Senior Performance Modeling Architect, CPU Fabric and LLC

NVIDIA; United States; On-site; USD 152,000 - 288,000; Full time

Job summary

NVIDIA Corporation seeks a Performance Modeling Architect to define and improve CPU cache hierarchies and interconnects. You will build source-of-truth models that govern data movement across silicon, ensuring L3 and coherent fabrics meet ambitious performance goals.

You will model, analyze, and optimize cycle-accurate interconnects using C++/SystemC, Python for data processing, and work across automotive and data-center domains. A PhD/masters and 5+ years' experience are required.

Qualifications

• Master's or Ph.D. in Computer Engineering, Electrical Engineering, or Computer Science (or equivalent) with focus on architecture and 5+ years of experience.
• Strong understanding of CPU microarchitecture, memory consistency models, and cache coherency protocols.
• Proven experience in C++ or SystemC for cycle-accurate or functional modeling.
• Proficiency in Python or similar scripting languages for data processing, visualizations, and automation.
• Understanding of NoC topologies (Mesh, Ring, Torus), credit-based flow control, and arbitration logic.

Responsibilities

• Develop and maintain high-fidelity, cycle-accurate performance models (C++/SystemC) for coherent interconnects and large-scale shared caches.
• Model and analyze performance bottlenecks across scales from automotive SoCs to large data center architectures.
• Evaluate performance impact of coherency protocols (CHI, ACE, or proprietary) and snooping filters.
• Run and analyze industry benchmarks (SPEC, MLPerf, automotive suites) to drive architectural trade-offs.
• Collaborate with build/verification to correlate models with silicon and work with software teams to optimize drivers.

Skills

C++

SystemC

Python

CPU microarchitecture

Memory consistency models

Cache coherency protocols

NoC topologies

Education

Master's or PhD in Computer Engineering/ Electrical Engineering/ Computer Science

Job description

We are looking for a highly skilled Performance Modeling Architect to lead the architectural definition and improvement of our next‑generation CPU Cache Hierarchies and interconnects. This is an outstanding chance to create scalable solutions that connect two fast‑paced domains: the high‑reliability, low‑latency needs of Automotive and the massive efficiency, high‑density demands of Data Center systems. You will build the "source of truth" models that govern data movement across our silicon, ensuring our next‑level caches (L3/System Cache) and coherent fabrics achieve ambitious performance goals.

What you'll be doing:

• Developing and maintaining high‑fidelity, cycle‑accurate performance models (C++/SystemC) for coherent interconnects and large‑scale shared caches.
• Modeling and analyzing performance bottlenecks across varying scales, from small‑cluster automotive SoCs to massive, multi‑mesh data center architectures.
• Evaluating the performance impact of different coherency protocols (e.g., CHI, ACE, or proprietary) and snooping filters.
• Running and analyzing industry‑standard benchmarks (SPEC, MLPerf, Automotive‑specific suites) to drive architectural trade‑offs.
• Collaborating with build and verification teams to correlate performance models with silicon and working with software teams to optimize drivers for the underlying hardware topology.

What we need to see:

• A Master's or Ph.D. in Computer Engineering, Electrical Engineering, or Computer Science (or equivalent experience) with a focus on architecture with 5+ years of experience.
• Strong understanding of CPU microarchitecture, memory consistency models, and cache coherency protocols.
• Proven experience in C++ or SystemC for cycle‑accurate or functional modeling.
• Proficiency in Python or similar scripting languages for processing large datasets, generating performance visualizations, and automating simulation sweeps.
• Understanding of Network‑on‑Chip (NoC) topologies (Mesh, Ring, Torus), credit‑based flow control, and arbitration logic.

Ways to stand out from the crowd:

• Cross‑Domain Versatility: Practical experience managing the functional safety (ISO 26262) requirements of automotive chips alongside the power‑performance‑area (PPA) limitations of data center hardware.
• Hardware Performance Counters: Experience defining or using PMU (Performance Monitoring Unit) events to debug performance on real silicon or emulators.
• Formal Methods: A background in using formal verification or mathematical modeling to prove the correctness of complex coherency state machines.
• Custom Tooling: A history of building your own internal tools or frameworks to accelerate architectural exploration rather than just using off‑the‑shelf simulators.
• Advanced Memory Systems: Knowledge of emerging memory technologies like CXL (Compute Express Link) or HBM (High Bandwidth Memory) and how they collaborate with coherent fabrics.

Your base salary will be determined based on your location, experience, and other objective job‑related factors. The expected base salary ranges for this position are: 152,000 USD - 241,500 USD for Level3; and 184,000 USD - 287,500 USD for Level4.

Applications for this job will be accepted at least until November3,2026.

This posting is for an existing vacancy.

NVIDIA uses AI tools in its recruiting processes.

NVIDIA is committed to fostering an inclusive work environment and proud to be an equal opportunity employer. As we highly value diversity in our current and future employees, we do not discriminate (including in our hiring and promotion practices) on the basis of race, religion, color, national origin, gender, gender expression, sexual orientation, age, marital status, veteran status, disability status or any other characteristic protected by law.