technology
rugged industrial computer markets
COTS hardware - ruggedized
We make use of high performance, multi-core processors with a guaranteed upgrade path. We deploy these on compact form factor, industrial base boards with an industry standard mini-ITX or ETX layout, but customise the peripherals and connectors to increase ruggedness and thermal properties.
In order to achieve a greater level of environmental ruggedness, we deploy the computing hardware in a tough, shock proof, conduction cooled chassis. We replace industrial connectors with MIL spec equivalents in order to deploy general purpose computing capability into military systems.
GPUs are the new massively parallel computers
In the last three years, commodity graphics processors (GPUs) have evolved from fixed-function graphics units into powerful SIMD processors. These streaming processors are capable of sustaining computation rates of greater than ten times that of a todays typical CPU.
Our engineers are working in the evolving field of general purpose computation on graphics processors (See GPGPU.org) have demonstrated mappings to these processors for a wide range of computationally intensive vision tasks.
Using the Vision4ce Gripworkx framework, we are able to stream video from analogue, digital and gigabit Ethernet cameras into the GPU texture memory for real-time processing and display, as well as some other useful high-level functions.
the benefits of GPU acceleration
- 10x CPU Bandwidth & Computation
- Math: 472 GFLOPS (8800 Ultra)
- Bandwidth: 86 GB per second (Tesla C870)
- Highly Parallel Math Intensive Architecture
- Thousands of threads
- Tightly bound 35GB/sec GPU Memory
- PCIexpress (x16) 8GB/sec
- Tremendous scope for innovation and acceleration
- Infinite flexibility in display options
- fast development times
algorithms
We are constantly working on new capability within this rapidly changing field of research, harnessing the power of the GPU to massively accelerate processing tasks. The current GRIPworkx application framework can be configured to support and accelerate algorithms such as:
- Image enhancement and noise reduction through intelligent spatio-temporal integration
- Dynamic local contrast enhancement and multi-sensor alignment and fusion
- Motion analysis and optic flow for motion detection and analysis
- Whole scene tracking, stabilization and mosaic formation
- Brute-force small target detection for low SNR environments
- Other numerically intensive vector algorithms
