ForumsFlash Games

Gemcraft CS: semi-optimal combining method for mana gems and amps

Posted Jul 10, '14 at 7:46am



11 posts

Good to know...

Also just now did a quick test in my current endurance run. I started from a 32-spec mana-gem with and w/o red component. Both have 14 black G1 and 18 orange vs. 17 orange plus 1 red. This decreases the mana-leech value by 3.8% before multiplying with the dual/triple-color gem bonus.
After 64-combining both 10x to G55 gems, the mana-leech value of the triple-color-gem is still 3.8% less (again, if you take out the dual-/triple-color bonus). By combining 1 R-O-Bl and 63 R-O G55 gems to a triple-color G60 with the 64-combine method, I gain 3.7% of mana-leech back and lose only 8.2% of the chain hit length. I think Astroshak calls this "squeezing out the red".

Because the chain hit length still grows in the next upgrades even though further growth is not necessary, gaining back the mana-leech even if it is only 3.7% is nice. This triple gem basically grows like he dual-color gem with the triple-color gem bonus and a significant chain hit length due to the blood-bound modifier. No need to squeeze it out further, as that does not gain a better mana-leech value.

In short: you can gain back the "inefficiency" of using a R-O-Bl 32-spec mana gem in terms of mana-leech value at any time by combining 1 triple with 63 dual color gems in the 64-combine method. You should do this as soon as the number of perma-frozen monsters is smaller then the chain hit length...


Posted Jul 10, '14 at 4:43pm



267 posts

@TheMalT ... when I "squeeze the red out" I start out with two gems per area (IE - two manafarm gems, two killgems). One is the "psorek standard" 32 spec triple gem, the other is a modified, dual gem (no red). I'll do the 64 gem upgrade on the triple gem, using 1 triple and 63 duals. I'll then do the 64 gem upgrade on the dual, keeping both the triple and the dual at the same level.

When doing this, I can squeeze the red down a lot. I have to stop squeezing it out after displayed grade 35 though, or I'll lose it completely. Dual gem with triple modifiers and no chain hit at all is no good to me! At that point, it is just a regular 64 gem upgrade on the triple gem, no more duals.

As you have noticed, squeezing the red out of the gem reduces the chain hit (only matters if the number of monsters on the trap exceeds the chain hit amount) and increases both the black and the orange/yellow, which increases the overall effectiveness of the gem.

What I do not know, though, is whether you are stuck with only being able to squeeze the red out (1ROB:63OB) six times only using the 64 gem upgrade, or if starting out a couple of upgrades not squeezing it out will allow you to squeeze more out later on. I've also never tested to see if using a different ratio (2ROB:62OB, or 3 ROB:63OB) would allow me to ultimately squeeze more red out without eliminating it entirely.


Posted Jul 11, '14 at 4:55am



11 posts


When you divide the mana-leech-value of the triple gem by 1.5 and by the blood-bound specials-multiplier (the full one, not the one per hit level), you get the base-mana-leech value. To get that base value for the dual gem, you divide by 2.1 and the blood-bound multiplier. 150% and 210% are numbers derived from the "true colors" skill, so could be different for you, if you are not maxed at 45+15 skill levels.

These base values are growing by upgrading, e.g. with the 64-combine method. By exchanging a single orange G1 in the 32-spec mana for a red, you lose about 4% of the of the mana-leech base value. The black is not affected, as it has the same contribution to the 32-spec mana gems. A single "squeezing" out combining of 1 triple with 63 dual gems gains you back most of that 4%. Doing more than 1 squeezing out will significantly lower the red component to the point it drops out, without gaining significantly better mana-leech.

If you want to test that, compare the following gems:
64x (1x 32-spec triple + 63x 32-spec dual) = G15 triple
1x (1x 32-spec triple + 63x 32-spec dual) + 63x (64x 32-spec dual)

If both resulting G15 triple gems have 0 hits, they both have almost identical mana leech values but the first one has a better - even if unnecessary - chain hit length. In other words, don't bother to keep around and keep upgrading the 32-spec dual color gem after the first round of red dillution...

This of course depends a little on how you mix the 1 triple + 63 dual gems. You should avoid putting the triple in a part of the 64-combine scheme that gets duplicated -- like the 8.1-8.4 intermediate results. Otherwise the dilution of the red component is not optimal...


Posted Jul 11, '14 at 6:22pm



267 posts

I just went and took to grade 35 three (ok, four) gems. One was a 32 spec, one was a 32 spec that I squeezed the red out once, and one was a 32 spec I squeezed the red out each time. Grade 35 is the point I stop squeezing it out, because further 1ROB:63OB combinations eliminate it entirely. (The fourth was the dual OB gem.)

The difference in mana per hit, using Grade 34 amps, was small - less than 1 percent between the no squeeze and my nearly-full squeeze. Oh, I'm sure there's still that 4% or whatever difference between the base amounts for the two, but the difference was between 1 and 2 million mana, when the total amount leeched per hit was over a billion. The difference in chain hit was a lot more significant, over 400 in the no squeeze gem, some 340ish in the "squeeze once" gem, and 113 in the nearly-full squeeze gem. At each level, the more squeeze I put on the red, the more mana leech there was, but as you noticed, it is a small amount when read as a percentage.

With the idea that the Bloodbound is the only real improvement to the Chain Hit, eventually no matter how much you squeeze it out, you are going to have more chain than monsters. In my current game, I have a Grade 70 managem (no squeeze) with 941,441,608 hits - whatever the HL on THAT is - and its chain is 182,136.51. The killgem, Grade 65 (almost full squeeze) and one or two HL's below the managem, is at 43,842.83 chain hit.

Personally, I figure to almost kill the red. The difference between "no squeeze", "one squeeze", and "almost eliminate" is not very big, but it is still a difference, and when chain gets over the number of monsters on the trap, additional chain is useless and the only benefit you can get is more mana leech, either through squeezing the red down or by increasing the gem's grade.

Most manafarm attempts people have spoken of, on these boards or in the kiwi, that used permafreeze involved permafreezing only a few waves. You would need 44 waves to saturate the chain of my killgem - and a whopping 183 waves to saturate the chain of my managem. I do not know how feasible permafreezing that many waves actually is. My decent PC is showing some lag at just under 44k monsters (I've actually got enough monsters on field to saturate my killgem's chain, by about 140). I do not figure that a supercomputer that could handle 100k+ monsters at once is in most people's hands ... I don't expect my computer to handle that many with any degree of actual playability. The lag I'm experiencing is enough to turn the 100 second Manalock into something much longer; I guesstimate about 15 minutes, meaning I'm lagging bad enough that everything is running at 1/9 the normal speed.

So .. squeeze the red out, until one more squeeze would eliminate it entirely.

last edited Jul 11 2014 06:23 pm by Astroshak

Posted Jul 13, '14 at 7:26am



645 posts

I forget the value that psorek gave, but according to him, he calculated that the 16 gem plan is some 20-30 percent stronger than just U by grade 75 or so (that's the cost of a grade 75 gem, not an actual displayed grade 75). The 64 gem upgrade was comsiderably better, at some 50 or 60 percent better, I think. I forget, its been a while, but it was somewhere around there.

This is w/o 32 spec first or with? With it, at G50e, difference is about 4.35x bigger for 16 combine alone. I expect ~6.5-7.5x for 64 combine by then compared to U.

Note that to compare it to a pure U-ed gem, you can't spec it however, to compare it you must 32 spec the supergem because its taken for granted that you use it. Also, just input the ratio of Black/Orange/Red or just BO. 10/21/1 or 10/22.
Reply to Gemcraft CS: semi-optimal combining method for mana gems and amps

You must be logged in to post a reply!