GSI Technology's APU Matches GPU Performance with 98% Lower Energy Consumption

GSI Technology announced a paper by Cornell University researchers confirming that their Associative Processing Unit (APU) CIM architecture achieves GPU-level performance in AI applications while reducing energy consumption by 98%. The Gemini-I APU matched NVIDIA's A6000 GPU in throughput on RAG workloads and consumed significantly less energy. The APU was also found to be faster and more efficient than CPUs.