AMD should guide tech reviewers to evaluate Zen4 thoroughly and consistently.
AMD should guide tech reviewers to evaluate Zen4 thoroughly and consistently.
Years back I shared my thoughts on how tech critics ought to assess Zen 1 in the AMD subreddit. The AM5 release is coming soon, and we should hope it improves upon Zen 1. We should encourage reviewers like Linus and creators such as level1techs and hardware unboxers to read this. Keep in mind that this piece might appear biased toward Ryzen. AMD faces marketing challenges. As gamernexus noted, focusing only on the subreddit where people favor flashy emojis or vague 'jebaited' posts isn't sustainable. When it comes to gaming benchmarks, prioritize 1080p and possibly 720p. Emphasize high graphics settings since these games demand significant CPU power. Be sure to include game version, driver, Windows, and DirectX details. For single-player titles like GTA 5, Total War, Civilization 6, Watch Dogs, and Red Dead Redemption 2, note that reviewers often overlook multiplayer titles. Test games like Fortnite, PUBG, Fortnite Battle Royale, CS:GO, Rust (with intense raids), and Arma 3 in multiplayer mode to evaluate frame consistency and micro stutter. For multiplayer experiences such as Battlefield, Planescape, CS:GO, and Arma 3, Intel’s efficiency cores tend to handle micro stutter better unless it's CS:GO. In MMOs like Warcraft, Eve Online, Big Battles, and Elder Scrolls Online, older games often relied on two-core setups but now struggle. For intense destruction-focused games like Destruction, Battlefield, or Red Faction Remastered, CPU can become a bottleneck. Pay attention to visuals during heavy destruction sequences, as some hardware may reduce animations or effects to stabilize frame rates. Make sure to properly configure games and software—games such as ARA, Arma 3, and certain older titles require specific settings for Ryzen 16 24 threads and Intel efficiency cores. Monitor GPU clock speeds too, since NVIDIA Boost 3.0 can cause inconsistency in some setups. Windows 11 with Rebar/SAM and HAGS might deliver the best results, while Windows 10 could lag behind. For graphics technologies, Ray Tracing deserves special benchmarking in Cyberpunk Tomb Raider, especially with DLSS. Reviewers should also consider games like 3DMark, Bright Memory, Boundary, and NVIDIA demos. Performance can vary widely depending on hardware capabilities. When it comes to destruction-heavy titles, ensure you check for visual fidelity and stability. Don’t forget to test game modes like Principled Technologies’ Ryzen Master, which may affect benchmark accuracy. Monitor GPU clock, cache usage, and memory configurations—especially with games that demand high cache or specific turbo modes. Game settings such as PBO2, XMP, or AMD’s turbo modes can greatly influence outcomes. For AI-driven games like Total War: Warhammer 40000 or Civilization VI, massive AI presence can stress systems; consider how these affect frame rates and consistency. Lastly, keep an eye on cooling solutions—Wolfhawk, Deepcool, Hyper-212, and AIO coolers all play a role in maintaining performance under load.
And reviews like this are precisely why opinions or AMD should consider insights from everyday users. I’d like a reviewer to try something just as a regular person would: set up a PC, install Windows 10/11 drivers, run a game and see how it performs. Most buyers don’t understand all the technical details about memory speeds or disabling vertical sync—those things you mentioned. Regarding choices, look at Steam Hardware Surveys to see what the majority use. They usually have 6-core CPUs and mid-range graphics like the RTX 2060/1650. About 73% still run Windows 10, unless the AM5 chipset demands Windows 11 features. Reviewers should compare both versions, especially since some users might avoid TPM or face changes in Windows 11. Many have 1080p displays, with around 20% using 1440p, so testing at 1440p could be useful for future trends and extra CPU load. Also, dual-monitor configurations are common, so those should be included in the test.
I think that's an interesting comment to throw at the end of a LONG list of things that "should" be done. The big problem is that there is a practically infinite number of variables, but people end up wanting a single number to show if A is better than B or not. If you focus on one detail, you lose context of the wider field. If I were to buy a new purely gaming focused system today, I'd probably get 5800X3D, but for a balanced general purpose system, 12700k would be far more interesting. Imagine if a video review covered all the points listed. It would go on for hours, not to mention how long the testing would have to be to even have the data to support it. Day 1 reviewers will have limited time and thus limited scope. More points will be revealed over time. Not forgetting there's often many bios/driver updates right up to and beyond launch date, so results will be a moving target early on. And as I have said many times before, there are often a lot of non-performance related considerations that will also be taken into consideration when deciding what to buy. Current system is not perfect by any means, but hard to see it being much different. Launch day reviews will be a quick snapshot to get ball park info for wider context. Longer term reviews will focus on more details. Some details will be so niche as not to be covered at all. Personally I find it annoying no one focuses on AVX family FP64 performance. I got most of that from my own testing, but had to get indirect data on Zen 3 which I didn't own until I got one in a laptop much later, and I'm absent data on Alder Lake.