Verifying RISC-V Platforms for Space

Breker Verification Systems and the Case for Bullet-Proof Verification

Categories: AI, Breker Verification Systems, EDA, RISC-V

The New Space Economy Demands Absolute Reliability

The space industry is undergoing a dramatic transformation. Launch costs to Low Earth Orbit (LEO) have fallen from nearly $20,000/kg to around $2,000/kg, driven by commercial competition led by players such as SpaceX, and costs are projected to fall even further. This shift has unlocked entirely new opportunities—most notably large-scale satellite constellations enabling global SATCOM, broadband, and IoT connectivity from providers such as Starlink, Amazon’s LEO initiative, and similar efforts underway in China.

Standardization efforts through 3GPP, already visible in 5G-Advanced and expected to mature in 6G, are accelerating interoperability. This makes direct-to-smartphone satellite communication a realistic, mainstream capability rather than a niche emergency service. Beyond connectivity, space-based systems are becoming increasingly critical for defense, climate monitoring, disaster response, and scientific research.

As this new space-based economy expands, so does the demand for electronics that can operate flawlessly for years in an environment where repair is impossible.

Space-Ready Is More Than Radiation Hardening

Radiation exposure in space—whether from solar events or deep-space cosmic rays—poses severe challenges for electronics. Radiation-hardened processes, redundancy, error correction codes (ECC), and fault-tolerant architectures are essential. However, rad-hard alone is not sufficient.

Unlike terrestrial or automotive systems, satellites cannot be serviced. If a failure cannot be recovered remotely through a reset or reconfiguration, the mission is effectively over. This reality places extraordinary emphasis on verification quality during design. A level of confidence acceptable for consumer electronics—or even automotive systems—is simply inadequate for space.

The Challenge of Comprehensive Specification Verification

Modern processors are defined by extremely complex specifications. The RISC-V ISA, for example, spans roughly 1,400 pages, covering a wide range of behaviors, interactions, and corner cases. Verifying compliance is not just about validating custom extensions or isolated features; it requires ensuring correct behavior across interacting subsystems, under real-world contention, latency, and stall conditions.

While unit tests for features such as ECC or redundancy are manageable, the real difficulty lies in cross-verification—ensuring that multiple behaviors interact correctly when exercised simultaneously. These are precisely the conditions under which the most dangerous bugs hide, often surfacing only after billions of cycles in operation.

Breker’s System-Level Verification Approach

Breker Verification Systems addresses this challenge by shifting verification from isolated testing to system-level modeling. Their methodology begins with abstract, implementation-independent test models that can be composed and combined to reflect realistic system behavior.

Breker provides a range of Verification IP (VIP), including RISC-V tests, coherency compliance checks, and security validation. These models can be extended using the PSS standard or conventional C++, enabling randomized scenarios—particularly valuable for testing custom ISA extensions. By running multiple models concurrently, engineers can generate high-stress, interwoven traffic patterns that expose subtle interactions and corner-case failures.

When the Specification Itself Becomes a Risk

Verification ultimately measures compliance against the specification—but translating a linear, human-written document into a complete and correct set of tests is inherently fallible. Even highly refined specifications like RISC-V can contain distributed dependencies that are easy to miss.

A notable example involves fence instructions, which control memory ordering between cores. While their behavior is clearly defined early in the specification, additional critical details appear later in unrelated sections. Missing such references can lead to coherence bugs that surface only in extreme scenarios—exactly the kind that can doom a satellite mission.

Breker identified such an issue in practice, helping a customer avoid a costly redesign. This highlights the need not just for rigorous verification, but for deeper understanding of specification intent across its entire structure.

AI as a Second Line of Defense for Specifications

Specifications are living documents, and perfect clarity is unrealistic. Breker is exploring the use of AI-based natural language processing (NLP)—rather than large language models—to analyze specifications and uncover hidden relationships and dependencies. Early results have already demonstrated success in identifying distributed requirements, such as those related to fencing behavior in RISC-V.

This approach offers a promising second step in closing the verification loop:

1. Ensure the specification itself is authoritative and precise.

2. Ensure all interdependencies within that specification are fully understood and reflected in verification models.

Conclusion

As space systems grow more capable and more complex, verification quality becomes mission-critical. RISC-V’s openness and flexibility make it attractive for space applications, but they also demand verification approaches that go far beyond traditional testing.

Breker Verification Systems’ system-level modeling, combined with emerging AI-assisted specification analysis, represents a compelling path toward the bullet-proof verification required for space-ready RISC-V platforms. In an environment where failure is not an option, this level of rigor is not just beneficial—it is essential.