Originally Published June, 2022 - Gigantor Technologies in Melbourne Beach, Florida, develops earth-shaking technology for neural network machine vision models. GigaMACS™ solves latency problems and enables high-speed frame rates regardless of frame size or model workload.
Gigantor proudly holds three patents, all issued within months of filing due to the unique non-von Neumann architecture. Continuing the remarkable trend of immediate patent approvals, Gigantor’s latest issued Patent, 11,354,571, Parallel Implementation of Deep Neural Networks Applied To Three-Dimensional Data Sets, was issued on June 7th, 2022.
Don Gaspar, the CEO, said, “GigaMACS™ already solves numerous performance and latency problems that plague TPUs and GPUs due to architecture, and now, our technology adds an additional dimension. Real-time inference for 3D neural network models, high frame rate and HD or 4K frames, and beyond.”
GigaMACS™ provides the power for neural networks to accept live video or data from high speed cameras, LIDAR, and RADAR systems while returning inference outputs in real-time. Now, this revolutionary technology is available for three-dimensional convolutional neural networks, also known as 3D CNNs.
Neural network models designed for 3D imaging must process massive quantities of three-dimensional data. With GigaMACS™, 3D CNNs or DNNs can now accept data directly from real-time scanners, process the data as presented, return the inference immediately, and maintain consistent high performance with near-zero latency. Use cases include medical data, point clouds, and volumetric voxels; 3D CNNs can greatly improve human interpretation of these difficult-to-visualize forms.
Mark Mathews, the CTO, said, “Our latest Patent brings the benefits of GigaMACS™ to higher dimensional neural network models. MRI and CT scans directly sample a three-dimensional object (the patient) which allows a 3D model to recognize and localize diagnostic aberrations imperceptible to the human eye.”
Today, AI researchers design and teach 3D models to accurately identify multiple forms of cancers using excessively large 3D data sets. With the complete GigaMACS™ system, these models will locate abnormalities and provide diagnostic targets in real-time before the patient leaves the office. Larger, more complex models needed for medical tools are possible because GigaMACS™ will not buckle under the pressure of heavy workloads. GigaMACS™-powered medical instruments can help save lives by providing accurate results immediately.
About Gigantor Technologies Inc.
Don Gaspar and Mark Mathews founded Gigantor Technologies to boost neural network performance into real-time frame rates for complicated Machine Learning and Artificial Intelligence models. GigaMACS™, takes an existing TensorFlow or other neural network format and converts the model into a high-performance circuit for FPGAs or custom ASICs. GigaMACS™ will accelerate the neural network model to keep up with any high-speed input.