DFKI GmbH
Evidence profile
Within the supplied evidence, DFKI GmbH appears through the affiliation “Cyber-Physical Systems, DFKI GmbH, 28359 Bremen, Germany” in technical papers on Instruction Set Simulator and processor verification.[C1][C3]
In the 2019 paper Verifying Instruction Set Simulators using Coverage-guided Fuzzing, the DFKI affiliation is assigned to Daniel Große and Rolf Drechsler; Vladimir Herdt and Hoang M. Le are listed with the Institute of Computer Science, University of Bremen only in that paper.[C1] In the 2022 paper Cross-Level Processor Verification via Endless Randomized Instruction Stream Generation with Coverage-guided Aging, the DFKI affiliation is assigned to Vladimir Herdt and Rolf Drechsler; Niklas Bruns is listed with the University of Bremen and Eyck Jentzsch with MINRES Technologies GmbH.[C3]
Instruction Set Simulator verification
The related paper Verifying Instruction Set Simulators using Coverage-guided Fuzzing frames verification of Instruction Set Simulators (ISSs) as crucial because ISSs serve as executable specifications in processor design flows, while simulation-based ISS verification requires comprehensive test sets.[C2]
The paper proposes a coverage-guided fuzzing approach for ISS verification. In addition to code coverage, it integrates functional coverage and a custom mutation procedure tailored to ISS verification. As a case study, the approach is applied to three publicly available RISC-V ISSs and reports finding several new errors, including one in the official RISC-V reference simulator Spike.[C2]
Cross-level RTL processor verification
The 2022 paper Cross-Level Processor Verification via Endless Randomized Instruction Stream Generation with Coverage-guided Aging proposes a cross-level verification approach for Register-Transfer Level processor verification. Its foundation is a randomized coverage-guided instruction-stream generator that produces one endless, unrestricted instruction stream that evolves dynamically at runtime.[C4]
The same paper uses an Instruction Set Simulator as a reference model in a tight co-simulation setting. It continuously updates coverage information based on the ISS execution state and employs Coverage-guided Aging to smooth the coverage distribution over time. The abstract states that a case study with an industrial pipelined 32-bit RISC-V processor demonstrates the approach’s effectiveness.[C4]
Related people and work in the evidence
- Verifying Instruction Set Simulators using Coverage-guided Fuzzing — related ISS-verification paper connected to DFKI GmbH through the listed DFKI affiliations of Daniel Große and Rolf Drechsler.[C1][C2]
- Daniel Große — listed with the Cyber-Physical Systems, DFKI GmbH affiliation in the 2019 ISS fuzzing paper.[C1]
- Rolf Drechsler — listed with the Cyber-Physical Systems, DFKI GmbH affiliation in both supplied papers.[C1][C3]
- Vladimir Herdt — listed with the Cyber-Physical Systems, DFKI GmbH affiliation in the 2022 cross-level processor-verification paper.[C3]