binary solo said:
So 200ms (i.e. 0.2 of a second) latency just for doing lighting effects (150ms for lower quality). That's 6 frames at a 30fps frame rate, or 12 frames at 60fps. With that sort of performance the cloud will only be able to be used for the most passive of visual elements, which won't actually do much to significantly improve the graphics. But yes they show that it can be done, which means performance can be improved (i.e. reduced latency) eventually to the point that it could be used to significantly improve the visuals of games that are primarily rendered on the local box. However don't be expecting that in 2014 or 2015. But the video also shows that for best cloud processing performance you also need well spec'd local hardware. So the ability for the cloud to enchance Xb one games is still limited by the specs in the Xb one. How does the Xb one's graphics specs compare to the GPUs in the laptop and PC of that video? And let's not forget that there is nothing magical about the Xbox cloud. Every console with the ability to connect to the internet can use cloud technology to improve aspects of a game. It's just that in the case of non-Xbox consoles the developer or publisher needs to hire a 3rd party cloud service. So MS's secret sauce is that they just make it easier for developers/publishers to utilise the cloud. So by the time the Xbox cloud is technically capable of mindblowing graphics enhancement, the cloud will be so ubiquitous and cheaper to access that it will be feasible for developers to utilise cloud processing on all versions of its games. Indeed because MS has invested in such a massive cloud infrastructure using today's hardware tech it might be that MS lags behind independant cloud services in upgrading hardware.
|
The technology that both AMD and Nvidia have demonstrated is that rendering of graphic features such as lighting and backgrounds can be rendered, sent back down to the console via a video file, decompiled by the discrete stream engine and put into a form the GPU can use then applied to the visualization the GPU is using.
The benefit, besides off-loading GPU intensive tasks, is that it frees up the GPU to do more work than it normally would be able to handle.
So yes, it does actually improve the graphic fidelity because the local GPU is freed up from doing some of the scene rendering, thus it's free to use that extra capacity to provide greater fidelity. The majority of work in processing the stream is taken up by the streaming engine (a discrete processor). This is part of the reason why both the Xbox One and the PS4 include them. The difference is that the Xbox One has two. One for compression, the other for decompression. So while the Xbox One can compress gameplay video for the Game DVR, it can in the same instant, decompress streaming video.
The biggest challenge to cloud computing is network reliability. Latency is a minor issue compared to network availability and server uptime.
As for governance, that is maintaining server hardware. No. Just no. This argument is completely stupid and doesn't even realistically taken into account how data centers are maintain. Server hardware is constantly being rotated out, if not on a monthly basis, on an annual basis based on the age of the hardware. Not to mention, long before the cloud would be used for graphics rendering, both the data center and network availability would be tested to ensure that the experience would be sutible for the consumer.
And in reality, the biggest issue with a cloud GPU is the capability of the GPUs installed into the server and how many are installed. Cloud GPUs have multiple high-power GPUs. They're not a stock standard server. So, while the CPU isn't necessarily the most powerful CPU, the GPUs would be extremely powerful.
