Launch of high-speed, large-capacity DDR5 DRAM for AI

Innodisk has just introduced the DDR5 6400 DRAM series with a single module capacity of up to 64GB. This product line is specifically designed to meet the needs of applications that need to process large data sources in the fields of artificial intelligence (AI), telemedicine and edge computing.

With a variety of options including CUDIMM (unbuffered, clocked dual in-line memory module), CSODIMM (cache-integrated temporary memory module) and RDIMM (RAM specifically designed for professional servers and workstations), the product line offers high speed, stability and capacity, the manufacturer said.

Specifically, the DDR5 6400 DRAM series offers speeds of up to 6400 MT/s, 14% faster than previous generations, and doubles the capacity to 64GB. Inside the product, Innodisk has integrated many technologies such as:

- Client Clock Driver (CKD): Enhances signal integrity through clock signal buffering, contributing to noise reduction and ensuring accuracy in high-speed data transmission.

- Electronic components used to control transient voltages (Transient Voltage Suppressor): Integrated protection against voltage fluctuations, ensuring maximum safety for fragile components.

- Electronic fuse protection system (RDIMM): Designed to interrupt the circuit when an overvoltage condition occurs, protecting components in the server system that require high voltage from damage.

- Wide operating temperature range: The DDR5 6400 series is optimized to operate in a temperature range of 0°C to 95°C (Tc), and will soon support an extended temperature range of -40°C to 95°C (Tc), suitable for harsh industrial environments.

- Anti-sulfation protection technology: With a thick gold plating layer, the product is protected against sulfation, ensuring durability.

The DDR5 6400 DRAM product line is designed to be compatible with Intel and AMD processors with CUDIMM, CSODIMM, ECC CUDIMM, ECC CSODIMM, and RDIMM options, with capacities from 8GB to 64GB, flexibly meeting configuration needs for different applications.

Samuel Daniel

Update 11 March 2025