Page 2 - A Closer Look, Test System
With a name like "FURY Beast", you may be expecting some intense RGB lighting or at least some flashy styling in its heatspreader. However, what we have here with the Kingston FURY Beast DDR5-5200 2x16GB is a pair of clean sticks of DDR5 memory. Its low profile heatspreaders are fully black with white text and a silver emblazoned FURY lettering. The Kingston and "DDR5" markings are found at the top of the black aluminum. This metal is lightweight and serves as a decent heat conductor. The FURY Beast is approximately 4mm taller than modules with no heatspreader at all, which should fit under practically any heatsink with a side-mounted fan. While having extra aluminum on the side of memory may or may not necessarily affect performance, it is almost impossible to find performance memory without any form of a heatspreader attached for quite a while now. They do definitely serve a purpose in dissipating heat while also having the added benefit of making the memory look a bit nicer to look at through a side panel window on your case.
The heatspreader design of the Kingston FURY Beast modules is asymmetrical when looked at straight on, but mostly symmetrical between the sides. There are two locking clips that ensure the heatspreaders stay together when in place over the memory module. Otherwise, one side shows off the aforementioned FURY logo with the "Beast" product name, while the other side has a specifications sticker. It lists the model number (KF552C40BBK2-32), voltage, and some certifications label. It also shows the assembly location, which is Taiwan.
From above, you can see the Kingston FURY Beast DDR5-5200 really does have a low-profile aluminum heatspreader, as it barely reaches above the rest of the memory module. The two aluminum sides are held together with strips of thermally conductive adhesive and are also interlocked at the top for more security. The adhesive force between the two heatspreaders and memory ICs is quite strong, so I would recommend a good heat source to warm up the adhesive strips and make it easier to pry off the aluminum pieces. You can also see how the aluminum pieces are designed with them being mirror images of each other. Since the pieces are made from aluminum, it does not hold a lot of heat, therefore dissipating the heat energy relatively quickly into the surrounding environment. These pieces are pretty thin, but they still feel solid and should not easily bend. Either way, you will probably never remove the heatspreaders as most aftermarket CPU heatsinks should easily accommodate memory modules of this height profile.
Looking at the black PCB itself, you can see the primary difference of DDR5 memory is the inclusion of a power management integrated circuit, or PMIC, in the middle of the module. DDR4 RAM power management is handled by the motherboard, however DDR5 has the PMIC directly on each stick of RAM. Kingston has placed a large blue thermal pad on top to ensure this area makes contact with the aluminum piece.
Taking a closer look at the memory chips on the Kingston FURY Beast DDR5-5200 2x16GB dual channel memory kit shows the eight Micron branded chips. Each are marked with the short code "D8BNJ", which is short for the full part number "MT60B2G8HB-48B:A". Each chip is 2GB in capacity and located on one side to make for a total of 16GB on each DIMM. As mentioned previously, this RAM kit runs at a frequency of DDR5-5200 with 40-40-40-80 latencies. These latencies are notably higher than that of DDR4 memory, but this is expected for DDR5. This module operates at a stock voltage of 1.25V, which is under the maximum safe limit of 1.35V set by Intel and AMD. Here are the listed features for the ICs, as obtained from Micron:
• VDD = VDDQ = 1.1V (nom)
• VPP= 1.8V (nom)
• On-die, internal, adjustable VREF generation for DQ, CA, CS
• 1.1V pseudo open-drain I/O
• TC maximum up to 95°C
– 32ms, 8192-cycle refresh up to 85°C
– 16ms, 8192-cycle refresh at >85°C to 95°C
• 32 internal banks (x4, x8): 8 groups of 4 banks each
• 16 internal banks (x16): 4 groups of 4 banks each
• 16n-bit prefetch architecture
• 1 cycle/2 cycle command structure
• 2N mode
• All bank and same bank refresh
• Multi-purpose command (MPC)
• CS/CA training mode
• On-die ECC
• ECC transparency and error scrub
• Decision feedback equalization (DFE)
• Loopback mode
• Command-based non-target (NT) nominal, DQ/DQS park, and dynamic WR on-die termination (ODT)
• sPPR and hPPR capability
• JEDEC JESD-79.5 compliant
Our test configuration is as follows:
CPU: Intel Core i5 12600K
CPU Cooling: Noctua NH-U12A chromax.black
Motherboard: ASUS ProArt Z690-Creator WiFi
Graphics: EVGA GeForce RTX 3070 Ti XC3 ULTRA GAMING
Chassis: Thermaltake Core P6 TG Snow
Storage: XPG Atom 30 1TB
Power: FSP Hydro PTM Pro 1200W
Operating System: Microsoft Windows 11 Pro
Compared Hardware:
- Kingston FURY Beast DDR5-5200 2x16GB @ DDR5-5200 40-40-40-80
- Patriot Viper Venom RGB DDR5-6200 2x16GB @ DDR5-6200 40-40-40-76
- XPG Lancer RGB DDR5-6000 2x16GB @ DDR5-6000 40-40-40-76
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 R23
10. Overclocking and Conclusion
