Memory Sub-system Performance Modelling Researcher
- Leuven, Belgium
- 07 Mar 2023
- End of advertisement period
- 28 Apr 2023
- Academic Discipline
- Engineering & Technology, Computer Science, Electrical & Electronic Engineering, General Engineering
- Job Type
- Research Related, Research Support
- Contract Type
- Full Time
Memory Sub-system infrastructure simulation using emerging technology: system technology co-optimization (STCO)What you will do
As a performance modelling and simulation engineer, you interact closely with device experts, technology experts, circuit as well as major foundry, fabless and EDA partners in imec s eco-system. This is an exciting opportunity to help drive future system architectures based on imec s emerging logic and memory technology. In the more than Moore era, memory wall bottlenecks pose imposing challenges to system performance for emerging AI applications. A memory sub-system simulation infrastructure will help model and benchmark the components for future systems, adhering to our System Technology Co-Optimization (STCO) roadmap. We analyze application driven, domain-specific architectures, their emerging usage models and benchmark novel memory technologies in conjunction with our design technology co-optimization (DTCO) research activities.
A performance modelling and simulation researcher will lead pathfinding research activities, encompassing application-architecture-technology co-design framework.
- Identify memory and communication bottlenecks in modern state-of-the-art SoCs and evaluate the impact of mitigation techniques such as memory-logic partitioning, package level partitioning, chiplet organization, hybrid integration solutions to scale systems more efficiently.
- Identify emerging applications exhibiting high memory access and compute requirements especially in domain-specific disaggregated computing architectures and map them to acceleration opportunities.
- Develop a simulation infrastructure spanning across various stages of the memory hierarchy (cache/main memory/storage), performance and technology models; optimize the infrastructure with respect to workload requirements; and propose efficient design of experiments (DoE) to substantiate the simulation methodology.
- Develop and manage software frameworks, repositories and programs. This also involves maintaining, releasing, documenting and improving existing software by collaborating with partner universities on new features, and testing and evaluating new tools from the community.
- Keeping up to date on recent developments in the field. You do this by studying literature and interacting with your colleagues.