My main interests lie in the intersection between computer architecture, memory systems, hardware security and reliability, and AI.
- Memory Systems and Processing in Memory
- Hardware Security and Reliability
- Designing efficient computing systems specifically for AI and emerging applications
- Other research interests
1. Memory Systems and Processing in Memory
I’m interested in improving the performance and efficiency of DRAM (and other memory technologies), making memory more capable, and reducing the memory bottleneck.
Featured Papers:
- pLUTo: Massively parallel computation in DRAM via lookup tables
- CODIC: A low-cost substrate for enabling custom in-dram functionalities and optimizations
- FIGARO: Improving system performance via fine-grained in-DRAM data relocation and caching
- DAMOV: A new methodology and benchmark suite for evaluating data movement bottlenecks
Featured talk:
2. Hardware Security and Reliability
I’m interested in understanding and improving DRAM reliability and security. I experimentally characterize commodity DRAM chips to better understand RowHammer and related vulnerabilities. I’m also interested in developing low-cost mitigation techniques for secure and reliable computing, and in exploring new memory technologies that are fundamentally secure and reliable.
Featured Papers:
- SpyHammer: Understanding RowHammer behavior under temperature variations
- BlockHammer: A low-cost defense mechanism against RowHammer attacks Intel Hardware Security Academic Award Finalist
- D-RaNGe: Using commodity DRAM devices to generate true random numbers with low latency and high throughput
- Revisiting rowhammer: An experimental analysis of modern dram devices and mitigation techniques
- A deeper look into RowHammer’s sensitivities: Experimental analysis of real DRAM chips and implications on future attacks and defenses
Featured talk:
3. Designing efficient computing systems for AI and emerging applications
This research topic will become increasingly important in the future. I’m particularly focused on specialized architectures for AI and Processing in Memory (PIM).
Featured Projects/Papers:
- 1HealthAI European AI Factory: Leading an €82M initiative to scale AI infrastructure
- EDEN: Enabling energy-efficient Deep Neural Network (DNN) inference using approximate DRAM
- EcoFlow: efficient convolutional dataflows on low-power neural network accelerators
Featured talk:
4. Other research interests
Other topics of interest include SSD optimization, GPUs, transactional memory, and parallel debugging.
Featured Papers: