Research at the intersection of compilers, architecture, and intelligent systems.
We empirically assess modern large language models on representative HPC code optimization tasks, detailing accuracy, performance gains, and practical guidance for production integration.
DeepContext delivers cross-platform insights into deep learning workloads by combining context-aware profiling with actionable diagnostics for developers and performance engineers.
GPU programming is a critical component in AI system courses, which is notoriously difficult to learn and teach, given its unique features such as massive parallelism and data movement across memory hierarchies. This …
Tensor multiplication with learned weight matrices is the fundamental building block in deep learning models. These matrices can often be sparsified, decomposed, quantized, or subjected to random parameter sharing …