Reconfigurable tiles of computing-in-memory SRAM architecture for scalable vectorization
| Autor: | Subhasish Mitra, Maha Kooli, J.-P. Noel, R. Gauchi, Bastien Giraud, V. Egloff, Henri-Pierre Charles, Pascal Vivet |
|---|---|
| Přispěvatelé: | Commissariat à l'énergie atomique et aux énergies alternatives (CEA), Université Grenoble Alpes (UGA), Stanford University |
| Jazyk: | angličtina |
| Rok vydání: | 2020 |
| Předmět: |
[INFO.INFO-AR]Computer Science [cs]/Hardware Architecture [cs.AR]
Computer science SystemC/TLM 020208 electrical & electronic engineering Instruction Set Simulator 02 engineering and technology Parallel computing SRAM SIMD 020202 computer hardware & architecture Computing-In-Memory Reduction (complexity) Instruction set simulator SystemC Scalability Vectorization (mathematics) 0202 electrical engineering electronic engineering information engineering Near-Memory Computing [INFO]Computer Science [cs] Static random-access memory computer AND gate PolyBench computer.programming_language |
| Zdroj: | ISLPED 2020: ACM/IEEE International Symposium on Low Power Electronics and Design ISLPED 2020: ACM/IEEE International Symposium on Low Power Electronics and Design, Aug 2020, Boston, MA, United States. pp.121-126, ⟨10.1145/3370748.3406550⟩ ISLPED |
| Popis: | International audience; For big data applications, bringing computation to the memory is expected to reduce drastically data transfers, which can be done using recent concepts of Computing-In-Memory (CIM). To address kernels with larger memory data sets, we propose a reconfigurable tile-based architecture composed of Computational-SRAM (C-SRAM) tiles, each enabling arithmetic and logic operations within the memory. The proposed horizontal scalability and vertical data communication are combined to select the optimal vector width for maximum performance. These schemes allow to use vector-based kernels available on existing SIMD engines onto the targeted CIM architecture. For architecture exploration, we propose an instruction-accurate simulation platform using SystemC/TLM to quantify performance and energy of various kernels. For detailed performance evaluation, the platform is calibrated with data extracted from the Place&Route C-SRAM circuit, designed in 22nm FDSOI technology. Compared to 512-bit SIMD architecture, the proposed CIM architecture achieves an EDP reduction up to 60× and 34× for memory bound kernels and for compute bound kernels, respectively. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|