techbrief-government-digital-transformation

techbrief-government-digital-transformation

New York, NY, October 15, 2025 - ACM, the Association for Computing Machinery and IEEE Computer Society have named Saman Amarasinghe of the Massachusetts Institute of Technology as the recipient of the 2025 ACM-IEEE CS Ken Kennedy Award. The Ken Kennedy Award recognizes groundbreaking achievements in parallel and high-performance computing (HPC). Amarasinghe is cited for fundamental contributions pioneering high-performance domain-specific languages, exceptional mentorship, and service advancing the global computing community.

Modern hardware presents a complex model of computation that promises high performance, but only if the compiler can successfully bridge the challenging gap between software and hardware to fully exploit the hardware resources. Amarasinghe has produced language designs and sophisticated compiler algorithms that successfully bridge this gap.

His specific technical contributions include:

• SLP: The Superword-Level Parallelism (SLP) compiler optimization was developed by Amarasinghe in collaboration with his PhD student Sam Larsen. It automatically generates code optimized for modern short vector units such as AVX, Neon, and AltiVec. Today, SLP is included of all modern compilers such as LLVM and GCC.
• Halide: This image processing language was developed in collaboration with Amarasinghe's then-PhD student (now Associate Professor) Jonathan Ragan-Kelly and Amar Bose Professor of Computing Fredo Durand, both in MIT's Department of Electrical Engineering and Computer Science (EECS). By separating the program into an algorithmic language and a scheduling language, Halide enables programmers to create portable algorithms and rapidly iterate over different schedules to obtain the highest performance on each architecture. Halide is included in many popular applications, including YouTube backend, Android Camera pipeline, and Adobe Photoshop.
• TACO: The Tensor Algebra Compiler (TACO) was developed in collaboration with PhD student Frederick Kjolstad (now a Stanford professor). TACO is the first compiler to generate efficient code for arbitrary tensor expressions over multiple tensor formats. This is important because the best tensor formats vary across computations and data sizes.
• StreamIt: The stream processing language (Streamlt) was developed in collaboration with Amarasinghe's PhD student Bill Thies. It was one of the early domain-specific languages to directly support emerging multicore processors.

In addition to his technical mentorship, Amarasinghe has trained young people to start businesses, especially in emerging economies. For 15 years he has served as Faculty Director of MIT's Global Startup Labs (GSL). GSL has conducted hundreds of summer programs to guide university students through the process of launching information technology-based startups. These entrepreneurship programs have seeded a vibrant ecosystem of startups in Africa, South America, and South Asia.

Saman Amarasinghe is the Thomas and Gerd Perkins Professor in the Department of EECS at the Massachusetts Institute of Technology. He leads the Commit compiler research group in MIT's Computer Science & Artificial Intelligence Laboratory (CSAIL), which focuses on programming languages and compilers that maximize application performance on modern computing platforms.

Born in Sri Lanka and schooled at Royal College, Colombo, Amarasinghe is a graduate of Cornell University and earned his Master's and PhD degrees from Stanford University. He was elected as an ACM Fellow in 2019.