Intel’s Kernel Security Bug: What it Means for PC Gamers

Intel headquarters

The PC hardware world is exploding this week over news that the last 10 years of Intel CPUs all contain a massive security flaw, forcing a redesign of the kernel software in Windows, Mac OS, and Linux. This issue is more serious than most security flaws, because it is connected with how the hardware talks with the OS, which means that patching it may impact performance.

Reports have been stating that Intel CPUs may suffer a 5-30% drop in performance after the major OS manufacturers issue their patches. A drop of 5-30% certainly sounds both large and scary. But is it really going to be that bad in real-world situations, or is this alarmist sensationalism? Let’s find out.

The first real-world benchmarks were done by Phoronix for Linux-based systems. Those painted a positive picture, at least for Linux gamers. The tests were done with an Intel i7-8700K, and actually resulted in some games performing slightly better after the patch when compared to pre-patch.

In Phoronix’s Linux benchmarks, the most heavily impacted areas were servers and virtual machines, with performance degradation ranging anywhere from 0% to well over 30%.

One pretty troubling benchmark is speeds of super fast SSDs, such as the Samsung 950 Pro. In one specific test, file transfer rates fell by just over 50% after the patch. For the SATA SSD tested, the performance impact was essentially nonexistent.

On Linux-based systems, there is no noticeable performance impact in games. In productivity- and especially SSD-bandwidth related benchmarks, the performance does see noticeable degradation, though. The i7-8700K system is running an extremely fast Samsung 950 PRO NVMe SSD, while the 6800K system is using a slower SATA 3.0-based SSD. It appears that performance degradation is noticeable only when data transfer speed bottlenecks are removed. Source: Phoronix

So, in reality, on Linux-based the performance impact of the patch will only be noticeable when using extremely fast NVMe SSDs in specific workloads. What is the story with Windows 10 though?

Just today, Microsoft issued a hotfix for the Intel kernel issue, and HardwareUnboxed have already done the basic testing. It appears that the situation is very similar to what we saw in the Linux tests: The performance impact is noticeable only in specific use cases with extremely fast data transfer speeds. In gaming, however, no impact can be seen.

And thankfully, in most cases, the SSD performance impact showed relatively small (~3-8%) performance impact at worst on transfer speeds in these tests.

During sequential read/write tests, we can see how the patch impacts performance using a Samsung 950 Pro. With larger sized data chunks the performance delta increases. Source: HardwareUnboxed

 

Gaming performance, on the other hand, has not been noticeably impacted. Battlefield 1 gives a great overview of the general performance. Source: HardwareUnboxed

In summary, this issue has been blown out of proportion. The performance impact is there, but only in specific use cases and on extremely high-end hardware. If you’re impacted by this, you probably already know it.

For gamers though, nothing really changes. Intel is still the performance leader when it comes to gaming and single-threaded performance. The one interesting thing is random read/write performance on high-end SSDs for content creators. How will this specifically impact video production and scrubbing? Most likely not much, due to bottlenecks in other parts of the system. That said, if you feel your system may be or has been impacted by this, let us know!

How will the Intel kernel bug affect Logical Increments recommendations?

A lot of readers are already asking how this will impact the PC hardware recommendations on our homepage. Based on initial benchmarks, we have no plans to change our current recommendations. Of course, we will keep watching as new benchmarks release, and make adjustments if competing Ryzen CPUs acquire a performance lead.

All of our recommendations are based on real-world performance, so if performance drops in a way that makes another CPU better, you can trust that our recommendations will change accordingly.

These initial benchmarks do not cover much in the way of older Intel CPUs or lower-core count CPUs, such as the i3 lineup. So, we may learn additional information that causes us to change recommendations as it comes to light.

We anticipate having more coverage and analysis of this issue as it develops over the next few weeks. But for now, rest assured that initial benchmarks indicate that the sensational hype seems to be overblown in the early reporting of the bug’s patch.

Sources: