VM Performance Showing Mixed Impact With Linux 4.15 KPTI Patches

But not all is in bad shape with VMs... Here are the results from the leaner Xeon E3-1280 v5 system with its Ubuntu VM having access to all 16 CPU threads and 12GB of the system's 16GB of RAM. A few different tests were run on this VM while still exploring the performance impact of the KPTI kernel patches over the past two days and also being a bit surprised at first about the lack of change in some of the results.

In the CPU benchmarks, the performance is less noticeable than with the Xeon Scalable platform.

But in cases like Redis, there continues to be a measurable hit to performance.

At least for now it's looking like the performance impact of having KPTI support is more noticeable for more virtual machines than with bare metal desktop class hardware, but we'll see what further optimizations come to this code ahead of the Linux 4.16 cycle and what other insight security researchers and developers may be able to shed on this situation in the days ahead.

With these VM results so far it's still a far cry from the "30%" performance hit that's been hyped up by some of the Windows publications, etc. It's still highly dependent upon the particular workload and system how much performance may be potentially lost when enabling page table isolation within the kernel.

Still exploring this vulnerability now known as Meltdown+Spectre and carrying out additional KPTI Linux patches this week, such as testing with Xen, my other available Xeon servers, and also more enthusiast-oriented tests still coming, so stay tuned for any other discoveries and interesting performance tests.