At Siggraph AMD had some software news, specifically a renderer and a photo stitcher, to reveal. Of the two, SemiAccurate thinks the renderer is the one to watch.
That renderer is called Radeon ProRender, the new name for FireRender. In case you aren’t familiar with the professional graphics market, FireRender/ProRender is a raytracing engine that plugs into high-end graphics programs like 3DS Max, Solidworks, Rhino, and Maya. They have traditionally been the domain of expensive packages like iRay and that is the market AMD is aiming to challenge. It isn’t all that coincidental that Nvidia has a large revenue stream from this segment.
What is AMD doing to challenge the entrenched players? How about making it open source, free, and hardware agnostic. Remember AMD’s GPUOpen promise to do things in a more open and accessible way from now on? ProRender is a big part of that. Previously you needed expensive licenses coupled to specific hardware, now you can get for free with a click from the Autodesk store for example. No extra charges for program specific plugins, no lock-in, and it is open if you want to tinker. It will obviously be optimized for AMD hardware but unlike the competition, not purposefully de-optimized for competitors.
The other software release today is called Project Loom, a GPU optimized mulit-camera photo/video stitching program. Traditional stitching is not really a trick, your phone can probably do it with ease. When you start talking about multiple high-rez cameras streaming video for 360 degree VR solutions, things are a bit more complex. Add in realtime streaming to a VR headset and you have the proverbial ‘issues’.
That is where Loom comes in, it runs on a GPU and is obviously optimized for the task at hand. AMD wasn’t showing off demos at Siggraph but they claim to have it working in the lab with multiple high rez cameras streaming realtime 360 degree video to a VR rig without nausea inducing latency. More on this one when we can try it out but the idea is compelling.S|A
Latest posts by Charlie Demerjian (see all)
- HyperX ships it’s 60 millionth enthusiast memory module - Oct 15, 2018
- Bittware/Nallatech water cools 300W of Xilinx FPGA - Oct 12, 2018
- More on Intel’s 10nm process problems - Sep 17, 2018
- Intel puts out another 14nm 2020 server platform - Sep 11, 2018
- Why Can’t Intel Supply Enough 14nm Xeons? - Sep 10, 2018