Uppsats
Continuing Towards a Trustworthy Stack : Formal Verification of Correctness for Data Forwarding in an In-Order Pipelined Processor Circuit for a RISC ISA
Master-uppsats
KTH/Skolan för elektroteknik och datavetenskap (EECS)
Publicerad: 2024
Språk: Engelska
Nyckelord
klicka för att sökaSammanfattning
To increase the performance, modern processor pipelines heavily rely on data forwarding, a technique to avoid pipeline stalls by adding additional data paths that route computed values to future instructions elsewhere in the pipeline before the instruction that produced that value has been fully processed. However, this adds complexity, making it harder to trust the correctness of such pipelined circuits. This thesis presents an approach to improve the performance of an in-order pipelined RISC processor architecture using data forwarding with guaranteed functional correctness down to its circuit-level implementation. We show that the processor executes program instructions correctly according to their specification, by proving a refinement relation between the execution traces of the processor circuit and the abstract behaviour of the instruction set architecture. The proof is mechanised using the HOL4 interactive theorem prover. The formally verified circuit has been successfully synthesised as a small computer system, integrating other unverified components like a cache and interrupt handler, for the PYNQ-Z1 FPGA board. The benchmarks indicate that the presented processor circuit outperforms the original processor. To the best of our knowledge, this thesis presents the fastest open-source processor with guaranteed functional correctness down to the circuit level.
Information
- Författare
- Näf, Luca
- Lärosäte / institution
- KTH/Skolan för elektroteknik och datavetenskap (EECS)
- Publiceringsdatum
- 2024
- Uppsatstyp
- Master-uppsats
- Språk
- Engelska
Utforska vidare
Liknande uppsatser
Uppsatser med liknande ämnen och nyckelord.
Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)
Vanhainen, Erik
Publicerad: 2024
Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)
Marie Schmidt, Jule
Publicerad: 2024
Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)
Sevenhuijsen, Merlijn
Publicerad: 2024
Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)
Granqvist, Lukas
Publicerad: 2025
Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)
Manjikian, Hovig
Publicerad: 2023
Master-uppsats, KTH/Skolan för elektroteknik och datavetenskap (EECS)
Söderberg, Oskar
Publicerad: 2023