AMD's Bristol Ridge & Stony Ridge Detailed
If you look back over the last year of processor-centric reviews and articles here on Hardware Canucks, you’ll notice a trend: AMD products take up the vast majority of our coverage space. Despite being given up for an also-ran by many naysayers, this is a company in the midst of an epic turn-around and one that could end up being a legitimate competitor against Intel once again. While many of AMD’s architectures and products we’ve covered have failed to gain massive traction they’re obviously trying mightily to remain relevant and, more importantly, offer products that their customers (both end user and OEM) are looking for.
This brings us to today’s “official” announcement of the Bristol Ridge and Stony Ridge platforms. These are two of the three pedestals –the other being the upcoming Zen microarchitecture- upon which AMD is building their future initiative. While this may seem like treading down the same path again since I’ve already discussed Bristol Ridge’s comparative performance metrics and a few additional details in separate articles, that’s all over now. According to AMD systems featuring these APUs will be shipping in the coming weeks.
One of AMD’s primary goals going forward is to simplify their lineup in order to increase commonality and reduce confusion. Obviously they’re in a transitional phase right now as their 7th generation APUs are going to be utilized as a placeholder until the next generation architecture, code named Zen, is ready to roll. To that end there will be a new all-encompassing AM4 socket that strides across the upcoming Summit Ridge FX-series CPUs while also supporting desktop APUs. It will also morph into a scalable FP4 socket for mobile applications and it’s that market this article is going to focus upon.
To be clear this announcement marks the inception the 7th generation mobile APUs exclusively with the desktop variants still under various stages of development. In addition, the mobile Bristol Ridge roadmap is now being broken into two different segments, those being Bristol Ridge which is being used to replace Carrizo-based mobile APUs while the aforementioned Stony Ridge aims to replace the Mullins ultra low voltage family. Of these the latter represents perhaps the largest step forward since that architecture has been around for the better part of 30 months.
Summit Ridge meanwhile -the replacement for AMD’s age-old Vishera FX CPUs- isn’t quite ready for launch yet but when it does, AMD’s new Zen microarchitecture will be laid bare for the world to analyze in all of its eight core, 16 thread glory. We expect it to quickly cascade down into so-called 8th generation APUs sometime in 2017.
Bristol Ridge; Mobile, Refined
The desktop market is obviously the one which concerns many of the people reading these pages but AMD needs volume shipments to increase their market share if there’s any hope of long-term survival. That means targeting the burgeoning mobile market with energy efficient options across a broad range of categories; from mainstream performance to entry level. If that’s accomplished they will then be able to better funnel development funds into higher level projects. This is where Bristol Ridge gets factored into the equation.
In many ways Bristol Ridge is simply an evolutionary step for Carrizo; it is still based off of the old but now heavily refined 28nm manufacturing process and utilizes Excavator CPU cores. I’ve already gone in-depth about all the newness Carrizo brought to the table in a dedicated architecture overview and I recommend you read that for some additional background.
What we can’t do is call Bristol Ridge a rebrand since there’s quite a bit of newness here as well. These APUs will be the first to include a DDR4 memory controller with support for speeds up to 2400MHz. In addition, there have been even more manufacturing process improvements so AMD’s engineers have squeezed higher clock speeds out of the silicon without an associated TDP uplift.
What’s abundantly evident is that Bulldozer and the market’s lackluster response to APUs has pushed AMD into such a deep hole, they’re still struggling to hoist themselves back out half a decade later. Despite its very noticeable dearth of design wins in the key notebook market Carrizo represented what AMD refers to as a turning point. Those chips with their Excavator microarchitecture took some very important steps forward in terms of basic x86 performance throughput, efficiency and a number of other aspects. While Bristol Ridge may be a Carrizo derivative, AMD is hoping it continues gaining ground on Intel’s latest Skylake offerings and sets the stage for Zen’s even more significant IPC uplifts.
Stony Ridge – Puma Finally Replaced!
Whereas Bristol Ridge will only move forward the performance of AMD’s APUs by a relatively minor amount in the mainstream and performance notebook markets, Stony Ridge is supposed to be a quantum leap forward for its specific category. Like the Carrizo-L and Beema architectures it replaces, the APUs within the ULV (ultra low voltage) Stony Ridge lineup are hoping to make some serious inroads in the convertible tablet, slim & light and ultra portable segments. One area AMD isn’t touching upon though is a Mullins (sub-6W) replacement for tablets.
At the heart of Stony Ridge beats a pair of Excavator cores alongside three dedicated GCN 1.2 graphics units. While that might not sound like all that much, consider that both Beema and Carrizo-L used the Puma architecture, a design that had been around for years and one that just couldn’t keep up with the demands of modern workloads. The step towards Excavator will be a quantum leap forward for performance while the space savings from the more condensed architecture has allowed AMD to add another graphics unit to the equation.
Past the obvious benefits of moving towards Excavator, Stony Ridge also includes support for DDR4 memory up to 2133MHz (though only single channel) and substantially higher core frequencies without increasing power consumption. There’s also been updates to the video processing engines and added support for HDMI 2.0, MJpeg decode, H.265 hardware decode and VP9 Decode. With these changes AMD’s ULV parts have jumped from an also-ran to a bonafide competitor.
Now I completely understand that ultra portables aren’t cutting edge stuff but what AMD needs right now is design wins from system integrators to effectively drive volume sales. Stony Ridge is a solution that could accomplish exactly that and with performance and wattages that challenge some of the best Intel has to offer.
A Quick Look at AMD’s Mobile Product Stack
In the past AMD’s mobile lineup was an extremely confusing thing. Less than two years ago it consisted of no less than sixteen different products spread throughout their performance, mainstream, entry level and ultra portable segments. This has now (finally!) been condensed into a far more streamlined and sensible product stack that allows system integrators and end users to home in on specific products for each market or usage scenario.
The 7th generation product stack is broken into two distinct sections. The FX, A12 and A10 series parts are specifically being meant for the standard notebook space while A9, A6 and E2 APUs hope to offer ultra mobile options that focus on a synergy between CPU cores and GPU output.
One major change here is a move away from emphasizing these chips’ configurable cTDPs over actual constant TDP values. While partners can still modify every one of these parts so they can feature performance at a system-specific level, there’s some hope that these new TDP designations will help clarify AMD’s newest APUs for end users.
All in all it looks like AMD’s combination of Bristol Ridge and Stony Ridge could be a winning one but then again, we have been saying that since Llano was launched years ago. There’s no doubt that with Excavator and Zen past wrongs are being corrected and hopefully a new AMD will emerge which is leaner, meaner and better able to compete with Intel on a level footing. There’s no doubt these two common platforms are a step in the right direction but their ultimate success or failure will be determined by how many design wins are achieved rather than their relative performance.
|Latest Reviews in Processors|