Update on Ryzen branch prediction enhancements and their impact on DX9 performance.
Update on Ryzen branch prediction enhancements and their impact on DX9 performance.
I installed the newest Windows Update which seems to include a 24H2 Ryzen patch. Then I picked up Borderlands 2 again. The experience has improved significantly. My frame rate has stabilized at around 168fps, staying just under my monitor’s 170Hz refresh rate. It was a chaotic experience before, sometimes dropping to 80fps, but now it runs smoothly. I haven’t tested with 24H2 for DX9 titles, maybe because they’re outdated. Could anyone else verify this change? I wish I had done a benchmark before and after, but I wasn’t anticipating such a big boost.
EDIT: For context, my CPU is a Ryzen 9 5900X, GPU an RTX 2060 Super, RAM at DDR4-3333 CL16. I’ve also noticed my game’s CPU usage has dropped from double digits (15-20%) to almost single digits.
I wasn't surprised at all. My gaming experience was really let down with the 5950X, as the frame rate fluctuated a lot. It was interesting that it improved with the 7800X3D. I think whatever issue existed was fixed by V-Cache, but it might have masked a more serious problem.
I can't run a Firestrike or Cb benchmark for you. However, I can help compare performance improvements in Overwatch versus other games if you'd like. It's not necessary to do anything extra on Windows 11, and I can guide you through the process if needed.
I've already made the updates, and it's clear this change enhances gaming speed. On average, performance gains are around 10% across many titles. The fix from 24H2 was recently released as a standalone Windows 11 23H2 update. It's currently available only for Windows 11. This update likely positions Windows 11 ahead of Windows 10 in overall gaming performance, though I haven't seen a direct comparison yet.
CPU activity reported in Windows reflects processing time. Performance gains indicate it may have been idle longer than actively working. Further investigation would be needed to verify this. Beyond my ability or interest in exploring more. Both you and the original poster have two CCD components, and I question whether that matters more than the 3D cache. AMD's strategy for scaling Ryzen cores past eight has been challenging because of core separation. I consider eight cores sufficient for gaming and chose the 7800X3D, though I’d prefer twelve if available as a single CCX. In my view, it’s not worth the extra hassle to use multiple CCDs for a gaming-focused setup. I’m thinking about swapping in a 9950X for my productivity system, which is a different kind of optimization.
Consensus reached, though gaming on the 5950X involved trying to disable a CCD which didn't yield results. It seems the unit was faulty, likely causing lockup when idle but remaining stable under stress until it eventually failed to boot. Through testing, the 5950X showed significant gains after an update, though accuracy remained limited in synthetic benchmarks. For reference: https://browser.geekbench.com/v6/cpu/com...ne=5779889 and https://browser.geekbench.com/v6/cpu/com...ne=7548645
I'm interested in my 7800X3D. I'd like to run some initial tests before implementing that update, as it would take a lot of time and there are other priorities I'd prefer to focus on right now. That's why I'm checking out other reviews. No significant improvements in non-gaming areas and only minor gains in low-resolution gaming don't seem like a strong enough reason to go through with this.
I'm not overly concerned because with the 7800X3D I've experienced top-tier gaming performance I've ever had on my PC. That means minimal stutter or random drops in frame rate. This could significantly impact games that don't rely on V-Cache, potentially making them compete more effectively with V-Cache versions. It raises the possibility that AMD might have anticipated this need and included it to boost sales of X3D chips. While it's improbable it would improve every game, it might lessen the number of titles that truly benefit from extra cache. The 5950X stands out since it's the model I use for Topaz products or other software that can't be run on Linux. It offers a major upgrade on the Steam Deck in Windows, now closely resembling SteamOS. Here are some useful links for comparison:
- https://browser.geekbench.com/v6/cpu/com...ne=3890068
- https://browser.geekbench.com/v6/cpu/com...ne=3890374
This decision stemmed from wanting a premium gaming experience. While not the fastest option, it offered better value compared to pricier alternatives. I aimed for optimal memory performance at the 6000C30 setting, opting for dual-rank modules despite the higher cost. It's a bit disappointing that AMD hasn't shared more details about their design choices, but it seems they're addressing bottlenecks in CPU operations. The L3 cache is positioned further away, acting as a separate segment of the data flow. Improving one area can shift challenges to another. The 3D cache helps when the memory isn't enough for heavy workloads. All these factors are weighed together during the design process. A comment from the Y-cruncher writer, who got early access to Zen 5, mentions conversations with AMD representatives at a conference. More info is available on their forum post.