welcome to a quick and dirty test on the 4x16GB G.Skill TridentZ DDR4-3400 C16-16-16-36 1,35v (F4-3400C16Q-64GTZ).
![Image](http://abload.de/thumb/jd801429p5jst.jpg)
![Image](http://abload.de/thumb/jd801430tise1.jpg)
![Image](http://abload.de/thumb/jd801431dbsyd.jpg)
The kit is delivered in the standard G.Skill box for DDR4, in which are two memory containers which protect two sticks each. The standard TridentZ pattern applies for these, black PCB with the black andgrey brushed heatspreaders and the red application on top.
A fast visual inspection unveils that the sticks are dual sided, built of 16 IC bricks, and PCB is printed with 3400 16GB, which also matches the new pattern of memory releases of this brand. SN 1611A500 on the labels tells us these are built in march 2016 and use Samsung ICs, workmannship on heatspreaders and general impression matched what we expected from highend memory. Not much more to say about this, let´s go on to some tests.
The Kit was tested on two different mainboards, for initial 2x16gb and especially for 4x16GB full slot tests, we used an Asus Maximus VIII Hero, BIOS, BIOS 0014, and a random 6700K, second test board was an Asus VIII Impact, BIOS 0014 and 0018, with 4400 Pentium and also a random 6700K. First I tested specs on both boards, no problem so I will not even post this, I will contentrate on few results to show the mere quality.
Please note that z170 is a platform, at least for me, which is limited with dual sided sticks, I had this with E-die 8GB dual sided sticks before.
I tested all sticks single for configuartions I post here and they needed up to 0.1v less for settings than in dual channel, and on the Hero, at full slots and tight timings I only achieved 2T while single tests and also 2x16GB dual channel worked at a lot less volts with CR 1T
Results
Asus Maximus VIII Hero
2x16GB DDR4-3600 12-12-12-28 1T 220 tight 1,78v
![Image](http://abload.de/thumb/3600c1232m178vq8szt.jpg)
4x16GB DDR4-3466 C11-11-11-28 2T 300 1,8v
![Image](http://abload.de/thumb/3466c1132m64gbh9su2.jpg)
Asus Maximus VIII Impact
2x16GB DDR4-3600 16-16-16-28 1T auto 1,32v
![Image](http://abload.de/thumb/3600c16132v32mkpq9c.jpg)
2x16GB DDR4-3733 17-17-17-28 1T auto 1,35v
![Image](http://abload.de/thumb/3733c17135vkyk71.jpg)
2x16GB DDR4-3733 12-12-12-28 1T 220 tight 1,85v
![Image](http://abload.de/thumb/3733c12impact32mxtunzssx.jpg)
2x16GB DDR4-3866 C12-13-13-28 2T semi 1,92v (geekbench, xtu stable, no 32m)
![Image](http://abload.de/thumb/geebench3866impact12-u6pom.jpg)
When I draw a fast conclusion, for me it seems that these Samsung B-die based mems are what you expect at a high binned 64GB kit nowadays. You have to live with the fact that duall sided sticks and OC plus tight timings are a hard task for the skylake platform, which clearly showed at single vs dual and quad configuration tests, as stated above the single test needed around 0.08v-01v less at tight settings than dual channel, no matter which of the two boards. A voltage premium came on top when 4x16gb were used, and like on 8GB B-die and hero, we needed 2T CR. In general, I was surprised nontheless that these did stable 3733 c12 tight for all benchmarks on Impact (3600c12 on Hero) and even bootet and ran a couple of benchmarks at 3866 c12-13. Effiiciency at some benchmarks at 2x16gb was slightly better for me than at 2x8GB, and seeing my results, it might be very interesting to see these on 2011-3 and especially on new Broadwell-E, they have the potential to dominate there if the platform can handle these better than Skylake, which clearly is above limit with these
Thanks for visit at this quick test
![Lächeln :)](./images/smilies/icon_e_smile.gif)