With 2015 over, it’s time to look to the technology of 2016. Today’s topic: what’s in store for that loneliest segment of the tech world, fans of AMD’s processors?
That’s because AMD hasn’t released anything processor wise, at least not worth celebrating, in over four years; that changes this year with the release of AMD’s hotly anticipated Zen architecture.
It’s been a trying time for AMD these past few years, thanks to decreasing market share across the board in consumer and business sectors, as well as decreasing sales and lower budgets for R&D.
No wonder they named their new X86 CPU architecture “Zen” – they’ve needed to remain quite zen ever since the dismal release of their Bulldozer architecture.
Having to suck it up and make do with an architecture that couldn’t compete versus their immediate Intel equivalents was painful enough, but it has to hurt to see customers purchasing Intel i3 CPUs over AMD’s Phenom II X6 chips, even when the purchase is made for the sake of single threaded performance.
Now, after a few CPU revisions and budget based APU refashioning, the world will mostly say goodbye to the Bulldozer based architecture in 2016, as there are still a few holdouts, like the Excavator APU due later this year.
To usher in a new era, it’s only fitting that we list some of the reasons why you should get excited for AMD’s Zen CPU, the first of which is that AMD management gave a lot of freedom for their engineers to design Zen, which is a hopeful sign in itself when engineers, and not bureaucrats, are in charge.
From the get-go, AMD have made some bold claims about the performance increases their Zen CPU will achieve, promising a 40% increase in instructions per cycle (IPC) over their yet-to-be-released Excavator APU (their last Bulldozer-based architecture).
AMD seem confident in this benchmark, and we’re swayed to at least give them the benefit of the doubt since they are moving away from Clustered Multithreading (CMT) which was the bedrock of Bulldozer, and towards Simultaneous Multithreading (SMT), which is basically AMD going back to basics and implementing their own version of Intel’s Hyperthreading technology.
This means that AMD has pretty much gone back to favouring single threaded performance over multi-threaded performance, which although might be too little too late, it’s a gambit they have to take if they want to remain in the X86 market at all.
More so, for many people, AMD at best bent the truth about Bulldozer performance at the time, and at worst, straight up lied about Bulldozer being an “octo-core” CPU.
Suffice it to say, if AMD has any more goodwill left, they would do well now not flush it down the toilet with overeager performance figures or marketing hype.
Late last year, a leak by a former AMD employee merely said that AMD had finished testing their Zen CPU and it has “met all expectations” with “no significant bottlenecks”. Unsubstantiated it may be, it’s also mildly promising considering the shift in architecture design to favour SMT.
Now, although we are yet to even measure Excavator to project the IPC performance increase of Zen, there is a measurement we can use to help position Zen in the grand scheme of all things CPU, and it’s all thanks to Intel.
The last great generational advance in IPC was Nehalem to Sandy Bridge, with Sandy Bridge decimating the performance and price barometer, and Intel has been scurrying about in the shadows of their great accomplishment ever since, as victims of their own success.
Only recently has Intel’s Skylake been a substantial enough reason for gamers to retire their Sandy setup and move over to Skylake, with around 25-30% faster IPC figures over Sandy Bridge.
Looking at Digital Foundry’s comparison between the 2600k and 6700k on the majority of gaming titles, the performance differences via frame rates and frame times are not that severe, although in CPU dependent games like GTA V the performance gap is much bigger.
However, if we were to excuse GTA V and extrapolate those numbers and the claimed 40% increase in IPC for Zen, then we can make some guesstimations that Zen should fall at the very least within Ivy-Bridge/Haswell performance ranges, which is a far cry from what Bulldozer and its revisions have ever achieved.
Up to now, it’s been a healthy dose of speculative guesswork on performance, projection and a little underdogism, so what do we know about Zen that might incite educated choices to move from Intel in the future?
Well, we know that it will finally be on a new 14nm FinFET manufacturing process, which is comparable to Intel’s Skylake node. Zen CPU’s will use a new AM4 platform, which offers support for DDR4.
The great thing about the AM4 platform, besides finally bringing AMD motherboards up to speed with features like native USB 3 support, is that AMD will once again return to their friendly compatibility roots for consumers.
Instead of the maddeningly confusing AM3+,AM1, FM1, FM2, FM2 + motherboard options we’ve had in the past 4 years, both AMD Zen or “Summit Ridge” CPUs and future “Bristol Ridge” Excavator APUs will be supported on the AM4 platform.
This makes drop-in upgrades or downgrades much easier for consumers who might not feel the need to lock themselves into the Intel tick-tock master plan.
The stakes have never been higher for AMD. In fact, Zen may be the CPU that either saves the company or sinks it in the X86 marketplace.
Their renewed commitment to a high performance X86 processor is admirable, but also wrought with potential pitfalls, some created by AMD themselves and others laid down by the current market dominance of Intel.
If you excuse our underdogism, even if many of us at MyGaming are using Intel CPUs, it seems like the CPU marketplace needs a shakeup in terms of pricing, and if AMD can become competitive again with Zen, we might finally get that. Once Q4 2016 rolls around, we’ll know what’s in store for us.