Name
Title | ||
|---|---|---|
Handling large data sets for high-performance embedded applications in heterogeneous systems-on-chip |
Abstract | ||
|---|---|---|
Local memory is a key factor for the performance of accelerators in SoCs. Despite technology scaling, the gap between on-chip storage and memory footprint of embedded applications keeps widening. We present a solution to preserve the speedup of accelerators when scaling from small to large data sets. Combining specialized DMA and address translation with a software layer in Linux, our design is transparent to user applications and broadly applicable to any class of SoCs hosting high-throughput accelerators. We demonstrate the robustness of our design across many heterogeneous workload scenarios and memory allocation policies with FPGA-based SoC prototypes featuring twelve concurrent accelerators accessing up to 768MB out of 1GB-addressable DRAM.
|
Year | DOI | Venue |
|---|---|---|
2016 | 10.1145/2968455.2968509 | ESWEEK'16: TWELFTH EMBEDDED SYSTEM WEEK
Pittsburgh
Pennsylvania
October, 2016 |
Keywords | Field | DocType |
high-performance embedded applications,large data set handling,heterogeneous systems-on-chip,local memory,accelerator performance,technology scaling,on-chip storage,memory footprint,DMA,software layer,Linux,high-throughput accelerators,heterogeneous workload scenarios,memory allocation policies,FPGA-based SoC prototypes,concurrent accelerators,DRAM | Dram,Computer architecture,System on a chip,Computer science,Parallel computing,Field-programmable gate array,Robustness (computer science),Memory management,Software,Memory footprint,Speedup,Embedded system | Conference |
ISBN | Citations | PageRank |
978-1-4503-4482-1 | 2 | 0.38 |
References | Authors | |
28 | 5 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Paolo Mantovani | 1 | 106 | 10.58 |
| Emilio G. Cota | 2 | 53 | 4.10 |
| Christian Pilato | 3 | 329 | 32.19 |
| Giuseppe Di Guglielmo | 4 | 107 | 15.57 |
| Luca P. Carloni | 5 | 1713 | 120.17 |