NVDLA

The NVIDIA Deep Learning Accelerator (NVDLA) is an open-source hardware neural network AI accelerator created by Nvidia.[1] The accelerator is written in Verilog and is configurable and scalable to meet many different architecture needs. NVDLA is merely an accelerator and any process must be scheduled and arbitered by an outside entity such as a CPU.[2][3]

NVDLA is available for product development as part of Nvidia's Jetson Xavier NX, a small circuit board in a form factor about the size of a credit card which includes a 6-core ARMv8.2 64-bit CPU, an integrated 384-core Volta GPU with 48 Tensor Cores, and dual NVDLA "engines", as described in their own press release.[4] Nvidia claims the product will deliver 14 TOPS (tera operations per second) of compute under 10 W.[4] Applications broadly include edge computing inference engines, including object recognition for autonomous driving.

Nvidia's involvement with open hardware includes the use of RISC-V processors as part of their GPU product line-up.[5]

References

[edit]
  1. ^ Freund, Karl (March 29, 2018). "Arm Chooses NVIDIA Open-Source CNN AI Chip Technology". Forbes. Moor Insights and Strategy. Retrieved June 13, 2018.
  2. ^ "NVDLA Primer — NVDLA Documentation". nvdla.org.
  3. ^ Kwon, Hyoukjun; Pellauer, Michael; Krishna, Tushar (2018). "MAESTRO: An Open-source Infrastructure for Modeling Dataflows within Deep Learning Accelerators". arXiv:1805.02566 [cs.DC].
  4. ^ a b Franklin, Dustin (6 November 2019). "Introducing Jetson Xavier NX, the World's Smallest AI Supercomputer". developer.nvidia.com. Nvidia. Retrieved 23 September 2020.
  5. ^ Wong, William G. (5 December 2017). "The Rise of RISC-V on Display at Workshop". electronicdesign.com. Endeavor Business Media. Retrieved 23 September 2020. Nvidia gave a presentation about how its proprietary Falcon (Fast Logic CONtroller) core will be replaced by RISC-V cores. Falcon is a RISC CPU used in a number of Nvidia's chips, such as the Jetson TX2.
[edit]