http://www.computerandvideogames.com/38 ... ics-iwata/

'She cannae take any more, captain!'

It will only get worse if there are more devices using the internet. For example tablets are a growing segment of traffic and if they continue to grow then it could be hard for cloud gaming to have sufficient latency to make games playable.

I can't see cloud gaming working myself.

Nintendo want to maintain a business model that places very high barriers to entry (the exorbitant costs of developing a platform and using content to subsidise hardware).

But over time this will surely bring them into conflict with tech companies who have already overcome all the same issues that affect cloud adoption in the corporate marketplace. They have deduplication and TCP prediction tech for wan acceleration and to reduce inter-site replication traffic already. Distributed cloudy datacentres and content delivery networks put 90% of consumers within spitting distance of the content they are downloading, and network latency can be further reduced with simple QoS.

The costs of using all that tech to reduce latency between gamers and the already proliferating distributed datacentres for hosting the cloudy services and the content delivery networks would be lower across the industry than those of the business model above. The hardware for your home wouldn't need subsidy any more as it would only have to be able to draw to a screen, send user inputs backwards and forwards, and hold a table of maybe 10gb for the wan acceleration. If you need to support better graphics, higher quality video etc, all you have to do is take that table and make it bigger, network traffic would grow at a rate fractional to the quality of content being delivered.

So the games companies whose products are used to fuel all that subsidy for Nintendo hardware could sign a cheap deal with the likes of Amazon, Riverbed, VMware and the ISPs etc and probably still get to keep twice as much of the money you spend on their product than they do using Nintendo as the platform.

That's assuming you can stand to play your game with 15 to 20 ms of ping. If you can't, then you will become one of the declining number of hold outs desperately propping up the increasingly expensive (per customer) games console market.

It also assumes the developers are up to the job; some of the slowdown instances I've seen where the game has been installed to the HDD, never mind controller input! And that's after the traditional round of patches. For a game you actually own...

What is the slowest acceptable response time and video quality?

I have a 30mS ping to bbc.co.uk and my throughput is about 10Mbps; that's 20KB per frame at 60fps with a latency of 47mS or 40KB at 30fps with a latency of 63mS.

The most efficient caching algorithm is where every character graphic is stored locally along with all the backgrounds. However, it's then it's no longer "cloud gaming". You need video acceleration to assemble all the elements together at their correct positions in 3D. That's really no different to how existing on-line games work.

The only "cheap" acceleration we have is for pure h264, so you're looking at existing streaming video technology. Video rates for true 1080 video are around 25Mb/s while iPlayer HD is 720p at about 3.2Mb/s. Gaming at 720p 30fps gives pings of 50mS which are probably acceptable for many, but won't challenge proper hardware for serious gamers any more than a Nintendo Wiii does.

However, with FTTC and a good provider you could probably do it properly - it's about 10x faster than what I have. The tech is already here, it's just not in every house yet.

That's not how a wan acceleration or data deduplication algorithm works though. They just have have matching data tables in the store at each end of the line and replace chunks of whatever they are sending with pointers to sections of that table. If you find yourself sending much raw data over the link, your table needs updating or to grow. And it doesn't matter in the least what you are sending because it's just a load of ones and zeroes in sequences that tend to repeat.

I don't really see how cloud gaming faces any major challenges that aren't faced by other types of virtual desktop infrastructure. Some people are implying that latency and bandwidth issues are insurmountable. I merely point out that the challenges of running VDI over long wan links with rotten pings, and of transferring large amounts of data over insufficient lines very very quickly have been dealt with many times already. There's no special reason why the tech involved shouldn't be rolled out to consumers as part of a monthly paid service.

Such algorithms will be ineffective. Video data is already heavily compressed and very unlikely to repeat. h264 already squeezes as much quality out of the available bandwidth as modern hardware can handle. 3D video games bare no resemblance to a desktop, which is largely static with repeatable 2D objects being moved - just try running one over RDP!

Unless you're talking about an 8 bit platform game.

TBF, H265 is looking like it will get better compression than H264 can in real world scenarios. Whether that's able to help much in streaming games (and when/if H265 actually gets adopted at any sort of scale) is anyone's guess. The only viable way to do it may be to provide a box, like OnLIve did, rather than offering a software client app.

Jon

There will always be progress, in improved compression techniques as well as in cheaper more capable hardware. My point was, it's video compression technology that enables cloud gaming and you can look to things like iPlayer to get an idea of what can be achieved now.

I also think that there could be a market for ISP's to have a separate network for gamers. It could be priced in a way that ensures low latency and contention but also low caps to avoid iPlayer etc.

Well I stand corrected. Although VMware does have impressive tech that runs 3d gaming over vdi, it relies on clever use of video compression, so it won't accelerate in the same manner as older vdi setups. So I guess cloud gaming is going to rely on boring, glacial developments in broadband, qos, and codec technology after all.

My money remains on PS5 or whatever being delivered as a service rather than a platform still.

They might try but what proportion have a paying PS account?

And that's just the initial problem, currently.

£60 for a game that took so long to download you could have got a taxi into town, bought it for £40 and had installed with a beverage on the go in the meantime? And then lent it to a friend or sold it if you wanted, freed up space without having to wait hours if you decide to play it again... With the market for games (and their developers) collapsing, they haven't even the wit to make money on stuff that's already been produced and costs next to nothing to provide, so I wouldn't give them money for streaming unless it was like renting a game prices.

Personally I might be attracted to a £30 game that is playable on its own, and then if you want there are additional downloadable content a bit like Civ 5. So even if I sold the game the next buyer could also buy the additional content if they wanted. By having lots of potential content that is priced competitively so that the game does not have a short playing life then the resale market would be much smaller. The industry is going about the problem the wrong way.