DSPIMM: A Fully Digital SParse In-Memory Matrix Vector Multiplier for Communication Applications.
Amitesh SridharanFan ZhangYang SuiBo YuanDeliang FanPublished in: DAC (2023)
Keyphrases
- sparse matrix
- floating point
- coefficient matrix
- communication networks
- memory usage
- communication systems
- interprocess communication
- low rank matrices
- memory requirements
- eigenvalues and eigenvectors
- random projections
- rows and columns
- low rank matrix
- communication channels
- low rank
- eigenvalue decomposition
- transformation matrix
- sparse data
- state information
- high dimensional
- sparse coding
- information sharing
- interior point methods
- weight matrix
- linear algebra
- hardware implementation
- feature vectors