Despite no official confirmation, it's more than likely that we'll see new eight-core CPUs from Intel in the next month or so, delivering the ability to comprehensively outperform the already stunning Core i7 8700K - our pick as the fastest gaming CPU money can buy right now. The additional benefits for gaming in the short-term may be limited but once the new wave of Ryzen-powered consoles arrive, the need for faster, wider processors in the PC space should become abundantly clear.
First up, let's round-up the various rumours - which actually begins with a cold, hard fact. Motherboard supplier Asrock has attached stickers to its latest H310 motherboards, citing eight-core CPU support. Not only does this confirm that octo-core chips are coming, but that the existing range of Coffee Lake motherboards will support them. Beyond that, we are moving into rumour territory. Chinese site Coolaler has delivered leaked specifications for the three top-end ninth-gen Core processors, citing a new eight-core, 16-thread Core i9 9900K, along with an eight-core, eight-thread Core i7 9700K and a six-core, six-thread Core i5 9600K.
The leaked specs are laid down in the table below, and what's remarkable - and possibly rather suspicious in terms of credibility - is just how fast the new eight-core processors are in terms of clocks. More cores means more silicon area, which typically sees clocks drop, not increase as we're seeing here. Noted German tech site Golem.de has an explanation though, stating that the heat spreader covering is attached to the processor die with solder, not thermal paste. It's well known that this kind of connection transfers heat far more efficiently, and by extension, allows Intel to raise clocks. Regardless, a prospective Core i9 9900K able to run all eight cores and 16 threads at 4.7GHz is something else.
The notion of Intel releasing a Core i7 eight-core processor without hyperthreading has raised some eyebrows, but in terms of raw horsepower, the existing Core i5 8600K (six cores, no HT) offers a small performance uplift over the Core i7 7700K (four cores, eight threads) and in combination with its prodigious clocks, its eight physical cores stands a good chance of beating the current 8700K flagship. Beyond the arrival of the new i7 and i9 though, the ninth-gen Core line-up looks to be a fairly standard refresh to the existing Coffee Lake-S line, with incremental performance upticks in the 100MHz-200MHz range across the stack.
|Leaked Specs (Unconfirmed)||Core i9 9900K||Core i7 9700K||Core i5 9600K|
|Max All-Core Turbo||4.7GHz||4.6GHz||4.3GHz|
Intel's motivation in pushing out the boat on the higher end parts is clear - AMD's strategy of ramping up core counts and frequencies with the Zen core and the recently released Ryzen 2000 series refresh simply offers more value for almost any application outside of gaming. And while its robust financials may not reflect it, the loss of performance from patching the security nightmares of Meltdown and Spectre may also have played a part in motivating Intel to really push hard on its upcoming flagship products.
But from a gaming perspective, do we really need octo-core processors with insanely high clocks? Well, first of all, it depends on your use-case. Many monitors now support 75Hz, 120Hz, 144Hz, 165Hz or even 240Hz refresh rates, meaning that the need is there to run game engines faster than they've ever run before for those invested in these gaming-grade screens. Secondly, multi-platform PC ports in the here and now are based on code that's fundamentally designed to run on the low-power AMD mobile x86 cores found in PS4 and Xbox One - and change is on the way in the console space.
Most console ports can easily run at higher than 60fps on PC, even factoring in the oppressive overhead introduced by Windows, but with next-gen likely to feature Ryzen and possibly two quad-core modules for eight cores in total, games will inevitably become much more complex and the PC versions will see system requirements rise significantly for the top-end experience.
But even in the here and now, the importance of the CPU in gaming is very much under-appreciated - for two reasons. Firstly, the primary limiting factor in games performance is usually the GPU, meaning that for the most part, it's the speed of your graphics card that defines the majority of the experience. However, by extension, it also means that when the CPU is your bottleneck, it's rather difficult to know that this is the problem as the demands placed on your processor vary so dramatically from one moment to the next, even in the same game scene.
Crysis 3: Very High, SMAA T2x
Secondly, something that's bothered me for a while now is the highly limited amount of replicable tests that truly stress-test game engine performance on the CPU. The vast majority of built-in benchmarks included in games are GPU-bound in nature, meaning that many CPU reviews show relative performance between processors being very close indeed when the reality is that frame-rate averages are skewed by sections of the bench where the GPU limit is holding back the CPU. This is why I've spent some time now finding benchmarks - or areas of gameplay - that really stress the processor in order to demonstrate that a fast chip really can make a difference.
So consider the Crysis 3 benchmark above as an example of the challenge and where we need to look at the right data. I chose this game as an example as the engine is many-core aware and previous results have shown that it's very Ryzen-friendly - and not every title is. If you're familiar with our benchmarking widget (only visible on the desktop version of this page), you can play the YouTube video to see frame-rate and frame-time data play out in real-time and you can mouse-over the barcharts for fps data, or click to switch to the more useful percentage differentials. Those viewing on a mobile device get a table with frame-rate averages and lowest one per cent results.
First of all though, it's worth stressing that this is not really a CPU benchmark as such - in common with many games, Crysis 3 doesn't have one. It's an engine-driven cutscene I use that hammers the processor hard and scales beautifully across however many CPU cores you have available, something I discovered by playing the game and observing its behaviour with monitoring tools. Each processor is tested with an overclocked Titan X Pascal at 1080p and 1440p resolutions. At 1080p, the idea is to remove GPU as the limiting factor as much as possible, while at 1440p, the aim is to show how results shift when the GPU is more of a bottleneck - as is the case in most gaming PCs. In both cases, the results are illuminating.
Ashes of the Singularity Escalation: DX12 CPU Test
I'm testing the recently released Ryzen 7 2700X here, a fascinating CPU that effectively overclocks itself to something close to its limits out of the box (in this case 4.0GHz). AMD even gives you a great cooler in the package - it's everything you need really and manually tweaking only gives you a further 100-200MHz max, killing efficiency and requiring a better thermal solution. So in this scenario, the stock performance metric is all we need. Stacked up against the 2700X is the Core i7 8700K at stock speeds and also at 4.7GHz (the all-core turbo speed you get with the CPU plugged into a Z370 motherboard). Both of these products are immensely capable processors and both come highly recommended - the 8700K excels at gaming but you need to buy a great cooler, while the 2700X beats it in more productivity-based tasks and gives you the complete package out of the box.
The Crysis 1440p stats are nicely representative of how a standard CPU benchmark may appear in any given review. On average, the Ryzen 7 2700X is only five per cent slower than the 8700K, similar to the claims AMD made about the 2700X in its reviewers' guide - but this differential is skewed lower because much of the sequence is CPU-bound. But looking at the lowest one per per cent scores, the i7 is actually 23 per cent faster, rising to 33 per cent when running at 4.7GHz. Suffice to say that in gaming, it's the biggest dips in performance that are most keenly felt and why you spend more money on a good gaming CPU. Meanwhile, the 1080p scores see the gap widen in the averages, as you might expect, but the difference between AMD and Intel in the bottom one per cent scores rises further in Intel's favour. In all cases, average and highest frame-rate numbers are stratospherically high, but from the CPU perspective, it's the lowest frame-rates and highest frame-times that actually impact the quality of the actual experience.
I've also included an Ashes of the Singularity Escalation benchmark too. This is an interesting one because it is actually a processor stress test designed to reflect the game under heavy CPU load - precisely the kind of benchmark we really need more of in gaming. Developer Oxide Games tells me that it's four AI players battling each other on a very large map, the only difference from actual gameplay coming from the fact that units can't be destroyed. Oxide also tells me that the potential is there for even heavier load in some scenarios, but this isn't a synthetic bench - it's a stress test designed to put your CPU through its paces without your graphics card skewing the results. Not even the 8700K at 4.7GHz can sustain 60fps in this test and even though the Ryzen 7 2700X's 16 threads are all fully utilised here, it's even further off the pace. It's a reminder that PC gaming offers some use-cases where high-end CPUs can really make a difference, even on a 60Hz displays. For the record, Creative Assembly's Total War games also drop significantly beneath 60fps too, though its CPU utilisation isn't as efficient as that seen in Oxide's Ashes game.
In essence then, factoring out gamers with high refresh rate displays, buying a faster, more capable CPU is mostly about paying more for an overall smoother experience with less hitching and stuttering - basically providing the overhead to ensure your GPU is pushed to the limit, or that your chosen target frame-rate is more firmly adhered to in more taxing games. That's in the here and now. By the end of the decade when next-gen consoles arrive, it'll be about having the overhead on tap to run much more advanced games at 60fps and higher. Consoles define the baseline and they're about to get a big, big CPU upgrade - something that PC gamers will need to match. Right now, the Core i7 8700K remains an astonishingly fast piece of kit, but I awaited the arrival of the new ninth-gen i7 and i9 with bated breath - and it'll be fascinating to see how AMD responds.
Will you support the Digital Foundry team?
Digital Foundry specialises in technical analysis of gaming hardware and software, using state-of-the-art capture systems and bespoke software to show you how well games and hardware run, visualising precisely what they're capable of. In order to show you what 4K gaming actually looks like we needed to build our own platform to supply high quality 4K video for offline viewing. So we did.
Our videos are multi-gigabyte files and we've chosen a high quality provider to ensure fast downloads. However, that bandwidth isn't free and so we charge a small monthly subscription fee of £4.50. We think it's a small price to pay for unlimited access to top-tier quality encodes of our content. Thank you.Support Digital Foundry