Because of how close we are to the New Horizon event, leaks and details popping up were inevitable. Today, we finally have confirmation that AMD’s desktop lineup of Zen Processors (which run on the Summit Ridge AM4 platform) will be named ‘Rzyen’.
Ryzen appears to be a complete monster, with the top of the line processor featuring 8 Processor cores (16 threads), run at 3.4 Ghz and have a total cache size of 20MB. Oh – and the crazy part? They actually are smart enough to automatically overclock themselves with XFR (Extended Frequency Range), meaning that less technically inclined folks won’t need to worry about voltages, timings or multipliers.
This also confirms that AMD has packed in 16MB of Level 3 cache, and 4MB (total) of L2 cache in the processor. This amount of cache should reduce the number of trips Rzyen needs to make to the system’s main RAM, and thus reduce both latency and improve throughput.
Pure Power is the lynchpin in Ryzen’s overclocking arsenal, monitoring the power, temperature, clock speeds and other assets of the processor. From AMD’s own breakdown of Zen, the CPU will heavily use Clock Gating, which simply enables or disables parts of the processor based upon their usage. So, if a particular core, a particular unit (say the Floating Point in core 2 is doing no work, as it’s all Integer based calculation), it appears the CPU will automatically disable this until it’s needed again.
AMD’s Precision Boost is labeled as ‘Fine-grained’ frequency control, and should constantly be tweaking the clock speeds of the processor cores based upon power consumption, heat output, and TDP. It does this in 25MHz increments and happens in just milliseconds. In theory, this should allow the chip to draw roughly the same power, but clock dynamically based upon workload across the various CPU units, the number of cores and threads, and so on.
XFR (Extended Frequency Range) apparently scales based on the thermal abilities of the chip – so in theory at least, you should spot a rather large difference between going from a basic air cooler to a high-end water loop. Ryzen’s XFR will work alongside AMD’s Precision Boost to find the best clock speeds available for the chip, and at least in theory, your processor should operate at excellent clock speeds.
Naturally how well this works in real life, and how much manual control users have of Ryzen’s overclocking features remains a mystery – at least for now.
Looking over our Zen analysis, we’d mentioned quite frequently AMD were in the business of optimizing the CPU’s front-end, essentially being better at anticipating the way code was going to turn during branch prediction (for if / then statements for example), and with larger data stores and much faster caches, the processor should be able to keep important data on chip, rather than constantly farm out to RAM.
Well, AMD confirms this – plus much more with Smart Prefetch. If it is as effective as AMD are touting, it should quite literally learn the pattern of data access in a program and then be able to predict where and when it will need to fetch data from ahead of time. Quite simply, think of it as being a shopper in a supermarket you really know. You know longer need to think which isle you need to head down to buy oatmeal, you simply head there. But with this, imagine you have an army of helpers who are already heading to the isle, ahead of time, as they know your shopping habits and as you walk down the aisle people are grabbing the relevant products from the store and throwing them into your trolly.
AMD have also written once again, Q1 2017 as the release window for Ryzen, so we should (in theory), see the CPU hit shelves before April next year. With any luck we’ll get a more concrete release window during the New Horizon event, later on, today.
There’s a lot more we could go through regarding Ryzen, but due to the close proximity of New Horizon, we figure we’ll wait and see what else is revealed. Stick with us at RedGamingTech for more information and analysis.
Credit for images – VideoCardz.com