FEATURED G.Skill Trident Z 2x8GB DDR4-3000 CL14 - F4-3000C14D-16GTZ

[Woomack]

This short review will be based on the latest G.Skill Trident Z 2x8GB DDR4-3000 CL14-14-14 memory kit which is hidding under the F4-3000C14D-16GTZ product number.

As you can see in the below specification window, new Trident Z 3000 CL14 is based on about the same IC as Trident Z 3600 so the latest Samsung which overclocks great. Later I will show some results at higher frequency.





Here is screenshot showing stability after 3h AIDA64 tests.

 

[Woomack]

OC Results - 1.35V

3200 14-14-14-32 1.35V

Spoiler

3466 15-15-15-32 1.35V

Spoiler

3600 16-16-16-32 1.35V

Spoiler

3733 16-16-16-34 1.35V

Spoiler

3866 17-17-17-36 1.35V

Spoiler

4000 17-17-17-36 1.35V

Spoiler

4133 18-18-18-36 1.35V

Spoiler

So far 4133 is maximum as my MSI Z170I Gaming Pro AC motherboard has no higher memory ratios and bclk overclocking is causing instability. I guess this memory can make more so I wait till MSI fix the BIOS or I will test it on other motherboard when I get any better.
It's hard to complain at DDR4-4133 1.35V considering this Trident Z kit is designed to work at DDR4-3000 speed.

 

[Woomack]

OC Results - >1.35V and on X99

32GB 3466 16-16-16 1.35V on ASRock X99
TridentZ 3000 and TridentZ 3600 on the same Samsung IC.




32GB 3466 14-14-14 1.45V on ASRock X99
TridentZ 3000 and TridentZ 3600 on the same Samsung IC.




32GB 3466 13-13-13 1.55V on ASRock X99
TridentZ 3000 and TridentZ 3600 on the same Samsung IC.

 

[PolRoger]

Nice results.

Well I'm guessing G.Skill is now binning/using the same B revision ic for all these newer 2x8GB kits with matching primary timings... (3200C14/3600C16/3733C17 as well as on their 4000/4133C19 kits). I was kind of wondering why they don't offer a 3400 15-15-15-35 kit?

How much ~VCCSA/VCCIO voltage are you using for testing?

 

[Mandrake4565]

Yeah nice results Woomack thanks for sharing!

 

[Woomack]

4000+ kits are still on old IC. B die is in 3000 C14, 3200 C14, 3600 C16, 3733 C17 and only in 8GB and 16GB modules. 3866+ kits are based on 4GB single sided modules.

For tests I set 1.20-1.25V VCCSA/VCCIO. My board is limited to 1.25V even though it supposed to work up to 1.5V. I also can't set more than 4133+bclk but bclk adjustment isn't really working. I can set 102-103MHz but above that board is losing stability. I wonder how MSI tested 4266 G.Skill kit. This is the only MSI board which has 4133+ memory on the tested memory list what is also interesting as even their highest OC board doesn't have that.

 

[PolRoger]

... only in 8GB and 16GB modules. 3866+ kits are based on 4GB single sided modules.

I was thinking about these two 4000(+) 2x8GB kits... 8GB sticks single-sided... B-die?

http://www.newegg.com/Product/Product.aspx?Item=N82E16820232268

http://www.newegg.com/Product/Product.aspx?Item=N82E16820232267

 

[Witchdoctor]

Yikes, that stuff is expensive, Nice results Woomack

Appreciate you taking the time

Ripjaws using the same IC's ?

 

[Woomack]

I was thinking about these two 4000(+) 2x8GB kits... 8GB sticks single-sided... B-die?

http://www.newegg.com/Product/Product.aspx?Item=N82E16820232268

http://www.newegg.com/Product/Product.aspx?Item=N82E16820232267

These are probably double sided as I saw them for longer in the stores but maybe G.Skill is refreshing these kits too. I haven't seen anything 3733+ listed as new kits. I can be wrong here.

I'm trying to make some results at higher voltages but for some reason HyperPi 32M is crashing faster once I set more than ~1.45-1.50V. Maybe it's something with this MSI board.
ASUS Hero back from RMA but it couldn't make so high memory clocks as MSI. Will check that later.

Ripjaws V in some kits are using the same IC. I just had no chance to test any new Ripjaws V.

 

[PolRoger]

What kind of speeds can you get out of the full size atx Hero motherboard with these new Samsung B-revision kits?

I'm also been kind of curious about the how these new double-sided 32GB kits (2x16GB) kits might clock? I haven't seen any tests/results out on the web yet?

http://www.newegg.com/Product/Product.aspx?Item=N82E16820232206

 

[Woomack]

I asked couple of times but no one wanted to send anything 2x16GB for review. I'm also curious how these kits are overclocking.

On Hero I couldn't boot above 3866 and I had some issues with stability at tight timings so I back to MSI today. I will probably get ASRock OCF in couple of days.

Quick dual channel valid @4257 on MSI. I can't make it run much higher. Once it crashes I have to reflash BIOS or bclk adjustment isn't working.

 

[Woomack]

While I'm waiting for Z170 motherboard which is in RMA, I'm playing a bit with TridentZ 3000 and 3600 memory kits ( 32GB in total ) on ASRock X99M Killer motherboard. Result are in 3rd post in this thread. I will try to add some more.

 

[leopelocastano]

Hello, I am new in the forum, I wanted to ask if you can recommend me a motherboard that can achieve 4000mhz in 2x8gb and to be sli compatible. Besides I wanted to know if you have performed more test on this interesting memory. Thanks!!

 

[Woomack]

to the OCF

This memory is about the same as 3200 CL13-13-13/14-14-14/15-15-15, 3600 15-15-15/16-16-16 and some 17-18-18, 3733 17-17-17/18-18-18 and some higher kits. I was simply not adding results here as there are other threads with similar memory kits so it's no point to make the same tests on each kit. Most of these Trident Z and Ripjaws V will overclock almost the same. There are some differences but in general can expect almost the same.
I have 2x8GB TridentZ 3200 14-14-14 which I'm using for a while and there is no thread about it yet but it's overclocking the same as 3000 14-14-14 so I'm not in a hurry. Maybe will add results on new platforms, will see.

If you want 4000 then stick to ASRock or MSI motherboards. ASUS in most cases make problems above 3866 ( I had 3 Maximus boards ). Impact seems to overclock the best. ASRock Z170 OC Formula and it's micro ATX version are the best options for memory overclocking. I also have MSI Z170I Gaming Pro AC which can make 4000+ without issues.
Simply if manufacturer tested 4133-4266 memory kits and you can see them on the tested memory list in dual channel then this motherboard will work stable at 4000 ( not even guaranteed higher as there are some variables which affect overclocking ). However larger motherboard = lower maximum memory frequency. Exception is mentioned ASRock OC Formula and maybe 1-2 top MSI.
On the other hand if you want it for gaming then I see no point going above 3600. Better will be to set something like 3600 13-13-13/14-14-14.

 

[leopelocastano]

Thank you very much sir, your info is very apreciated.

Following your advice that 3600 mhz with lower latencies will be better for gaming is that I want to ask you, based on your experience, if those top tier motherboards (that allows very high frequencies) are better in order to achieve lower timings too or if depends only on the memory chip itself (leaving the IMC aside).

Because maybe instead of buying that kind of mobos, I can buy something that supports for example 3600 mhz and work on better timings. (with something cheaper, lets say Msi gaming M3).

Thank you in advance and sorry if this is not the adecuate thread to to be asking.

 

[Woomack]

Most motherboards will work at lower timings. Many simply won't pass ~3733 in dual channel. The only memory IC which can make 3400+ CL13-13-13/14-14-14 stable at reasonable voltages is Samsung "B" which is in Trident Z and some other kits ( but not many ). The best for overclocking are 8GB modules. About everything I mentioned in my last post so with timings like 13-13-13, 14-14-14, 15-15-15, 16-16-16, 17-17-17 and 18-18-18 will be on Samsung B but can be some exceptions like 3200 16-16-16 are on couple of different IC.
Trident Z 3000 14-14-14, 3200 14-14-14/15-15-15 and 3600 16-16-16 I was able to run @~3466 13-13-13 in quad channel on X99 motherboard. If it's working on X99 then I don't think there will be any issue to set the same on any Z170 motherboard where were tested 3600-3866 memory kits in dual channel.
Stability of CL13 or lower at higher clock depend on some other things like signal quality ( PCB quality etc ). At CL12 or less Windows has problems to keep stability. You can see results around the web at 3500+ CL12 but they're not fully stable. Memory can make it but other components not really.

I don't know what about M3 but I was testing Z170 Mortar which has the same max tested memory frequency and it was working at 3600 but not much above that.

 

[leopelocastano]

Thank you sir. Regards from Argentina

 

[Fasttrack]

With all due respect, the info on the 3733+ kits is inaccurate.

I own two sets of G.Skill Trident Z 4133, which are B-die single sided.

On Asus Impact 8, I am rock stable at 4000 / 12-11-11-28-1T-220trfc, very tight seconds and thirds, RTL and IOL manually set at 49/50 - 7-4. 1.91v, 1.18V, VccIO, 1.2750VccSA.
Same mems on same board, 4133 / 12-12-12-28-1T-240trfc, same seconds thirds RTL and IOL, IMC on cold, 2.01V, 1.25VccIO and 1.30VccSA.

SPI32m stable with copy wazza.

Thank you

 

[EarthDog]

4133 CL12... I'd love to see some validations... impressive! Is that on subambient cooling?

 

[Fasttrack]

4133 CL12... I'd love to see some validations... impressive! Is that on subambient cooling?

DICE. The memory controller is in great danger at these levels of voltages. Must be cool.
As for validation I have submitted a random Maxmem2 benchmark, at 4000 /12-11 on the M-OCF. If I remember correctly it was a world top 10 result