Why the GPGPU is Less Efficient than the TPU for DNNs
The GPGPU’s massive multithreading is unnecessary for DNNs, and imposes performance, area, and energy overheads. By avoiding such multithreading, the TPU is more efficient.
Archive of posts tagged: Machine Learning
The Future in Visual Computing: Research Challenges
The tremendous growth in visual computing is fueled by the rapid increase in deployment of visual sensing (e.g. cameras) in many usages ranging from digital security/surveillance and automated retail (e.g. smart cameras & analytics) to interactive/immersive...
DNN Accelerator Architecture – SIMD or Systolic?
While the concept of hardware acceleration has been around for a while, DNN accelerators are perhaps the first to see the light of commercial adoption due to the AI/ML wave. Giant software corporations, veteran hardware companies, and a plethora of start-ups have...
Intelligent memory architecture with new memory technologies
A previous blog titled “Blurring the lines between memory and compute” by R. Das was a nice summary of the history and the recent trends on addressing the memory wall challenges with process-in-memory (PIM) ideas. This blog would like to further highlight...
Why Machine Learning Needs Benchmarks
It’s a marvelous time in computer systems. For me, working in Deep Learning feels like living through a scientific revolution. Kuhn described this kind of change in his classic book, where a new paradigm takes hold, causing entire fields to change their standard...Subscribe
