In this article
Services that create and evaluate representations of an ASIC before or during silicon development. Options include virtual models, FPGA prototypes, emulation, MPW silicon and laboratory bring-up.
Choose a prototype for a question
An ASIC prototype is useful when it provides evidence about a specific uncertainty. Simulation, emulation, FPGA platforms and prototype silicon answer different questions. A software team may need realistic transaction sequences and long execution runs, while an analog team may need fabricated devices to characterize electrical behavior. Begin by listing what the prototype must demonstrate and what it cannot establish. This keeps a successful demonstration from being mistaken for complete verification and helps the team choose a platform that matches the decision at hand.
FPGA and emulation platforms
Programmable platforms can exercise large amounts of behavior and support hardware-software development before ASIC silicon arrives. Their mapping may change clocking, memory structures or execution speed. Multi-device partitioning can introduce additional constraints. Preserve a record of those differences and use them when interpreting results. A software defect found on the prototype can be highly valuable, but the platform does not prove final ASIC timing or analog performance. Design debug visibility deliberately so the prototype can reveal causes rather than merely reproduce a failure.
Prototype silicon and MPW
A multi-project wafer run can provide fabricated samples while sharing certain manufacturing resources with other designs. Availability, design rules, area limits and delivery terms depend on the selected service. Prototype silicon is particularly useful for evaluating process-dependent circuits, interfaces and complete-device behavior. It still requires a test and bring-up plan. Decide how the package, board, instruments and firmware will be ready when samples arrive. A fabricated die without suitable access or measurement infrastructure can leave important learning delayed despite a successful manufacturing step.
Use results to reduce uncertainty
Define pass criteria and preserve the test configuration, software revision and input data. When the prototype differs from expectations, classify the issue before changing the design: it may lie in the implementation, model, test setup or specification. Feed verified findings into the source design and regression tests. A prototype program should leave a clearer product specification and risk picture. Its value is measured by the decisions it enables and defects it helps resolve, rather than by the existence of a demonstration board alone.
Questions about prototyping
Is prototype silicon necessary for every early design decision?
No. Many architectural and functional questions can be explored earlier with models or programmable platforms. Fabrication is most useful when it addresses uncertainties those methods cannot resolve adequately.
Does a prototype replace production qualification?
No. Prototype learning supports development, while production readiness requires appropriate characterization, repeatability, test coverage and the release evidence defined for the product and application.
Requirements that can be checked
Turn each important behavior into a requirement with an observable result. Describe valid inputs, invalid inputs, ordering rules, throughput expectations and the response to interrupted transactions. Identify which clocks and resets are involved. A good review asks whether another engineer could write an independent test from the description. Where the answer is unclear, improve the requirement before adding more implementation detail. Keep assumptions about the surrounding system visible, including how long an interface can stall and whether input data can arrive while a block is unavailable. These details determine whether the implementation and its test environment are solving the same problem.
Verification across normal and unusual operation
Check ordinary workloads alongside stressful combinations of events. Reset during activity, a full queue, simultaneous requests and an error response can expose behavior that an ideal sequence misses. Use appropriate simulation, assertions or formal methods for the properties being examined. Preserve reproducible inputs and expected results. If a check is excluded, record why and who reviewed the decision. Coverage helps organize evidence, but its interpretation depends on what the tests actually check. A regression that executes many cycles without observing meaningful outcomes can provide less confidence than a smaller set of well-targeted checks tied to requirements.
Implementation handover
The source design needs constraints, technology assumptions and integration context before a receiving team can use it reliably. Include the intended operating modes, clock relationships, IP versions and any known limitations. Check that source files, models and reports describe the same configuration. Define how changes are communicated while implementation is in progress. When physical results challenge an assumption, route the finding back to the responsible architecture or logic owner. A controlled handover keeps the project from optimizing an obsolete version or interpreting an undocumented exception as an intentional requirement.
Evidence for the next decision
Separate predicted results from measured behavior and state the conditions behind both. An early synthesis estimate can support architecture exploration, but it is not equivalent to a final implemented result. A prototype can support software integration without proving final ASIC timing. Before a release, identify the evidence required for the particular stage and list unresolved issues with owners. Preserve the approved revision and enough context to reproduce the relevant checks. This makes later debugging, engineering changes and product maintenance more efficient than relying on a single successful run whose inputs are no longer clear.
Primary reference for further reading
A practical cross-discipline review
Before committing the next stage, bring the design, verification and product-engineering owners together around the same revision. Confirm the assumptions that pass between them: operating modes, interfaces, test access and the evidence expected at release. List unresolved issues with a responsible owner and a concrete way to close them. An open question is easier to manage when its consequence is understood than when it is hidden inside a general progress estimate. Preserve the decision record so a later change can be evaluated against the original reasoning. This review is useful even for a narrow specialist task, because the surrounding product still determines whether the delivered result is usable.
Include the receiving team's acceptance question in the review: what must be demonstrated before this result can be used safely in the next engineering stage? Agree how that demonstration will be recorded and how a subsequent change triggers a focused reassessment of the affected evidence.
Explore the related Avnet offering or discuss your project requirements.
