grade 108e
Also, a HUGE thank you to TheMalT for making the 1024 combine digestible enough for me to put into practice for the trap, which without it the gem wouldn't have the bloodbound it has.
Whoooo - the 1024c looks very impressive! One point makes me wonder: Your gem has 36bn hits... In my actual run, my managem (128_16s, 64c) has only 18bn hits, it leeches far less mana - as you have stated above- by comparable cost. Why has your gem so many hits? And: Thanks for the pic in the other thread, that makes 1024c "understandable"
I also wanted to let you all know that while generating combine recipes is computationally easy (hence those huge numbers), generating specces is immensely more difficult, so it'll not be possible to have 10k+ specces to match those combines (at least with my machine and my programming skill).
As of now my current farm is 53.8% better than my old 64c incarnation. The 4096c amps really help. Here is one of them:
Just before I was to get my last two amps, I felt a strong temptation to venture forth into the vast forest known as the 65k combine. I, the fool, who knows no bounds, set about learning this last and most complicated upgrade scheme in existence. It took me a few days to learn all the patterns and fix the errors in my personal guide, but, after all that toiling, I can now present you a final comparison at my favorite testing grade:
Going from a 64 gem upgrade plan to a 4096 gem scheme yields the 45.1% improvement as said before in my last post. Switching to the 65k plan, you gain another 13.6% more leech. The 65k combine is overall 64.8% better than the 64 combine. Worth it? If you plan on going past grade 120, sure, use it, but at the level I'm at right now? I would highly recommend first learning and using the 4096 combine instead and wait until you have learned the 65k combine well enough to start squeezing that last drop of power out of your gems. Unless, of course, you are like me, who likes the challenge of such ridiculously large combination schemes
Also, for fun, I did a calculation of the first and last gem and compared their leech rates if they fired at their true speeds:
Standard gem: 8.554e20 mana per second.
65k gem: 2.073e21 mana per second.
I never thought someone could actually do that, congratulations, borthel.
Sadly it also seems that it isn't possible to do much better in reasonable time: 65k orange combine successor, the 262k orange combine, is expected to yield 2.709e12 mana/hit at g97e, making it only a 3.6% improvement over 65k combine.
borthel: I somehow missed that post... Un-be-liev-ab-le! And you did it not only once, you did it till g85... Mr Patience... And nice comparing pics. Well done!
@thunderrider, just to put things in perspective, computing an orange 2^30 combine with my program will keep the currently best supercomputer busy for 23 full days (the 2^24 one will take only some seconds).
(data pulled from Wikipedia and some rough estimates done).
I can speculate about the leech of such gems using my last data fit and extrapolating:
The function here is: G(NC)=0.476791-0.000877444*ln(NC)+0.0150773*ln(ln(NC))
Computing leech at borthel's test grade we get:
NT=4.278320e30/54=7.922e28
Leech_power=P_1*NT^G(NC)
G(2^24)=0.5045846977
G(2^30)=0.505 (not from above formula, here it breaks down)