Direct GPU transfer

 • Sample programs for AMD DirectGMA and NVIDIA (CUDA) available.

 • Direct GPU transfer eliminates unnecessary system memory copies, lowers CPU overhead, and reduces

latency, resulting in 

significant performance improvements in data transfer times for applications.

 • Direct capture of image data to GPU memory is available using AMD’s DirectGMA. Compatible with

AMD FirePro W5x00 and 

above and all AMD FirePro S series products.

 General purpose I/O lines

 • Compatible with a wide range of sensors and motion encoders.

 • High-speed differential inputs: Quadrature motion encoder support up to 5 MHz.

 • Isolated current-sense inputs: 5V, 12V, 24V signaling voltages accepted, up to 50 kHz, individual

galvanic isolation up to 

250VDC and 170VAC RMS.

 • Isolated contact outputs.

 • High-speed 5V-compliant TTL inputs/ LVTTL outputs.

 High-performance DMA (Direct Memory Access)

 • Direct transfer into user-allocated memory and hardware boards that expose PCI addresses

 • Hardware scatter-gather support

 • 64-bit addressing capability

 Area-scan triggering capabilities

 • A trigger is used to start the acquisition when the part is in position. Hardware triggers come from the

Coaxlink’s I/O lines. 

Software triggers come from the application.

 • An optional trigger delay is available to postpone the acquisition for a programmable time.

 • A trigger decimation function allows to skip some of the triggers.

 • Camera exposure control allows the application to control the exposure time of the camera.

 • When the acquisition starts, at the appropriate timing, the Coaxlink board generates a signal to control

an illumination device connected to one of its output lines.