AMD, Vishera, and Beyond: New Design Philosophy Dictates a Faster Pace
Less Risk, Faster Product Development and Introduction
There have been quite a few articles lately about the upcoming Bulldozer refresh from AMD, but a lot of the information that they have posted is not new. I have put together a few things that seem to have escaped a lot of these articles, and shine a light on what I consider the most important aspects of these upcoming releases. The positive thing that most of these articles have achieved is increasing interest in AMD’s upcoming products, and what they might do for that company and the industry in general.
The original FX-8150 hopefully will only be a slightly embarrasing memory for AMD come Q3/Q4 of this year.
The current Bulldozer architecture that powers the AMD FX series of processors is not exactly an optimal solution. It works, and seems to do fine, but it does not surpass the performance of the previous generation Phenom II X6 series of chips in any meaningful way. Let us not mention how it compares to Intel’s Sandy Bridge and Ivy Bridge products. It is not that the design is inherently flawed or bad, but rather that it was a unique avenue of thought that was not completely optimized. The train of thought is that AMD seems to have given up on the high single threaded performance that Intel has excelled at for some time. Instead they are going for good single threaded performance, and outstanding multi-threaded performance. To achieve this they had to rethink how to essentially make the processor as wide as possible, keep the die size and TDP down to reasonable sizes, and still achieve a decent amount of performance in single threaded applications.
Bulldozer was meant to address this idea, and its success is debatable. The processor works, it shows up as an eight logical core processor, and it seems to scale well with multi-threading. The problem, as stated before, is that it does not perform like a next generation part. In fact, it is often compared to Intel’s Prescott, which was a larger chip on a smaller process than the previous Northwood processor, but did not outperform the earlier part in any meaningful way (except in heat production). The difference between Intel and AMD in this aspect is that as compared to Prescott, Bulldozer as an entirely new architecture as compared to the Prescott/Northwood lineage. AMD has radically changed the way it designs processors. Taking some lessons from the graphics arm of the company and their successful Radeon brand, AMD is applying that train of thought to processors.
The original Bulldozer die was pretty fascinating for the time, but that was before people knew that it was a bit of a dud.
So what exactly does this mean? For the past several decades we saw processor design follow a fairly simple routine. A new generation of architecture is released, and there are a few minor updates to the architecture other than moving to smaller process nodes. Around the seven year mark a brand new architecture is then introduced, updates are applied, and then the cycle starts over again. These massive jumps in technology are complex and expensive, and they take tens of thousands of man hours to complete. While the advantages of going with a clean sheet design are many, if a basic decision is made that turns out to be flawed or counterproductive, then years of design work are wasted. AMD appears to be trying to move away from this design paradigm, as the risks of making such a poor decision is nearly catastrophic for the company. Instead, AMD is looking at a more conservative, though accelerated, route. Essentially AMD is looking at major, yearly updates for their processor architectures. Instead of small updates over the years culminating in a massive redesign, they are taking many smaller steps in between. Much like AMD did with their process technology, they are applying this to the design methodology. This means small, meaningful changes on a very regular basis. This philosophy allowed AMD to stay within spitting distance of Intel and their formidable process lead, and their products (historically) have been in the same rough performance range with similar TDPs.
We have now seen the first fruits of this labor with the Trinity APU that was recently released. Trinity fixes a lot of the issues that plagued Bulldozer, and combined it with a solid integrated graphics part based on the HD 6900 series VLIW4 architecture. The first major issue that was solved was that of power. Bulldozer was really rushed to market, and as such it was not completely optimized for power consumption. While the top end part was rated at 125 watts, it fell well below where it was initially expected to be in terms of clockspeed. When the FX-8150 was taken above 3.8 GHz, we saw an impressive rise in power consumption. In my FX-6200 review, when that particular 3.8 GHz/125 watt TDP part was taken to 4.8 GHZ, I saw a 100 watt increase at the wall. Needless to say, the design was a bit “loose” when it comes to power consumption. With Trinity AMD was able to take the time and truly optimize the design for power consumption.
Trinity, on the other hand, proved to be a very adequate CPU in a variety of performance scenarios.
Trinity also provided per-clock performance advances. These advances do not take the Piledriver architecture up to Ivy Bridge performance heights, but it is still a good 8% to 10% faster per thread/per clock than the previous Bulldozer. And it is able to achieve that at lower TDPs. Pretty exciting stuff, right? What is perhaps even more exciting is that Trinity was shown off in working condition a year ago, and was apparently quite production ready in late 2011. AMD held off Trinity production due to multiple factors, but primarily because they had finally fixed Llano production issues and that particular chip was selling very well in both desktops and notebooks. It also made little sense to introduce a new socket at that time, as FM1 was only then just gaining steam.