Ishita Chaturvedi

I am a computer architect at Cerebras, working on hardware-software co-design for next-generation systems.

I completed my PhD in Computer Science from Princeton University. I was very fortunate to be advised by Prof. David August.

My research focused on High-Performance Computing, specifically on efficiently utilizing computational resources and extracting maximum performance from CPUs and GPUs. I extensively explored the parallelism offered by both CPU and GPU architectures, as well as their inherent limitations. My goal was to bridge the architectural gap between CPUs and GPUs, bringing them closer together, a vision partially realized through GhOST: Out-of-Order execution on GPUs and SHADOW: Assymetric SMT on CPUs.

Before jumping head on into computer science with a focus on computer architecture I completed my Bachelors’ in Physics at Delhi University, India. As an undergraduate I worked on developing a Radio Transient detection pipeline at the National Center for Radio Astrophysics. This CUDA-based pipeline aimed to detect pulsars in real time. While working on this project I realised how hard it is to write efficient GPU code! While a GPU can theoretically give unprecedented speedup it is very difficult to realise in practical implementation. With my experience working with GPUs I decided to pursure computer architecture to make GPUs more efficient and easy to program on.