XPG Spectrix D50 Extreme DDR4: Elite Memory for Serious Enthusiasts

Specifications and Features

The XPG Spectrix D50 Extreme line is focused solely on pushing the frequency barrier of the DDR4 IC. To accomplish this, they’ve combined the best that modern ICs have to offer with an advanced PCB layout. Achieving high frequency is all about eliminating electrical noise and keeping the PCB traces short, so ADATA implemented the increasingly more common A2 PCB design.

DDR4 PCBs are broken down into three common types, two of which are common and widely used. The older design, called A0, has the ICs spaced evenly on the PCB and can limit the maximum frequency achievable. The newer A2 design places the ICs closer together and closer to the PCB connection edge. The A2-style PCB has become the new unofficial industry standard for high XMP ratings above 4266 MHz.

The D50 Extreme is not just about raw speed, it also includes addressable RGB LEDs. To round out the look, they released this special memory kit with a full-mirror heat sink. The overall theme is geometric with a sharp angular design that includes a triangle-shaped RGB light diffuser. Composed of thick 2 mm aluminum, the heat sink includes deep grooves on each side which add to its style as well as increasing surface area for better cooling.

One unique aspect of this memory that we simply haven’t seen before, is the rated voltage of 1.60 V. That simple fact puts this kit of memory in the extreme enthusiast category. The XMP 2.0 specification carries a voltage limit of 1.50 V.  So ADATA’s stepping pretty far outside of the normal limits with regards to voltage, however, they still offer an impressive limited lifetime warranty.

Due to this being an extreme enthusiast product, we are not surprised to see that the SKU list is quite limited. The product line consists of only two options, see the table below for the full specifications.

Below is a screenshot of Thaiphoon Burner, a free tool that allows users to read the Serial Presence Detect (SPD) firmware of the DRAM. The SPD information is critical in determining how the stick will perform and how the computer will recognize it out of the box.

As the Thaiphoon Burner screenshot shows, this specific kit of memory is composed of Hynix DJR ICs. The Hynix D-Die has gained a reputation for very high frequency overclocking with extreme cooling solutions such as liquid nitrogen. It has been used to break world records on multiple occasions and high-binned examples can easily reach over 6000 MHz with exotic cooling. We know from experience that the DRJ IC can be temperature sensitive and achieving frequencies of 5000 MHz with ambient cooling could prove challenging.

As a result of advancements in DDR4 frequency progression, motherboard manufacturers are now routing memory traces to coincide with the IC placement on A2-style PCBs. For testing today we will use the ASRock Z490 AQUA, which has been specifically optimized for A2-style memory PCBs, so we have a good chance at achieving high frequencies. That said, it’s important to understand that an XMP rating of 5000 CL19 is not achievable on most motherboards. In fact, ADATA only lists one motherboard model in the compatibility list below.

Packaging and Product Tour

The Spectrix D50 Extreme is an elite product designed for extreme enthusiasts and carries a price tag that matches its unique disposition. As such, we expect a premium unboxing experience to help offset the price. We’re happy to report that ADATA hasn’t overlooked the retail packaging, which consists of a classy box and sleeve design. The sleeve is orange with a mirror-like finish and includes all the typical info you’d expect to see on the front and back.

Removing the sleeve we are greeted with a sleek black box with a die-cut foam insert to house the memory safely. As memory packaging goes, it just doesn’t get much better than this.

Meet the Spectrix D50 Extreme

The heat sink is thick and feels quite substantial when holding it compared to many others who use thin, stamped aluminum heat spreaders. The design is simplistic, but it’s refreshing to see a somewhat understated RGB design when competitors taking things in the opposite direction with radical designs.

The color is listed as ‘Gunmetal Grey’, and the finish on the outside is a perfect mirror. The deep grooves cut into the heat sink, combined with an angular RGB light diffuser, give it a geometric look.

The light diffuser does not dominate the entire topside of the module. The heat sink extends up to the very top in spots, which means the light diffuser occupies about 75% of the top-down view real estate.

Spectrix D50 Extreme Illumination

The overall lighting effect is very subdued and mellow. There are no observable hot-spots or bad viewing angles. ADATA clearly isn’t trying to outshine the competition by having the biggest and brightest RGB lights, instead, they opted for a subtly refined look.

Software

We tested the RGB functionality with ASRock Polychrome Sync. The bundled motherboard software effortlessly communicated with the D50 Extreme memory modules and allowed us to control the light show. Additionally, ADATA also has its own standalone software suite – the ADATA RGB Sync App is designed to work independently of any motherboard control software. It works with all major motherboard brands including Asus, Gigabyte, MSI, and ASRock.

Software download: XPG RGB Sync App Beta

Software download: ASRock Polychrome Sync

Testing and Overclocking

If and when the XMP profile has been established to be stable, we will evaluate the memory from an overclocking perspective. As we pointed out in the specifications, the rated voltage of this kit is 1.60 V, however, we want to see what this memory can do with realistic voltages. Therefore, we will stick to what could be classified as 24/7 stable daily memory voltages for the memory and CPU memory controller. According to the XMP 2.0 certifications, the absolute maximum allowable voltage is 1.50 VDIMM. Thus, all overclocking endeavors will be conducted with no more than 1.50 V.

Below is the test system and resulting memory speeds that will be used to evaluate the memory and run the various benchmarks.

Today we’re are using a limited series ASRock Z490 Aqua OC motherboard. We validated this motherboard to work at 5200 MHz CL18 with Micron based memory at a low VDIMM voltage of 1.50 V. This engineering sample board has been specifically designed with extreme memory overclocking in mind and believe it’s one of the best motherboards for this unique task.

XMP Profile 1 ~ 5000 MHz CL19-28-28

Without touching any of the other settings, simply applying the XMP Profile 1 failed to work. In order to make it work so we could test it, we needed to tweak some settings. The first and most important task was to manually apply the secondary XMP timings. When you apply the XMP profile it only sets the primary timings and VDIMM. However, if you look at the Thaiphoon Burner screenshot above, you will see there are many other timings involved in the XMP profile, and the motherboard does not set automatically.

Here is what we did to make the XMP profile work:

1. Tested 6 different 10900K CPUs to find the one with the best memory controller for high-frequency memory overclocking.
2. Manually set all 8 XMP timings, note that some of the timings specified were too high for our motherboard to set, so we went with the maximum value for those.
3. VDIMM increase from 1.60 V to 1.65 V, which was verified with a digital multimeter
4. System Agent voltage increase from auto to 1.60 V (1.30 V is considered ’24/7 safe’ in the overclocking community)
5. IO voltage increase from auto to 1.40 V (1.25 V is considered ’24/7 safe’ in the overclocking community)
6. Decrease the DRAM termination voltage from 0.825 to 0.750 V
7. Placed a 2000+ RPM fan directly on the memory to help with cooling

By torturing the CPU’s integrated memory controller with incredibly high and unsafe voltage levels, we achieved relative stability with the XMP Profile 1. Please note, we would never condone this voltage profile for daily use, as it will most certainly harm the CPU long term. The System Agent voltage is the most important consideration when dealing with high-frequency memory overclocking.

XMP Profile 2 ~ 4800 MHz CL19-28-28

The XPG D50 Extreme comes loaded with two different XMP profiles, thankfully. Especially for a kit of this nature, it’s a great feature to add a secondary option if the primary XMP fails.

After that hard-fought battle with Profile 1, we had no problems at all with Profile 2. The profile applied on the first try and worked correctly with no additional tweaking required. ADATA did the right thing by including this memory profile. If the primary profile didn’t work, and we didn’t have this backup profile, then the memory would be running JEDEC speed of 2666 CL 19-19-19 at 1.2V.

Overclock 1 ~ 3600 MHz CL14-19-19

Now that we got those crazy 1.60 V XMP profiles sorted and working, we decided to have some fun and see what the memory could do with realistic 24/7 safe voltage levels. Many AMD users choose to run memory profiles around 3600 MHz so they can run 1:1 ratios with MCLK and FCLK. Therefore, we decided to create the best profile possible with a frequency of 3600 MHz and a voltage limit of 1.50 V.

We managed 3600 MHz CL14, which is pretty exciting. We tightened up the timings as much as possible, but given the nature of this IC they are still quite loose everywhere else except CAS latency.

Overclock 2 ~ 4800 MHz CL19-26-26

Looking back at XMP Profile 2, we see it’s also set to run at 1.60 V. That high voltage may have been needed for 5000 MHz, but it certainly wasn’t needed for 4800 MHz. We were able to lower the voltage down to the realistic working limit of  1.50 V, and also substantially tighten up the timings, except for the CAS latency. While the timing profile is much tighter than XMP, it’s still wouldn’t be considered ‘tight’ by any standard.

Benchmark Results

First up is AIDA64, perhaps the most common benchmark used to assess memory performance. When it comes to AIDA64, frequency is the most important factor for producing the best score. Our 3600 MHz profile for AMD users got crushed across the board, we expected this for AIDA64. However, we are very happy to see that our safe voltage 4800 MHz profile came out ahead in all of the tests.

Next, we used Geekbench 4 to test our memory profiles. It’s one of the best synthetic benchmarks at predicting the ‘real-world’ performance. This benchmark utilizes 25 unique sub-benchmarks of varying complexity and then collates them to create an overall picture of performance.

Focusing on the total score, we see that the ‘real world’ benchmark results look similar to AIDA64, but with a bigger emphasis on timings. In almost every test example, the two XMP profiles scored less than our overclocked profiles.

Lastly, we examined the performance using a few of the memory benchmark tests offered within the SiSoftware Sandra suite. The flagship product, known as Sandra, is a powerful suite of many different benchmarks used to evaluate the computer performance of all major components, including the processor, graphics, memory, and disk.

What we’ve seen so far is that Geekbench 4 favors tight timings while AIDA64 favors high frequencies. As you can see below, Sandra represents them both! The tight timing profiles showed a substantial gain in memory bandwidth performance. In the Cache & Latency test, we saw the best score come from the 3600 CL14 profile, which shouldn’t come as much of a surprise. Similar to Geekbench 4 results, our overclocked profiles beat the XMP profiles in almost every example, and with safe voltage levels!

XMP Comparison

So we’ve seen how the memory compares against itself when overclocked, but how does it compare against other XMP profiles on the market today? In this section, we exclude all overclocking results and use only XMP profiles for comparison against different brands.

We put the Spectrix D50 Extreme up against 5 other kits of varying speeds and latencies to see who was the fastest and to get an overall idea about performance. As we pointed about above, frequency is king for AIDA64, and expected, the D50 Extreme came out on top in every test. Looking at the write tests, we see that it’s nearly 9% faster than the next fastest kit, an impressive feat for sure.

We expected that the D50 Extreme would dominate the competition for AIDA64, however, we didn’t expect it to be nearly as competitive in Geekbench 4. Focusing on the total score, it came in second place behind the G.SKill 4266 CL16 kit, but it was a very close race.

Conclusion

The physical characteristics of the XPG D50 Extreme are impressive. With a stunning mirrored gunmetal grey heat sink, they certainly make a big impression. We appreciated their approach to RGB lighting and the overall effect. ADATA took things in a slightly different direction and built the D50 Extreme with a more refined and understated lighting effect. Style is always subjective, but overall the memory looks and feels like a premium product, and we couldn’t be happier with it from the cosmetic perspective.

Ultimately, the advertised speed rating did not work as intended out of the box. That is to say, we could not make it work without intervening and seriously tweaking the bios settings. We cannot fault the memory for this because simply put, we are not using the one and only officially compatible motherboard. To get them working, we needed to push the IMC to its limit and force unsafe voltage levels through the CPU. Furthermore, we also needed to bump up the VDIMM to 1.65 V, which is absolutely crazy. The factory rated voltage is already outside of the XMP 2.0 standard and is considered unsafe for daily usage, even within the overclocking community. However, we don’t have a problem with that because ADATA has given this memory their limited lifetime warranty, which is incredibly generous considering the target buyer is overclockers who will punish the memory.

It’s worth noting that an XMP profile of 5000 CL 19 would have been easier on the motherboard and IMC if the memory was outfitted with Micron ICs. We’ve seen that the ambient overclocking capabilities are better, on average. It’s not uncommon to achieve 5000 CL18 at 1.50 V with both AMD and Intel platforms with high-quality Micron E-Die. We’ve also observed that the Hynix DRJ is very temperature sensitive and generally more difficult to overclock on air without extreme cooling. When liquid nitrogen enters the picture, the power of Hynix DJR becomes more apparent. It’s easier to control when cold than Micron E-Die, and also can reach a potentially higher frequency looking at the averages. Unfortunately, we need to return this review sample, so we won’t be reaching the D50 Extreme’s maximum potential with extreme overclocking and liquid nitrogen.

There’s no denying that the D50 Extreme is sexy and fast, but is it worth it? With a price of $849.99 on Amazon, it’s among the most expensive 16 GB kits on earth. However, looking at the competition, we see all ratings of 5000+ MHz come with steep price tags. You can get the Corsair Vengeance LXP 5000 CL18 for a whopping $1079.99, or the Crucial Ballistix MAX 5100 MHz CL19 for $899.99. The competitor pricing puts things into perspective and ADATA’s 5000 MHz offering is actually the most cost-effective. It’s also the only RGB equipped option in the 5000 MHz market, which increases the value even more for some buyers.

Despite the difficulties we experienced, and the high price tag, we’re giving this memory our stamp of approval because it’s a special retail option for those who want to get serious about memory overclocking. The overclockers who buy this type of memory are serious enthusiasts who understand it’s not about the destination, it’s about the journey.

Click the stamp for an explanation of what this means.

David Miller – mllrkllr88

Related Reading:

• T-Force Vulcan Z DDR4 Review
• ADATA Spectrix D50 DDR4 Review
• Mushkin Redline Frostbyte DDR4 Review
• T-Force Xtreem ARGB DDR4 Review
• Crucial Ballistix Elite DDR4 3600 Review