AMD’s latest 9000-series CPUs have sparked a wave of discussion, as their performance has not significantly outpaced the previous 7000-series chips, despite a notable increase in price. In a recent blog post, the company provided insights into the discrepancies between its own gaming performance data and that of independent reviewers. Initially, AMD had touted a “9% average generational uplift in 1080p gaming” compared to the Ryzen 7000 series, alongside a 6% performance advantage over competitors in over 30 games.
Understanding Performance Metrics
However, the conversation took an intriguing turn when Hardware Unboxed referred to a “hidden Administrator account” that allegedly enhances performance. AMD suggested that this account should be utilized, as the new Ryzen chips may not perform optimally under standard user accounts, particularly for bursty workloads. This raises a critical question: why would performance testing be conducted under such a hidden account, given that the vast majority of users operate under standard profiles?
One might cynically speculate that this approach could inflate performance metrics by bypassing certain system security protocols. Alternatively, a more generous interpretation could suggest that AMD underestimated the impact of using this account for initial benchmarks. Nevertheless, it seems peculiar for anyone to adjust Windows settings to enable a hidden Administrator account without a clear rationale.
Interestingly, while AMD has made some reasonable decisions regarding testing protocols—such as maintaining Windows Virtualization-based Security (VBS) during evaluations—there are still nuances to consider. Tech YouTuber Level1Tech pointed out that running a game under the hidden Admin account could effectively negate the VBS overhead, complicating AMD’s assertion that VBS was enabled during testing.
This leads to further speculation about whether the optimizations related to branch prediction in Windows profiles might mimic the effects of disabling VBS. Such a scenario would render the choice of testing under a hidden profile even more questionable, as it does not reflect typical user experiences.
Ultimately, it appears that AMD recognizes the shortcomings of its testing methodology, as evidenced by its revised gaming performance assessments. Fortunately, there is now a pathway to leverage improved branch prediction code without altering Windows profiles. AMD has announced that this optimized code will be available in Windows 11, version 24H2, accessible through the Windows Insider Program or via an ISO download.
While the anticipated performance boost from this update is modest—estimated at only 2-3%—in a landscape where the 9000-series chips have shown minimal enhancements over their predecessors, every incremental improvement could be valuable to users seeking the best possible performance.
