Amd Ryzen And Vega Coming Soon

http://www.forbes.com/sites/antonyleather/2017/02/01/amd-confirms-ryzen-processor-and-vega-graphics-card-release-dates/#cb26628552bf

Whether you are a fan of AMD or not, these two products coming on the market soon will be the first low cost alternatives to compete with Nvidia and Intel in years and indicate a new race for the bottom where the two compete for low cost video gaming solutions. For just playing a decent game at decent frame rates and high resolution the six core R5 model of Ryzen is all you need and which version of the Vega video cards might be the best bang-for-your-buck is yet to be determined, but their high end offering should offer up to 22 terabytes of raw computing power and most people won't need half of that just for casual gaming at normal high resolutions.

However, those facts speak more to the current limitations on video game hardware and software which will change dramatically over the next couple of years. For example, later in the year AMD will be releasing their Raven Ridge version of the same chip which will include HBM2 memory which the entire industry has adopted as a new standard. Basically, they are adding simple processing capabilities to the memory itself in order to reduce the traffic back and forth to the cpu cores and the Raven Ridge chip using HBM2 can provide the equivalent of a current PS4 in a single chip. The important thing to remember, is these are the low cost alternatives coming on the market first and if you wish to future proof your investment it would be best to wait for the first eight core Raven Ridge chips to come out by next year at the latest. Memory and bandwidth are rapidly becoming the biggest bottleneck in computing and all sorts of new solutions, including improved algorithms, are coming on the market allowing manufacturers to mix and match them as they please and make even a game only designed for four cores use more.

Bandwidth and AI are rapidly emerging as the decisive factors that will replace the traditional concerns about things like cpu latency and Intel remarked over a year ago that what is required is an entirely new computer architecture. Moving simple repetitious tasks into the memory itself lowers the latency, but so can something as simple as using an optical wave guide on a chip to allow cores to communicate faster and using small circuits here, there, and everywhere that provide AI that can replace many common functions that might otherwise lug cpu cores and graphics cards alike. Its pretty much up in the air at this point as to what will come on the market three years from now and until Intel and Nvidia release their next generation products its just not worth speculating on what will come out within the next three years other than to say it should be interesting to say the least.