Accelerated-X™ Summit Series
3D Summit Series

OpenGL                        

Xi Graphics offers high-performance OpenGL v1.5 with some later extensions. Our OpenGL rendering pipeline is fully hardware-accelerated, and was developed "from the ground up" in-house.

Some OpenGL v2.0 extensions are available by request. However, since our typical customer is not a render farm or a gamer, but rather a medical researcher, seismic engineer, chemical scientist, simulation builder, and so on, there has not been the need to go all the way to v2.0. And v3.0 is probably not something we will offer in the near future, if ever.

XiG Direct Access Mode
When the OpenGL client application is on the same machine as the X server, the X Window System protocol can be bypassed, allowing the application(s) to "grab control" directly of the graphics hardware. This greatly speeds up 3D performance. We call this "XDA" for XiG Direct Access with OpenGL"

There are some pitfalls, of course, and some restrictions. Since the X server is not protecting you, your application can really screw things up, so be careful. It is fast though ;-).

Xi Graphics' approach to Direct Rendering is completely different from that adopted by XFree86 and subsequently by Xorg X servers - an approach termed "DRI/DRM".

Xorg opts to have the OS kernel heavily involved in the OpenGL process, and puts much of the involvement in the kernel itself. This results in a maintenance nightmare, as well as poor rendering performance.

The Summit Series OpenGL Direct Rendering approachs maintain the UBIX philosophy that graphics is an application in User Space, and is always under the supervision fo the kernel, but does not require the kernel to get involved with details that involve what graphics hardware is doing, or what the Xprotocol packets contain.

While the Xorg approach brings down the performance of the entire system substantially - in the name of "Security" - Xi Graphics' approach is to continue the philosophy of using tight access to the comuters if security is a concern, and not have the kernel snoop each data packet of X.