A programmable parallel accelerator for learning and classification
Srihari Cadambi, Abhinandan Majumdar, Michela Becchi et al. · International Conference on Parallel Architectures and Compilation Techniques · 2010
For learning and classification workloads that operate on large amounts of unstructured data with stringent performance constraints, general purpose processor performance scales poorly with data size. In this paper, we present a programmable accelerator for this workload domain. To architect the accelerator, we profile five representative workloads, and find that their computationally intensive portions can be formulated as matrix or vector operations generating large amounts of intermediate data, which are then reduced by a secondary operation such as array ranking, finding max/min and aggregation. The proposed accelerator, called MAPLE, has hundreds of simple processing elements (PEs) laid out in a two-dimensional grid, with two key features. First, it uses in-memory processing where on-chip memory blocks perform the secondary reduction operations. By doing so, the intermediate data are dynamically processed and never stored or sent off-chip. Second, MAPLE uses banked off-chip memory, and organizes its PEs into independent groups each with its own off-chip memory bank. These two features together allow MAPLE to scale its performance with data size. This paper describes the MAPLE architecture, explores its design space with a simulator, and illustrates how to automatically map application kernels to the hardware. We also implement a 512-PE FPGA prototype of MAPLE and find that it is 1.5-10x faster than a 2.5 GHz quad-core Xeon processor despite running at a modest 125 MHz.
Cited by 96 works. doi:10.1145/1854273.1854309
Authors
References
- Gradient-based learning applied to document recognition
- Some methods for classification and analysis of multivariate observations
- A unified architecture for natural language processing
- Least squares quantization in PCM
- Fast Training of Support Vector Machines Using Sequential Minimal Optimization
- Large-scale deep unsupervised learning using graphics processors
- Face recognition: a convolutional neural-network approach
- A Survey of General‐Purpose Computation on Graphics Hardware
- Image retrieval
- High Performance Convolutional Neural Networks for Document Processing