National

ETV Bharat / technology

Samsung Unveils World's First 24Gb GDDR7 DRAM For Next-Gen AI Computing

Samsung has announced the world's first 24Gb GDDR7 DRAM, offering unprecedented speed and efficiency for next-generation AI computing and high-performance applications.

By ETV Bharat Tech Team

Published : 5 hours ago

Samsung's new GDDR7 offers industry-leading capacity
Samsung's new GDDR7 offers industry-leading capacity (Samsung)

Hyderabad:Samsung has announced the development of world's first 24Gb (gigabit) GDDR7 DRAM for next-generation AI computing. In addition to industry's highest capacity, the new GDDR7 from the South Korean tech giant claims to feature the fastest speed.

In a newsroom post, Samsung says that the validation for the 24Gb GDDR7 in next-gen AI computing systems from major GPU customers will begin this year. The company plans to start commercialisation early next year.

Samsung says that the 24Gb GDDR7 will be utilised in fields that require high-performance memory solutions, such as data centres and AI workstations. Additionally, it will also be used in traditional applications, including graphics DRAM in graphics cards, gaming consoles, and autonomous driving.

Notably, Samsung developed the industry's first 16Gb GDDR7 last year. The new 24Gb GDDR7 maintains the same size as its predecessor and utilises 5th-gen 10nm-class DRAM for a 50 per cent increase in cell density.

The 24Gb GDDR7 uses 3-level Pulse-Amplitude Modulation (PAM3) signaling to achieve a reported industry-leading speed for graphics DRAM of 40 Gbps (gigabits-per-second), which translates to a 25 per cent improvement over the previous version. Based on the usage environment, Samsung says that the GDDR7's performance can be further enhanced to up to 42.5Gbps.

It also claims to enhance power efficiency by applying the same technologies used in mobile products to graphics DRAM for the first time. Samsung claims to have reduced unnecessary power consumption to achieve an improvement of over 30 per cent in power efficiency by implementing methods like clock control management and dual VDD design.

For All Latest Updates

ABOUT THE AUTHOR

...view details