The RTX 2080 Ti requires a compatible CPU to run games smoothly.
The RTX 2080 Ti requires a compatible CPU to run games smoothly.
Conducting tests at ultra is logical. Using my own presets leads to a major issue, making it hard to repeat them later. Ultra is much simpler to recall.
Don't measure against past data from different times since updates and system issues can skew results. It's better to focus on testing in Ultra and Low configurations without complications. The CPU performance indicator in Ultra isn't reliable, even at 720p, often giving misleading or "limited" outcomes.
The tests are meant to show the progress you make through upgrades. That’s why they’re spaced out over months or years. These changes aren’t about the original games you play; they focus on your building projects since you often exceed what your hardware can handle. With these upgrades, it’s impossible to know instantly if a part is an improvement unless you purchase and construct yourself.
I see, but there are still some points I’m unclear about. You only check for GPU performance? How would you confirm a CPU upgrade? For example, your 5800X is clearly limited by the 3080Ti in 1080p Ultra. Even switching TSAA on or off affects performance by about 6% in this test, despite both settings being at their lowest. If you run the analysis, it might help pinpoint where the speed differences originate.
Interesting note: 8th and 9th generation CPUs perform similarly in gaming, even though the latter has more cores. The 5800X actually has a larger L3 cache, which could point to bandwidth issues rather than CPU speed. Higher core clock speeds require more bandwidth, especially when the cache is limited. Under normal conditions, a 5800X should be roughly 10-15% faster than a well-tuned 8700K in games. In your case, it was 27% faster—way beyond what’s expected.
@Mister Woof, for those curious about Alderlake’s speed:
The challenge with evaluating a game such as SotTR at 1080p with the lowest configuration is that it moves far beyond typical expectations. Most players won’t play at such low resolutions unless they’re experiencing performance issues like dropping below 30 frames per second. While extremely low settings can reveal CPU performance variations, they eventually lose relevance. Upgrading from 9600K to 5900X helped me compare several titles with a 2060 Super. For the configurations I actually employ, differences were minimal. I also tried Tomb Raider 2013 at both 1440p High and 720p Low. At the lowest setting, the 9600K averaged around 450fps and hit peaks above 600. The 5900X averaged nearly 700fps, reaching almost 1100. Does that suggest the 5900X is nearly twice as fast as the 9600K? Not necessarily. Looking ahead 15 years, if someone on a budget checks used 9600K or 5900X systems paired with an RTX 9050 Ti for a modest rig, the 5900X might indeed outperform significantly—but that’s not the year 2036. Generally, most games are optimized for GPU performance, with eSports exceptions being the rule rather than the exception.
I'm a high-resolution gamer and at that scale none of the processors are upgraded from the 2018 i7 8086k. I can even include my i9 10900kf in that list. The more cores the CPU has, it feels noticeably better, though this might just be thanks to extra cache. For the older models I frequently use saved game files from modded titles or simulations that become unplayable. I began this practice back in 2000 with MS Flight Simulator, and it's been a long journey. I add features like enhanced airports, weather effects, and aircraft until the game becomes unplayable. Then I save it and test it again after upgrading hardware. Now I rely on a Space Engineers save and my modded Fallout 4. This particular scene stalls with stutter and pauses when using the 8086k and 9900k without a 5GHz overclock. With overclocking it works but isn't smooth. It runs perfectly with a stock 5800x/5900x, offering a clear view of the CPU gap. It's somewhat surprising how visually appealing these settings appear on OLED screens even at 1080p. At 4K it just looks like a slightly outdated game. I managed around 207 frames per second at 4K, but it wasn't completely GPU-limited. I suspect the issue lies with the L3 cache since single-core games still feel smoother on 8-core processors. Whether this is realistic or not, the numbers come from screens like yours, adjusted to settings I actually use. The 5900x at stock performance is roughly 120 frames down, which is impressive. I plan to build with an i9 12900k next year unless AMD alters my strategy.
I saw Linus' video about RAM and gaming. The outcome wasn't impressive, so I didn't pay much attention. You wonder if quicker RAM would make a difference—I'll check it out later.
Low CPU testing for comparison works well, particularly when aiming for future upgrades. It helps estimate performance with newer GPUs as you mentioned. Still, I think it’s mostly theoretical and doesn’t justify spending too much time on it. Those planning for future GPUs strong enough to count usually won’t be overly concerned, and they’re more likely to upgrade a CPU regardless.