Pipeline Stage Resolved Timing Characterization of FPGA and ASIC Implementations of a RISC V Processor
Fuente:
arXiv
Saved in:
| Main Author: | |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909966039777280 |
|---|---|
| author | Darvishi, Mostafa |
| author_facet | Darvishi, Mostafa |
| contents | This paper presents a pipeline stage resolved timing characterization of a 32-bit RISC V processor implemented on a 20 nm FPGA and a 7 nm FinFET ASIC platform. A unified analysis framework is introduced that decomposes timing paths into logic, routing, and clocking components and maps them to well-defined pipeline stage transitions. This approach enables systematic comparison of timing behavior across heterogeneous implementation technologies at a microarchitectural level. Using static timing analysis and statistical characterization, the study shows that although both implementations exhibit dominant critical paths in the EX to MEM pipeline transition, their underlying timing mechanisms differ fundamentally. FPGA timing is dominated by routing parasitics and placement dependent variability, resulting in wide slack distributions and sensitivity to routing topology. In contrast, ASIC timing is governed primarily by combinational logic depth and predictable parametric variation across process, voltage, and temperature corners, yielding narrow and stable timing distributions. The results provide quantitative insight into the structural origins of timing divergence between programmable and custom fabrics and demonstrate the effectiveness of pipeline stage resolved analysis for identifying platform specific bottlenecks. Based on these findings, the paper derives design implications for achieving predictable timing closure in processor architectures targeting both FPGA and ASIC implementations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_13866 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Pipeline Stage Resolved Timing Characterization of FPGA and ASIC Implementations of a RISC V Processor Darvishi, Mostafa Signal Processing Hardware Architecture This paper presents a pipeline stage resolved timing characterization of a 32-bit RISC V processor implemented on a 20 nm FPGA and a 7 nm FinFET ASIC platform. A unified analysis framework is introduced that decomposes timing paths into logic, routing, and clocking components and maps them to well-defined pipeline stage transitions. This approach enables systematic comparison of timing behavior across heterogeneous implementation technologies at a microarchitectural level. Using static timing analysis and statistical characterization, the study shows that although both implementations exhibit dominant critical paths in the EX to MEM pipeline transition, their underlying timing mechanisms differ fundamentally. FPGA timing is dominated by routing parasitics and placement dependent variability, resulting in wide slack distributions and sensitivity to routing topology. In contrast, ASIC timing is governed primarily by combinational logic depth and predictable parametric variation across process, voltage, and temperature corners, yielding narrow and stable timing distributions. The results provide quantitative insight into the structural origins of timing divergence between programmable and custom fabrics and demonstrate the effectiveness of pipeline stage resolved analysis for identifying platform specific bottlenecks. Based on these findings, the paper derives design implications for achieving predictable timing closure in processor architectures targeting both FPGA and ASIC implementations. |
| title | Pipeline Stage Resolved Timing Characterization of FPGA and ASIC Implementations of a RISC V Processor |
| topic | Signal Processing Hardware Architecture |
| url | https://arxiv.org/abs/2512.13866 |