AMD clarifies discrepancies in gaming performance between official data and reviews

The company says that Windows patch will improve gaming performance, explains why ‘admin’ account was used for internal testing.

AMD published a blog post discussing the disparity between the official gaming benchmarks for the Ryzen 9000 series and those reported by reviewers. Unfortunately, the data did not match what AMD presented, and in many cases, gaming performance was not as high as expected. Several factors contributed to the discrepancies, including the following.

AMD’s test suite may differ from what reviewers used. The company relies heavily on built-in benchmarks in most of its automated tests, which may not accurately reflect real-world usage. Additionally, AMD is using specific Intel configurations, including DDR5-6000 memory and Intel’s new Default Settings (Baseline), which may not be representative of all review systems.

Ryzen 9 9950X Performance, Source: AMD

Moreover, AMD uses Admin mode for its automated testing on Windows 11 24H2. This mode enables new branch prediction code optimizations that were not present in the systems used by reviewers. These optimizations are only available in the latest version of Windows 11, which is currently accessible only through the Insider Program (Build 26100). In other words, gaming performance is expected to improve once the newer version is released.

Here’s a full quote on the topic:

Why AMD-generated gaming data differs from reviewer data: For our Ryzen 9000 launch, AMD internal labs generated data that showed a 9% average generational uplift in 1080P gaming versus Ryzen 7000 Series, and an average of 6% higher performance across more than 30 games when compared against the competition’s best. Not all reviews are seeing these results, and this reflects the complexity of high-performance PC testing today given the number of system and software variables. There are several factors specifically creating these differences in some Ryzen 9000 reviews:

• The AMD gaming test suite includes a broad set of esports, AAA, and popular older games, which are a combination of CPU- and GPU-bound titles. Game performance conclusions can be influenced significantly by the makeup of the test suite.
• AMD tested Intel configurations using comparable DDR5-6000 memory as well as Intel default settings-baseline power profile which can have a small impact on gaming performance.
• AMD also tests with Windows Virtualization-based Security (VBS) enabled. This is the default Windows behavior and Microsoft recommends activating VBS to improve security, however it can affect gaming performance.
• The “Zen 5” architecture incorporates a wider branch prediction capacity than prior “Zen” generations. Our automated test methodology was run in “Admin” mode which produced results that reflect branch prediction code optimizations not present in the version of Windows reviewers used to test Ryzen 9000 Series. We have a further update on accessing this performance for users below.

AMD has shared a chart that explains the difference between the 22H2 and 23H2 versions of Windows 11. The patch is said to deliver a 3% to 13% increase in gaming performance. The company remains committed to its claims of a 10% improvement in productivity, a 25% improvement in AI workloads, and a 5-8% improvement in gaming performance over the Ryzen 7000 series.

Reviewers can now test AMD’s claims using a pre-release preview of Windows 11, which is expected to be released to everyone soon. “We’re collaborating with Microsoft to roll out this optional update to all Windows 11 users soon,” said AMD in a press release. It’s worth noting that AMD confirms these branch prediction optimizations will not only affect Zen 5 processors but also Zen 4 and Zen 3.

Source: AMD via KitGuru