Here’s a bold statement: the future of computing hinges on bridging the gap between software and hardware, and one MIT professor is leading the charge. Saman Amarasinghe, the Thomas and Gerd Perkins Professor at MIT, has been awarded the prestigious 2025 ACM-IEEE CS Ken Kennedy Award for his groundbreaking work in domain-specific languages and his unwavering service to the global computing community. But here’s where it gets fascinating: Amarasinghe’s innovations aren’t just theoretical—they’re powering some of the most widely used applications today, from YouTube’s backend to Adobe Photoshop. Let’s dive into why this matters and how his work is reshaping the industry.
The Ken Kennedy Award, named after the late high-performance computing pioneer, recognizes individuals who push the boundaries of programmability, productivity, and community service in computing. Amarasinghe’s contributions are nothing short of transformative. Modern hardware promises incredible performance, but only if software can fully leverage its capabilities. This is where Amarasinghe’s genius shines. His designs and compiler algorithms act as a bridge, ensuring that software can tap into the full potential of hardware resources. And this is the part most people miss: his work isn’t just about speed—it’s about making complex systems accessible to developers worldwide.
Let’s break down his key technical achievements:
SLP (Superword-Level Parallelism): Developed with his PhD student Sam Larsen, SLP optimizes code for modern vector units like AVX and Neon. It’s now a staple in compilers like LLVM and GCC, quietly powering performance gains in countless applications.
Halide: Created with Jonathan Ragan-Kelly and Fredo Durand, Halide revolutionizes image processing by separating algorithms from scheduling. This allows developers to write portable code and fine-tune performance across different architectures. It’s no surprise Halide is used in YouTube, Android Camera, and Photoshop.
TACO (Tensor Algebra Compiler): Co-developed with Frederick Kjolstad, TACO is the first compiler to efficiently handle arbitrary tensor expressions across multiple formats. Why does this matter? Because tensors are the backbone of machine learning, and TACO ensures optimal performance regardless of the computation or data size.
StreamIt: One of the earliest languages to support multicore processors, StreamIt was developed with Bill Thies and paved the way for modern parallel computing.
But Amarasinghe’s impact extends far beyond code. For 15 years, he’s served as Faculty Director of MIT’s Global Startup Labs (GSL), mentoring young entrepreneurs in emerging economies. Through GSL, he’s helped launch hundreds of startups in Africa, South America, and South Asia, fostering innovation where it’s needed most. This blend of technical expertise and social impact is what makes his work so remarkable.
Controversial question: As we celebrate Amarasinghe’s achievements, should the computing community prioritize performance optimization or accessibility? His work seems to balance both, but is this the right approach? Let’s discuss in the comments.
Born in Sri Lanka and educated at Cornell and Stanford, Amarasinghe’s journey is a testament to the power of global collaboration in tech. His election as an ACM Fellow in 2019 further cements his legacy as a leader in the field. As he continues to lead MIT’s Commit compiler research group, one thing is clear: his influence will shape computing for decades to come.
About ACM: The Association for Computing Machinery is the world’s largest computing society, uniting professionals to advance the field through dialogue, resources, and recognition of excellence. Learn more at https://www.acm.org/.
About the Ken Kennedy Award: Established in 2009, this award honors those who advance programmability, productivity, and community service in computing. With a $5,000 honorarium, it’s a fitting tribute to Ken Kennedy’s legacy. More details at https://awards.acm.org/kennedy.
This press release was distributed by WebWire, your go-to source for news release services. For more information, visit https://www.webwire.com/OurServices.asp.
