Page 2 - A Closer Look, Test System
The XPG Spectrix D50 DDR4-3600 2x8GB comes equipped with a set of medium-profile heatspreaders. The tungsten grey color provides a clean look, which is further complemented by a plastic light diffuser at the top. The aluminum found on both sides is light and makes for a decent heat conductor. The height of the entire module itself is relatively low, measuring around 4.5cm. There should not be many issues when trying to fit a CPU cooler over top of the RAM modules if you keep in mind the size and design of the heatsink on the cooler you wish to install. Aside from looking good, the heatspreaders serve an important role in dissipating heat generated by the memory module. Since integrated memory controllers exist on Intel and AMD CPUs, the memory modules are limited in what you can do with the voltage. Because of this, removing the heatspreaders will not mean they will suddenly overheat. Either way, why you would remove something that provides so much beautiful color?
The design of the heatspreaders for the XPG Spectrix D50 is asymmetrical when viewed from both the front and the side. Looking straight at the front, the XPG logo can be found along the bottom edge. There is an uneven triangular curve down the center top of the RAM module, which displays some more RGB goodness. A specifications label can be found on one side listing the model number, bandwidth, latencies, rated voltage, and memory capacity. The manufacturing location is not included on this list. With some digging, I was able to find out that the XPG Spectrix D50 DDR-4 2X8GB was manufactured in China.
Using a bit of heat and a small nylon pry bar, the heatspreaders can be removed. Taking a closer look at the memory module itself, we can see that the XPG Spectrix D50 has a clean black printed circuit board. The LEDs can be seen on the PCB itself and, as the retail box displays, can be controlled by the lighting software on all of the popular motherboards. The heatspreader pieces are each attached to the memory module by a long strip of thermally conductive adhesive and are manually aligned rather than being physically locked together.
From the photo above, we can also see the specific design of the heatspreaders. The heatspreaders are perfect mirror images of each other when looked at directly. The aluminum pieces do not hold a lot of heat and thus, the heat energy is quickly dissipated into the surrounding environment. Do not let the thin look of the aluminum fool you -- they feel very firm and should not easily bend. With the RGB LED lighting being one of the primary selling points of the XPG Spectrix D50 DDR4-3600 2x8GB, there is really no reason you would need to remove the heatspreaders. If the scenario arises that these RAM modules will not clear the heatsink on your CPU cooler, which is basically non-existent nowadays, it would be better to just find another kit to use.
Taking a closer look at the integrated circuit chips, we can read the code "43QA-0622H SP20440" on each IC present. While it is not immediately clear, I was able to eventually find out that these are Micron manufactured chips identified as MT40A1G8SA-062E:E, with eight 1GB chips on a single side, adding up to a total of 8GB on each DIMM. As mentioned on the previous page, these memory modules are programmed to run at a frequency of DDR4-3600 with 18-20-20-42 latencies. The performance will be covered very soon, which will allow us to see how these latencies affect the XPG Spectrix D50 DDR4-3600 2x8GB when compared to another set of memory modules. The stock voltage of the Spectrix D50 is 1.35V, which is the maximum Intel safe limit along with the recommended safe limit for AMD. The specifications for this specific IC chip are listed below from the datasheet found on Micron's website:
• VDD = VDDQ = 1.2V ±60mV
• VPP = 2.5V, –125mV, +250mV
• On-die, internal, adjustable VREFDQ generation
• 1.2V pseudo open-drain I/O
• Refresh time of 8192-cycle at TC temperature range:
– 64ms at -40°C to 85°C
– 32ms at >85°C to 95°C
– 16ms at >95°C to 105°C
• 16 internal banks (x4, x8): 4 groups of 4 banks each
• 8 internal banks (x16): 2 groups of 4 banks each
• 8n-bit prefetch architecture
• Programmable data strobe preambles
• Data strobe preamble training
• Command/Address latency (CAL)
• Multipurpose register READ and WRITE capability
• Write leveling
• Self refresh mode
• Low-power auto self refresh (LPASR)
• Temperature controlled refresh (TCR)
• Fine granularity refresh
• Self refresh abort
• Maximum power saving
• Output driver calibration
• Nominal, park, and dynamic on-die termination (ODT)
• Data bus inversion (DBI) for data bus
• Command/Address (CA) parity
• Databus write cyclic redundancy check (CRC)
• Per-DRAM addressability
• Connectivity test
• JEDEC JESD-79-4 compliant
• sPPR and hPPR capability
• MBIST-PPR support (Die Revision R only)
Our test configuration is as follows;
CPU: AMD Ryzen 5 3600 @ 3.6GHz
CPU Cooling: AMD Wraith Stealth Stock Cooler
Motherboard: MSI MAG B550 Tomahawk
Graphics: NVIDIA GeForce RTX 2060
Chassis: Antec P10 FLUX
Storage: Western Digital Blue SN550 NVMe SSD 1TB, Samsung 970 EVO 1TB
Power: Corsair CX650M 650W
Operating System: Microsoft Windows 10 Pro
Compared Hardware:
- XPG Spectrix D50 DDR4-3600 2x8GB @ DDR4-3600 18-20-20-42
- G.Skill TridentZ RGB DDR4-3600 2x8GB @ DDR4-3600 18-22-22-42
Since this is my first time reviewing a memory kit on this computer, I am only able to provide one other memory kit for comparison. However, I think the G.Skill TridentZ RGB DDR4-3600 2x8GB makes for a solid one-on-one comparison, as it shares the exact same memory capacity at 2x8GB and clock speed of 3600MHz as the XPG Spectrix D50. There is only a slight difference in latencies, which could make this comparison even more interesting and an excellent reference baseline.
Page Index
1. Introduction, Packaging, Specifications
2. A Closer Look, Test System
3. Benchmark: AIDA64 CPU
4. Benchmark: AIDA64 FPU
5. Benchmark: AIDA64 Memory
6. Benchmark: PCMark 10
7. Benchmark: 3DMark
8. Benchmark: PassMark PerformanceTest 10
9. Benchmark: SuperPI 1M, Cinebench R20
10. Overclocking and Conclusion