This latest offering has an increased (16GByte) memory and ECC memory support. The fourth generation Tensor core increases AI throughput and delivers structured sparsity and FP8 precision for higher inference performance for AI-accelerated tools and applications, says Nvidia.
The third generation RT cores in the GPU accelerate ray tracing and the CUDA cores used deliver up to 1.5 times the FP32 output of the earlier generation to improve graphics and compute performance.
The GPU operates at 70W, in common with the earlier generation, but doubling the graphics, rendering, AI and compute workload performance. For AI-powered graphics, real time rendering is enhanced with the use of DLSS3 which generates additional frames. There is also an eighth generation encoder (NVEC) with AV1 support which is claimed to be 40% more efficient than H.264 and an increase by up to 300% for virtual reality workflows, compared to the previous generation, said the company.
The RTX Enterprise driver is available to support the RTX 2000 Ada GPU.
The NVIDIA RTX 2000 Ada is available now through distribution partners, e.g., Arrow Electronics, Ingram Micro, Leadtek, PNY, Ryoyo Electro and TD Synnex, and will be available from Dell Technologies, HP and Lenovo from April.
