In our last article we wrote about how Cloud Gaming or Gaas (Game As A Service) progressed since the first introduction of the term back in 2013. Today is a though about two players which synthetize perfectly two singular approach of this business: NVIDIA and SHINRA. One is business first driven while the other is tech specification first driven. Those two actor provide a fairly portrait of what Cloud Gaming is today and what it can be tomorrow.
By providing both hardware and middle ware solution the GPU vendor insure himself a main global position covering almost all the layers of the sector. Put simply they “own” the solution from hardware virtualization to the user store. By creating a cloud based store platform and partnering with publishers to have their title available directly in the cloud they force the shift of the industry directly to their cloud solution. NVIDIA said that they designed the solution at every level but we may wonder how the game development itself is re-designed to take advantage of the cloud specification. From a publisher point of view targeting the cloud remain a challenge if not rightly supported by the solution vendor. Same B2B deal about game engine. What we want to point out there is that Cloud Gaming have to be considered from a design challenge perspective: Multi-user, wide open world, synchronization are a few of the technical design that have to be re-thought to fully take advantage of the cloud. Otherwise we would be in a VDI business and not Cloud Gaming. So let’s try to illustrate this matters with an example. Until now open game world where built by streaming the extra world data from a ROM media to the GPU, buffering only the necessary focused player area. The same technique from a cloud solution wouldn’t make sense due to the latency of memory bandwidth as well as a lost of shared asset between users. So entire part of the game architecture or game engine architecture have to be re-design for the cloud target. We bet all the player are at work on each one of those challenge as well as NVIDIA itself with cloud-ready middleware. But for now the giant is insuring partnership first, then will for sure upgrade the entire solution design to allow the full potential of the cloud. And to have a grasp of what we are missing with this business first strategy we would like to turn to a different approach which prioritized design challenge: Project SHINRA.
Project SHINRA, formerly “Project Flare” was launched two year ago by Square Enix to pioneer the Cloud Gaming sector. From a conceptual point of view, Cloud Gaming, by abstracting the technical specification of the hardware, shouldn’t have any limit in its game world. Asset should be shared massively, full synchronization of IA agent, user agent as well as Physic event in a massive pool of real-time computed world shared by theoretically unlimited number of users. This should be the technical specification that any publisher would expect from a GAAS platform. SHINRA started from ground zero such a GAAS platform with those goals as root for their design. Now SHINRA tech demo really demonstrated some achievement. The solution provide large scalability and clear the synchronisation bottleneck in a 2 step process: computing then rendering. The computing phase is designed on a CPU cluster with all data loaded in RAM. Rendering pass then handle viewport management for each stream. Game developer can leverage cloud implementation to the Shinra team which work as B2B partner on a title basis. They also provide an other option, equivalent to the VDI like pattern of actual GPU cloud service: deployment of an existing title “as it is” on the their platform and store. We are avidly waiting for their CCDK that should be available during this spring to have our hand on this technology that promise best the future of the gaming industry.