performance under the microscope

We have compared the two versions of the Microsoft operating system on numerous benchmarks and video games. Is there really a very marked difference?

Thunderclap, with the release of Windows 11: “The new system from Microsoft reduces in-game performance by 25-28% “Headline some of our English-speaking colleagues. In question : reinforced security measures, in particular the use of VBS (Virtualization Based Security). As the name suggests, virtualization-based security takes advantage of hardware virtualization (Intel VT-X and AMD-v) to create and isolate a secure area of ​​memory, detaching it from runtime. normal ”from Windows. Clearly, this area allows code to be executed without any vulnerabilities or malicious attacks compromising the integrity of your data and “contaminating” the operating system.

Still among the advanced security features, HVCI (Hypervisor-Enforced Code Integrity) relies on VBS to strengthen code integrity control. The technology aims in particular to check the drivers, binaries and system files before they start, again in a virtual memory managed by the Windows hypervisor, in order to prevent any malicious code execution. During its presentation, Microsoft indicated that the new security features of Windows 11 (TPM 2.0, VBS and HVCI) will help to “Reduce threats by almost 60% compared to Windows 10”.

To read also | Windows 11: Guided Tour of Microsoft’s New System

Windows 11 and VBS / HVCI: the real impact on performance

The technology is not new and is already in the x64 Pro edition of Windows 10: pull down the Start menu and click Settings> Update & Security> Windows Security> Device Security> Kernel Isolation> Kernel Integrity memory to check its status. What is new, however, is the fact that this feature is enabled by default on all new PCs compatible, that is to say for the purchase of a new computer pre-equipped with Windows 11.

If you are performing a “simple” update from Windows 10, VBS and HVCI should not be enabled (you can check this in Settings> Privacy & Security> Windows Security> Device Security> Kernel Isolation> Kernel Integrity> Memory). On the other hand, if you install Windows 11 “clean”, without updating and starting from scratch, it is VBS without HVCI which should be activated.

Faced with the swelling controversy, we wanted to verify in practice the performance of Windows 11 by comparing it to that of Windows 10. Our test platform is based on an AMD Ryzen 9 3900X processor and an Asus ROG Strix X570-E Gaming motherboard, with 2 × 8 GB of DDR4-3600 G.Skill TridentZ CL16. We installed Windows 10 and Windows 11 on a 1TB PNY CS3030 SSD (in M.2 NVMe) and our machine is running with an AMD Radeon RX 6700 XT graphics card, a Thermaltake Toughpower iRGB Plus 1050 W power supply and a BeQuiet cooling kit! Pure Loop 280.

We ran 13 classic CPU benchmarks in Windows 10 (without VBS or HVCI), then compared them to three scenarios in Windows 11 (without VBS or HVCI, with VBS but without HVCI and with both technologies enabled). Except with 7-zip compression, where the difference reached 25.62%, most of our tests show a relative similarity between the performance of Windows 11 and Windows 10. It should still be noted that out of the 39 tests run under Windows 11, only three results are better with the new Microsoft system. If we take into account the results of 7-zip, which weigh heavily in the balance, Windows 11 is on average -2.84% (VBS and HVCI off), -5.81% (VBS on, HVCI off) and -6.49% (VBS and HVCI on) slower than Windows 10 (VBS and HVCI disabled). By removing the compression, the difference falls to -0.72%, -3.96% and -4.40% respectively.

Image 1: Windows 11 vs Windows 10: performance under the microscope

We then continued the tests with 11 recent games as well as the various 3DMark benchmarks. We have primarily tested in Full HD, with settings chosen for stress in priority the CPU and not the GPU (all the parameters appear on a case-by-case basis, in the graph below). The results tighten markedly between the two versions of the operating system, which allows us to state that, contrary to what the press headlines suggested, no Windows 11 does not strongly penalize video games. On average, the performance gap is -2.02% (VBS and HVCI on), -2.46% (VBS on, HVCI off) and -0.71% (VBS and HVCI off). If we look at the results of each game individually, we see that in many scenarios the difference is less than 1%, which is within the margin of error. We find the most significant differences with Far cry 5 (-8.40%), Ashes of the Singularity (-7.04%) and Shadow of the Tomb Raider (-6.85%). The most surprising thing is that, unlike the CPU benchmarks, this is the scenario here with VBS enabled but HVCI disabled (i.e. the default mode after a “clean” installation of Windows 11, without updating from Windows 10!) which is the most penalizing.

Image 2: Windows 11 vs Windows 10: performance under the microscope

Leave a Comment