CPU Cache Design Engineer
The job description
Tech stack. Cache controller design, SystemVerilog, coherency protocols, SRAM macro integration, PPA analysis, Python, verification of concurrent protocols, timing closure
About the role
You will design CPU cache controllers and hierarchies at a semiconductor company, implementing the tag arrays, data arrays, replacement logic, and coherency engines that keep the processor fed with data. Cache designers work on some of the most concurrency-intensive logic in the chip: multiple outstanding requestors, in-flight transactions, snoop handling, and race conditions that must resolve correctly under every possible interleaving. Your designs must be simultaneously high-performance, area-efficient, and provably correct, since a cache bug can corrupt data silently.
What you will achieve
- Design cache controllers for L1, L2, or last-level caches including tag and data pipelines, miss status handling registers, fill and eviction logic, and snoop response paths.
- Implement coherency protocol engines supporting the program's chosen protocol, handling all transient states, race conditions, and deadlock scenarios with rigorous attention to correctness.
- Optimize cache access latency and throughput through pipeline design, banking strategies, and arbitration policies, meeting the aggressive timing targets these latency-critical structures demand.
- Integrate SRAM macros including redundancy and repair schemes, ECC protection, and timing closure of array interfaces in partnership with memory compiler and physical design teams.
- Drive verification of your cache designs to the quality required for first-silicon success, with coverage models that genuinely reflect protocol behavior, concurrency corner cases, and error scenarios.
- Support post-silicon validation of cache performance and correctness, diagnosing subtle coherency or performance anomalies with systematic measurement and analysis.
What you will bring
Must-haves
- 5 to 8 years of experience in cache, memory subsystem, or high-performance control logic design with shipped silicon.
- Deep understanding of cache organization, coherency protocols such as MESI variants, and the concurrency challenges inherent in cache controller design.
- Strong SystemVerilog design skills with demonstrated experience on complex, timing-critical control logic.
- Knowledge of SRAM characteristics including access timing, power behavior, ECC, and redundancy, and how they constrain controller design.
- Rigorous verification mindset appropriate for logic where a single escaped bug can cause silent data corruption in customer systems.
- Timing closure experience on high-frequency cache pipelines with tight array-interface budgets.
- Patience and precision for the detailed protocol work this domain demands, combined with the paranoia that keeps products correct.
Nice-to-haves
- Experience with formal verification of coherency protocols or cache controllers.
- Familiarity with advanced cache features such as way partitioning, QoS enforcement, or data compression.
- Knowledge of multi-level cache hierarchy performance tuning and modeling.
- Exposure to security aspects of caches including side-channel attack mitigations.
Apple
NVIDIA
Intel
AMD
Arm
SiFive