HOME                ABOUT                TEAM                PRODUCTS                CONTACT


 

We accelerate convoluted neural networks

Gigantor Technologies was founded by Don Gaspar and Mark Mathews with the goal of making neural network performance real-time for complicated Machine Learning and Artificial Intelligence models. Our first product, GigaMACS, will take your existing TensorFlow or other NN model that you have developed and make a high-performance bitstream for you to use with your FPGA or ASIC. GigaMACS will accelerate your NN model to levels where it will perform as fast as your input, to real-time. Have a faster camera, radar, lidar? GigaMACS will have your model keep up with it. Unless you have a Tardis, you won’t manage your system clock as well as GigaMACS will.

 

Only a Tardis manages your clock better

Throwing faster hardware to speed up your neural network model is not addressing the problem, and provides only small incremental gains. GPUs and TPUs, and more memory is not the answer, generates too much heat and is still too slow. Gigantor can optimize your existing neural network model to new performance heights, and even help you migrate to an ASIC for the ultimate performance. Only a Timelord from Gallifrey could make a neural network leverage time better.