We may earn a commission when you buy through links in our articles. Learn more.

Intel exec throws shade on AMD “adding core count just because you can”

A post on core-scaling in games points out that going beyond 8-cores is irrelevant for games... forgetting 10-core Comet Lake...

Intel 8-core CPU

Intel’s chief performance strategist, Ryan Shrout, has been throwing thinly veiled shade on AMD’s push for ever higher CPU core-counts on the desktop… and inadvertently on his own company’s marketing team and products. He has posted a piece to his Medium account under the guise of analysing core-scaling for PC games, and the data itself if very interesting. Written by an independent journalist, something he once was, there would have been some definite value in doing such analysis.

Basically Shrout looks at relative gaming performance using a Core i9 9900K CPU, locking it to 4GHz and testing it in four, six, and eight-core modes. There is a definite bump in performance in the six games he’s tested, with half showing between a 20 – 30% frame rate increase at 1080p. But the move to eight cores shows far less of a boost, with three of the tested titles seeing no benefit at all.

His point is that straight-line CPU frequency and single-threaded grunt is what today’s game engines need and that “adding core-count just because you can… doesn’t result in better performance.” That looks a lot like some shade being thrown at AMD just before it releases its first 16-core gaming CPU, the Ryzen 9 3950X, the chip that it’s bin-sorting to be the AMD’s “best gaming part.”

The problem is that Intel has been marketing it own multi-core processors specifically at gamers, with the proviso that everyone wants to be a streamer today. And if you want to stream you’ve got to have a high-end, high core-count CPU to manage both the game and the video encoding shizzle at the same time.

It’s also difficult because if Shrout were doing this in his previous independent journalistic role he would have been able to actually figure the competition into his testing too. Seeing how games scale across multiple cores, and across different CPU architectures would be something to see. And may even swing further in Intel’s favour.

Drop an eight-core Ryzen 7 3700X into the mix to see how that matches up would be fascinating, from a performance point of view, but also given the relative pricing of the two levels of Intel and AMD processor.

“For gaming, 8-cores is the optimal spot for performance scaling in modern PC gaming,” says Shrout, “and at 8-cores we get the best frequency scaling out of our Coffee Lake products and 14nm process technology today. And clock speed is what feeds the hungry primary threads of game engines today!”

With the new Core i9 9900KS Intel has released a chip which is running at 5GHz on all cores, all the time. That should be great for gaming, right? Unfortunately it barely makes a difference. Sure, in raw numbers you can argue that it is the fastest gaming CPU around, but I’m not sure how much that matters these days. Especially when it costs well north of $500 for the special edition part.

He also throws in an 18-core chip to test what the core-scaling is like shifting up to 16 cores – the same number as in the upcoming AMD Ryzen 9 3950X… make of that what you will – and he conclusively proves that Intel’s own i9 9980XE is pointless for gamers. Shifting to 16 cores some games actually perform worse.

YouTube Thumbnail

“Adding core count just because you can,” he concludes, “without a corresponding increase in sustained frequency and architectural design decisions necessary to feed these cores (like low latency memory systems), doesn’t result in better performance. The software engines that power games across the PC ecosystem scale best with frequency and IPC, and Intel plans to lead in this space for years to come.”

He does have a valid point about frequency and IPC, those are really important things for today’s game engines, and Intel leads in those areas. But he can’t pretend Intel isn’t doing the same thing re. core counts. After all are we not factoring the upcoming Intel Comet Lake chips into this?

Y’know, the range which will introduce a 10-core processor, just because Intel can, likely without a corresponding increase in sustained frequency or architectural design decisions necessary to feed those cores?

The other struggle is that when you’re purely talking about gaming you have to remember that the even the best CPU actually has very little impact on gaming performance these days. With resolutions pushing beyond 1080p it’s very rare for games to be CPU limited to such an extent that frame rates are affected. It’s far more likely that the performance bottleneck for your PC surrounds the speed and architecture of your GPU.

I understand testing at 1080p, that’s how you highlight the differences in CPU design and performance, but it’s also not necessarily indicative of real-world performance. Those in glass houses shouldn’t throw shade, someone might need to remind Intel’s chief performance strategist of that at some point…