Back
LLM and ViT Acceleration on Quadric GPNPU
HPC
AI & ML
Staff Aug
The ACCELR team has expertise in AI inference acceleration on truly exotic compute. This is demonstrated by our contribution to the Transformer Kernel library developed for the highly parallel Quadric Chimera GPNPU. Talk to our team for more information on our AI acceleration experience.
Inference acceleration of LLM and ViT networks
As members of Quadric's extended application development team, our engineers were able to contribute to accelerating the LLM and ViT transformer networks on the Quadric Chimera GPNPU
Optimized to maximize performance
Achived by was of grasping the intricate details of the hardware design as well as by gaining knowledge of transformer network internals
