Since introducing its Ryzen series of desktop processors into the PC market, AMD has consistently increased core counts and core clocks, shrunk its transistor size to 7nm, and modified its architecture to be more efficient all on the same motherboard socket. The scaling AMD achieved between its Zen and Zen 2 architecture generation was impressive enough, and AMD has done it again with its new Zen 3 architecture and latest series of Ryzen processors. It’s done it again so hard it’s surpassed Intel in single core performance.
Yes, that’s right. AMD’s Ryzen 9 5950X, it’s current top-of-the-line processor of the series, beats Intel in single core performance. It’s so good, I can’t even... The clock speeds, the gaming frame rates, the thermals, the power consumption are all great. The only thing that will stop some people from buying this processor is the price or if they literally built a new PC yesterday. If you are part of the latter, oof, why? Didn’t you know these CPUs were coming?! This is the best Ryzen processor AMD has made to-date, and this generation of Ryzen processors should be everyone’s first pick the next time they build a PC or buy a pre-built desktop. Other than the price, I cannot find anything worth complaining about.
At $800, the Ryzen 9 5950X is a decent chunk of change more than Intel’s Core i9-10900K, which was supposed to cost around $488 at launch. But global supply chain disruptions due to the covid-19 pandemic caused all sorts of production delays across many industries, and unfortunately that meant Intel was releasing its 10th-generation of desktop processors in the middle of all it. The i9-10900K became hard to find not too long after it hit the shelves. I remember going into Micro Center and seeing a healthy supply of 10th-gen Core i5s and some i7s, but the i9s were missing.
All that is to say the Core i9-10900K goes for an average of $580-$700 these days, sometimes higher. So when you look at what you gain with the Ryzen 9 5950X in comparison, spending $100-$220 might be worth it if you have the money to burn and you’re upgrading from a Ryzen 2000-series or Intel 7th or 8th-gen yet. Oh hell, even if you have a Ryzen 3000-series or Intel 9th-gen, I’d consider it.
At $550, the Ryzen 9 5900X (12-cores/24-threads, 3.7 GHz base clock/4.8 GHz boost) might even be the best option for most people working within a more modest budget. We haven’t had the opportunity to check that one out, but given all the benchmark results for the Ryzen 9 5950X below, there’s enough of a performance gap between it and the Core i9-10900K to say that the Ryzen 9 5900X could be on par, maybe slightly better than Intel in terms of single core performance and gaming performance. Just every bit of performance, really.
AMD’s Ryzen 9 5950X is a 16-core/32-thread monster with a boost clock up to 4.9 GHz, although I consistently saw a few cores hit up to 5.0 GHz. Intel’s i9-10900K 10-core/20-thread processor in comparison can get up to 5.3 GHz, but in experience I’ve only seen the processor reach 5.1 GHz. So depending on whatever core or cores need a little frequency boost in the middle of whatever your doing both the Ryzen 5950X and Core i9-10900K have the same effective boost clock speeds—and yet AMD still outperforms Intel by a lot thanks to the architectural changes Zen 3 has ushered in.
One of those big changes was how the cores accessed the cache on the chip. With Zen 2, Ryzen chips had two separate 16MB 4-core cache complexes (aka the part of the processor that helps reduce the time it takes to access data on the SSD. The cores assigned to their respective caches could only communicate within that complex. Zen 3 has a single 32MB 8-core cache complex now, so all the cores can talk directly with the cache. AMD mentioned during its stream that this architectural change reduces latency and boosts processing speed.
So all those benchmark numbers AMD threw around last month? AMD was not exaggerating. For comparison, I retested the Intel Core i9-10900K and AMD’s Ryzen 9 3900XT, which has 12-cores/24-threads, a 3.8 GHz base/4.7 GHz boost clock, and built on Zen 2. My test system included an: RTX 3080 GPU, G.Skill Trident Z Royal 16GB (2 x 8GB) DDR4-3600 DRAM, Samsung 970 Evo NVMe M.2 500GB SSD, Seasonic Focus GX-1000 PSU, and a Corsair H150i Pro RGB 360mm AIO for cooling.
Both AMD systems used an Asus ROG Crosshair VIII Hero (wifi) X570, and the Intel system used an Asus ROG Maximus XII Extreme Z490. For all tasks, I enabled XMP/DOCP on all three systems, which overclocked the DRAM to 3600 MHz instead of the base 2133 MHz it normally sits at.
I should also note that I ran these tests on a fresh Windows 10 install with no programs running in the background—ideal conditions—which is not the case if you’re gaming and want to use a program to take screenshots, capture video, or stream live. That will always eat up some performance. I imagine most people, myself included, usually do not game under ideal conditions. I have Google’s Back Up and Sync, Adobe’s Creative Cloud Desktop, Slack, and one or two other programs that are usually open whenever I use my PC.
Anyway, here’s some numbers!
We don’t normally use Cinebench R20 in our testing, but since AMD claimed a 87 point lead over the i9-10900K in its event, I needed to see it happen on a screen in front of me. The Ryzen 9 5950X does indeed decimate Intel’s Core i9-10900K in single core performance, 636 to 546, which is a full 90 points higher. The gap is much bigger in multi-core/multi-threaded performance, too, although that’s not surprising considering how many cores the Ryzen 9 5950X has in comparison. I’m just including those scores so you can see how fast the performance is. And some people like numbers.
The Ryzen 9 5950X faired extremely well in our usual Geekbench 4 test, too. With the DRAM overclocked, it was 649 points higher than the i9-10900K—a larger gap than between the i9-10900K and the Ryzen 9 3900XT. Even with the DRAM at a normal speed the Ryzen 5 5950X has a 601 point lead over Intel. Rendering a 3D image in Blender and transcoding a 4K video to 1080p at 30 fps in Handbrake is super fast, too.
This is what makes me believe the Ryzen 9 5900X might have the same or better single core performance than the i9-10900K, even with a boost clock of 4.8 GHz. It’s all in the architecture.
AMD’s promise of increased frame rates from its previous generation also holds true. For the gaming benchmarks, I left the DRAM overclock enabled, and that’s represented in the benchmark scores below. On average, the Ryzen 9 5950X gains up to 30 fps to 40 fps over the Ryzen 9 3900XT at 1080p. At 1440p, those gains are around 10 fps to 25 fps. Metro Exodus is the outlier, with a smaller fps increase between processor generations. 4K performance across the board is even.
Also, you’ll see in some games the Ryzen 9 5950X either gets near the same frame rate as the i9-10900K, or it’s a decent amount ahead. Faster raw single core performance and level with Intel in gaming? I’d say that’s pretty good. Pretty damn good.
The 5950X does suffer from some of the same bottlenecking as the i9-109ooK when paired with the RTX 3080, though. When comparing frame rates with the DRAM overclocked and not overclocked, I got between a 3 fps to 10 fps boost at 1080p, so the Ryzen 9 5950X is not quite powerful to tame the RTX 3080, but it’s really the most minor of complaints.
It’s exciting to see AMD make all these generation-on-generation improvements, and it’s really stepped up its game to compete with Intel on all levels as far as gaming desktop processors are concerned. It will definitely be interesting to see how Intel’s next generation of Rocket Lake (11th) desktop processors compare to AMD’s Ryzen 5000-series, especially with all the advancements its made with its 11th-gen Tiger Lake mobile processors and it (finally) moving from 14nm to 10nm. We won’t see those hit shelves until sometime in the first quarter of 2021, though
Intel could pull ahead again in single core speed with Rocket Lake, and we could have a game of silicon tug-o-war with every processor generation that gets released after that. Intel has had issues moving off its 14nm node though, and there’s a lot of questions about how good its 10nm node will be, or if Rocket Lake will even be on the 10nm node at all. Recent rumors point to Rocket Lake having boost clocks of up to 5.6 GHz on a single core, which I’m highly-highly skeptical about.
In a few short years AMD has gone from trailing behind Intel and relying on budget processors to increase adoption, to surpassing Intel on nearly every front (and keeping its processors competitively priced). If you’ve never given AMD any thought in your life, now is the time to pay attention.
README
- A nearly perfect CPU.
- Equal or better single core performance than Intel, depending on the task.
- Huge architectural improvements from the previous generation.
- Competitive performance per dollar.
- But still wish it wasn’t $800!