Overclock.net banner

Team Dark Pro 32GB Memory Overclocking Analysis and Review

10K views 11 replies 6 participants last post by  mllrkllr88 
#1 · (Edited)





In recent months consumers have seen Samsung based DDR4 prices steadily increasing and there seems to be no light at the end of the tunnel. However, I am happy to report to you today that there is hope.

In the overclocking world, Samsung B-Die based memory modules have become ubiquitous with tight timings and high frequency. The market demand has grown exponentially for B-Die memory because of its many advantages, but the downside to this is the constantly increasing prices. Teamgroup Inc. decided to meet the demand of this market by launching their Dark Pro line of memory which appears to be a budget conscious memory module based around highly overclockable Samsung IC’s. While not all of the memory in the Dark Pro lineup may not be equipped with the coveted B-Die, I have confirmed that my test kits do indeed use it.

You may not be familiar with them, but Teamgroup Inc. is not a new company. They were founded in 1997, and they have been a top contender in the memory market since the days of DDR1.





The dark pro lineup offers a few different choices to meet your demands, all are certified XMP 2.0. For the purposes of this research project and review, I will be looking at the kit of memory that is confirmed to be Samsung B-Die. While other modules in the lineup may be highly overclockable and extremely efficient for gaming, my objective is the maximum overclock potential. Therefore, he kit of memory I will be looking at today is the DDR4 3200c14 kit, which runs XMP 2.0 timings of 14-14-14-31.


Teamgroup Inc. Dark Pro Lineup:




Packaging is often an overlooked element to a finished product, but it shouldn't be. The packaging is just a delivery mechanism to ship the product safely, but it also serves as the first impression of quality. We have all opened products with cheaply made packaging and whether you acknowledge it or not I believe that it has an effect on the first impression of the product.

The Dark Pro packaging does not disappoint. While it is incredible simple and effective, it does lend a feeling of quality.


The heatsink design and look is incredibly subjective, but one thing is certain, when you hold the memory in your hands you get the feeling of quality. It has weight and the overall fit and finish feel excellent. On a personal level I strongly dislike anything RGB. I think the memory should be neutral and blend in to the system without standing out. I think the the Dark Pro modules fulfill that requirement perfectly with the predominant color being black and the highlight colors being either gray or red. They blend into the system and allow you to focus lighting efforts in other places.

Dark Pro in Red


With just a small flash of red, they blend in nicely with almost any system. Here you can see then paired up with the ASRock X299 OC Formula.




The overall objective is memory overclocking for benchmarking purposes. Many of the benchmarks today can benefit from both increased memory speeds and tighter timings. With this kit of memory, the XMP is already incredibly efficient combining both high frequency and low latencies. However, my goal is to take things one step further and find out the percentage of gain I can achieve in benchmarks from purely memory overclocking.

To evaluate the performance I will use Geekbench 3. This type of benchmark is purely 3D calculation based, there is no graphical processing element so it's a great analytical tool to evaluate memory performance. Specifically speaking, “Geekbench 3 features new tests designed to simulate real-world scenarios. This helps make Geekbench an invaluable tool to determine how your current computer (or your next computer) will handle your tasks and applications.” (quote taken from www.geekbench.com)

There are two main areas of usage I want to address with this review. In the first part I will look at what can be accomplished using the absolute maximum voltage. According to the XMP 2.0 certifications, the absolute maximum allowable voltage is 1.50v VDDR. Anything above this will void the warranty and potentially kill the memory. In the second part of this review I will evaluate any potential gains to be had from driving MORE than 1.50vddr. In some situations there can be exponential gains in frequency and latency by driving up to about 2.10v VDDR.

Note: I have been overclocking DDR4 just about on a daily basis with over 2.0v and there has been no detrimental effects to the CPU, MB, or memory in over a year of hard use.




Both of my test systems are not contained within a case, they are open air test rigs. This gives me the flexibility to quickly change configurations and run exotic test setups without being constrained to a computer case. The ambient air during the testing was right around 80-85F.

Dual channel memory test system
CPU: i7 8700k
Motherboard: ASRock Z170M OC Formula (Modded to allow Coffee Lake CPU’s)
Memory: 16GB Teamgroup Dark Pro DDR4 3200Mhz
Power Supply: Seasonic 1200W Platinum PRIME
Graphics: 9400GT (I use a very basic GPU for 2D testing)
Storage: OCZ Vertex 2 (64GB)
Cooling: Custom water with 2x360 radiators
OS: Windows Server 2012 R2, 64bit
Maximum memory: Restricting the memory seen by OS to 1.8GB (Required for CL12 tests)

Dual channel memory test rig



Quad channel memory test system
CPU: i9 7940X
Motherboard: ASRock X299 OC Formula
Memory: 32GB Teamgroup Dark Pro DDR4 3200Mhz
Power Supply: Seasonic 1200W Platinum PRIME
Graphics: 9400GT (I use a very basic GPU for 2D testing)
Storage: OCZ Vertex 2 (64GB)
Cooling: Custom water with 2x360 radiators and Kingpin cooling T-Rex for subzero testing
OS: Windows 7 Embedded, 64bit
Maximum memory: Restricting the memory seen by OS to 4.3GB (Required for CL12 tests)+


Quad channel memory test rig





For each platform I will examine the overclock potential under various conditions as described above. However, I thought it would also be interesting to look at how Geekbench responds to increase in memory speeds. Looking at score improvement with only memory overclock is not just a curiosity, it's a real world representation heavy load daily functions.


High End Desktop: X299 Platform, Quad Channel
Test 1: XMP 2.0 (Stock Voltage)
Score: 58568



Test 2: 3733 14-14-14-31 (1.50v or Under)
Score: 59314



Test 3: 3800 12-11-11-24 (Under 2.1v, Actual = 1.85v)
Score 61665




Mainstream Desktop, Z170 Platform, Dual Channel


Test 1: XMP 2.0 (Stock Voltage)
Score: 29986



Test 2: 3733 14-14-14-31 (1.50v or Under)
Score: 30432



Test 3: 3733 12-11-11-24 ((Under 2.1v, Actual = 1.80v)
Score 31537




Below is the calculations showing percentage gain of overclocking memory



I used this memory when I overclocked my i9 7940X with Liquid Nitrogen. I was curious to see if I would be able to achieve any noteworthy increase in memory overclock potential with the CPU under LN2. In this particular case, with X299, I was not able to increase the memory overclock potential. It should be noted that gains from LN2 CPU should be more noticeable with the Z170 platform, but at this time I have yet to test that configuration. Here is the quad channel system in the process of running some LN2 tests




I have spent a lot of time testing other brands of memory modules with both X299 quad channel and Z170 dual channel. In the quad channel configuration, the Dark Pro kit of memory is on par with the best DDR4 available on the market today. I have compared them to quad channel kits that exceed $650 and the overclock potential is exactly the same, this is really quite impressive. However, in the dual channel configuration, with Z170, they do not have the same overclock potential. With other memory modules, testing in the same Z170 MB/CPU configuration, I have reached 4133 CL12 vs Dark Pro 3733 CL12.

As I have shown above, the real world gain in performance from memory overclocks beyond XMP is really not practical. The maximum performance I was able to achieve was about 3.5% but this required voiding the warranty and severely limiting the life of the hardware.

I highly recommend this memory for quad or dual channel configurations. The overclocking potential in Quad channel is really quite impressive, right in line with the absolute best. If you are a power user who is using subzero cooling for highly competitive extreme overclocking in dual channel mode only, then perhaps this memory is not for you.

At the time of this review, you can purchase a 16GB (2x8) kit of this memory on Newegg for $210.00


Pro: Overclocking Potential, Design, Price
Cons: Not on par with the very best B-Die when used in Z170/270/370


 
See less See more
23
#2 ·
Since we had an OCN review of the Team DDR4 I've been wanting to get some when I upgrade. Nice review/test :specool:
 
#4 ·
Great to see some quality reviews on actual RAM overclocking potential without getting too crazy. Keep up the good work.
 
#6 · (Edited)
Thanks for the comments guys! If you get this memory show them off in here ;)


Great review! I decided to pick up the Grey color version of the 16GB 3200c14 kit as it's on sale at Newegg for $179.99.

TDPGD416G3200HC14ADC01
Nice man, that is one heck of a deal for great B-DIE!!!

Edit: I just saw the gray kit is on sale right now for $180 Newegg
 
#8 · (Edited)
They use the RGB PCB, which has different IC spacing and different overclock potential compared to the older B-DIE PCB's. The new PCB spec just doesn't clock as well with high frequency and tight timings.

As for getting them to 4000c12, I binned all 4 sticks and composed a kit of the best two, and still 3866 12-12 was not stable enough for geekbench3. As you have seen, I am using the legendary ASRock Z170 MOCF and even considering the improved coffee lake IMC, its not possible. With other kits I have no problems running 4133 12-12 on air and 4200 12-11 with LN2 CPU for benches like geekbench3.
 
#10 ·
No tests with XP 32-bit 600MB maxmem for SuperPi 32M?

Sent from my Pixel 2 XL using Tapatalk

Not on this kit of memory no. I think perhaps some other Team kits of memory have a better chance of 4000c12 32M wazza. If I can get my hands on a different kit, I will show some 32M wazza testing. :)
 
#12 ·
Thats awesome! Yes, I have also done basic 4kc12 (needed over 2v real) on X299 OCF with this memory, though not to the extent that you have. Nice work!!:thumb:
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top