Nvidia corporation (20250125819). ENERGY-EFFICIENT DATAPATH FOR VECTOR-SCALED HIERARCHICAL CODEBOOK QUANTIZATION
ENERGY-EFFICIENT DATAPATH FOR VECTOR-SCALED HIERARCHICAL CODEBOOK QUANTIZATION
Organization Name
Inventor(s)
Rangharajan Venkatesan of San Jose CA US
Reena Elangovan of Milpitas CA US
Brucek Kurdo Khailany of Austin TX US
Brian Matthew Zimmer of Sunnyvale CA US
ENERGY-EFFICIENT DATAPATH FOR VECTOR-SCALED HIERARCHICAL CODEBOOK QUANTIZATION
This abstract first appeared for US patent application 20250125819 titled 'ENERGY-EFFICIENT DATAPATH FOR VECTOR-SCALED HIERARCHICAL CODEBOOK QUANTIZATION
Original Abstract Submitted
vector-scaled hierarchical codebook quantization reduces precision (bitwidth) vectors of parameters and may enable energy-efficient acceleration of deep neural networks. a vector (block array) comprises one or more parameters within a single dimension of a multi-dimensional tensor (or kernel). for example, block array comprises 4 sub-vectors (blocks) and each sub-vector comprises 8 parameters. the parameters may be represented in integer, floating-point, or any other suitable format. a vector cluster quantization technique is used to quantize blocks of parameters in real-time. hardware circuitry within a datapath identifies an optimal codebook of a plurality of codebooks for quantizing each block of parameters and the block is encoded using the identified codebook. during processing, the identified codebook is used to obtain the quantized parameter and perform computations at the reduced precision.