This update appears to target Intel CPU performance improvements.
This update appears to target Intel CPU performance improvements.
I might be a bit outdated, but I only came across this update yesterday. KB4497165: Intel microcode updates. It seems optional now, and it’s version 4. I ran some basic tests to see if the performance drop on my i7-4790k was real—turns out I didn’t even notice the 4790k in the patch notes. (Other Haswell chips like 4770k were mentioned too.) I still installed it, and here are the results: (see attached image). Just a reminder, I only tested once. I also have the newest BIOS for CPU microcode on Windows 10 1909. Before “Windows 10.0-KB4497165-v4-x64,” a fresh reboot scored 1908. Using “InSpectre.exe” to turn off Spectre & Meltdown gave 1930, then re-enabling it brought the score back to 1946. Whatever this means, I’m okay skipping any patches for performance reasons. It’s labeled V4, and Microsoft suggests reinstalling it if needed.
It seems you're referring to a term related to precision. Could you clarify what you mean by "margin of error"? I'm here to help with your question!
Yes, but your benchmarks haven’t shifted much. This just addresses the performance drop you’re seeing. With an i7-4790k, it won’t exceed the stated speed limits because my system is using standard DDR3 1600. Since it’s not slower now, I’m not concerned about bugs or patches—game on!
You don't need to be concerned about the performance drop from the updates—it's very small.
The evaluations are complete. Several trials were executed, with at least three repetitions performed. The highest scores observed in theory suggest factors that may hinder benchmark performance, yet none appear to enhance it. As runs increase, results approach the theoretical limit. The system runs Windows 10 version 1909 with February updates applied. Earlier hardware details indicate an i5-4570S processor and a motherboard BIOS last refreshed in 2015, meaning microcode updates for Spectre/Meltdown protection aren't available. Prior to Cinebench R20, scores were 1195 P95 29.8b6 with 4096k FFT; after the update, they improved slightly to 1209 (+1.2%) with minor adjustments (29.8b6 4096k FFT: 122.62, -0.3% 128k FFT: 8016.93 (+0.9%).) A subtle variation was noted, likely due to repeated testing rather than a significant change. Neither of the cases showing ~1% gains were heavily influenced by RAM speed. Cinebench shows minimal reliance on memory bandwidth, and Prime95 128k FFT fits comfortably within most CPU caches. Prime95 4096k FFT exceeds typical cache sizes, making it more likely constrained by RAM speed for slower processors (like dual-core CPUs). This could be the main constraint, or the update might slightly reduce RAM efficiency. I also checked the Get-SpeculationControlSettings PowerShell tool; one entry changed post-patch, but its impact is unclear. OS support for MDS mitigation remains disabled in both versions.
They disabled MDS mitigation since it was vulnerable to attacks, and now it’s back in place because the updates improved performance. I’m assuming they thought the new patches were more effective. The MDS issue was related to a specific vulnerability, but I didn’t go through the full details. For clarity, my previous i7-2600 performed well with my RX 580, and the newer i5-4570 handled it much better.