South Korean company SK Hynix is currently engaged in an intriguing project aimed at revolutionizing the interaction between GPU and VRAM. Holding about 29% market share in the memory industry, it stands behind Samsung at approximately 40% and ahead of the U.S.-based Micron with around 26%.According to the South Korean media The JoongAng , SK Hynix has a plan in hand that seeks to catapult the performance of the next generation of graphics processors.
Please follow us on Twitter and Facebook.
NVIDIA and SK Hynix Vision: Redefining GPU-VRAM Interaction
SK Hynix’s primary goal is a radical redesign of the constant interaction between GPU and VRAM to significantly boost graphic subsystem performance. To achieve this, the company is enlisting CPU and GPU design experts, collaborating closely with NVIDIA engineers to find the optimal solution for memory-GPU communication.
The strategy devised by SK Hynix engineers resembles AMD’s 3D V-Cache technology, allowing chiplets to be stacked on top of each other rather than side by side. This stacking significantly increases level 3 cache capacity while reducing latency within the subsystem.
SK Hynix and NVIDIA’s joint strategy involves directly stacking HBM4 memory onto the GPU to optimize their communication. However, this intricate design requires a deep overhaul not only of the interface facilitating their communication but also the manufacturing process for both memory and GPU.
Collaboration and Future Implications
Eliminating the traditional intermediary interface involves GPU manufacturers like TSMC or Samsung. It’s likely that in this project, SK Hynix and NVIDIA are working closely with one of these companies responsible for manufacturing NVIDIA’s next-generation GPUs.
This breakthrough has the potential to benefit AI application chips and gaming GPUs. The HBM4 memory will utilize a 2,048-bit interface directly connected to the GPU. However, concerns about overall packaging cooling have been raised, indicating potential challenges despite the promising advancements. Eventually, distinctions between memory and chips might fade in the coming years.
The collaboration between SK Hynix and NVIDIA represents a groundbreaking effort to redefine the core interaction between GPU and VRAM, potentially reshaping the landscape of graphics processing.
Read Also:
AMD Officially Introduces Zen 4c Cores on Ryzen for Greater Efficiency
Gamer Installs 22 Fans In PC, Now Worries: ‘It’s Like a Submarine About to Implode!’
Vigor Gk41 and Gk41 Dusk: MSI Presents Two Affordably Priced Mechanical Keyboards.
