Latest question:
How do you think the new GigE standards will influence the machine vision industry?
Respond or ask your question now!




By Larry Adams
Editor
Faster data transfer speeds, higher resolution, real-time processing. These are all goals of most imaging users. These also require bandwidth, a commodity in short supply as devices gobble up CPU resources and create traffic jams waiting for information to cross to and from a device.
The PCI boards and its antecedents have generated a fast highway of communication that have helped to move data at fast speeds, but their asynchronous or serial format means that the data must first move from place to place before other sets of data can make its journey.
“Conventional PCI buses that have been used on host computers are parallel in nature,” says Inder Kohli, product manager, DALSA Coreco (St-Laurent, QC, Canada). “That means that both address and data move in parallel and everything has to go to each and every connector. As there is demand for higher and higher performance, the only way to achieve that is by clocking it much faster or by a combination of clocking faster and increasing the width of the data path itself. It is literally analogous to a highway. You raise the speed limit and increase the number of lanes on the highway.”
Suppliers also say that the conventional PCI, as well as PCI-X and AGP, is not able meet its theoretical bandwidth. While there are several reasons for this, one contributing factor is that the available bandwidth is shared by all the devices on the bus, resulting in lower bandwidth available per device.
“Look at the cameras coming out now. For a while, it was okay to have analog cameras that give 30 frames per second and give you that VGA image,” says Tony Iglesias, National Instruments’ (NI, Austin, TX) Vision group manager, who worked on the original Camera Link standard and was a designer of NI’s new PCI products. “Look at the sensors and cameras. We have 11-megapixel cameras out there. A user does not have to go very fast to need a ton of bandwidth to acquire that image. This trend is not going to stop. People are going to want to acquire larger images and process them faster.”