We recently took an interesting look at the progress of Intel CPUs over the past six years, testing 5th Gen Core against the performance of 10th Gen processors. We covered two angles: first, we separated the benefits of increasing the pace, and second, what we achieved by adding more cores. This gives us a clear picture of the architectural improvements made by Intel in the 14nm process.
Over the years, of course, they have acquired much more operating frequencies and increased the number of cores. In this case, we also examined how increasing 6, 8, and even 10 cores improves game performance, although we ignored frequency frequencies because they are well studied at this point and the frequency scale was more predictable. Finally, we found that Intel made limited 5th to 10th generation optimizations in the 14nm process, which is a 10% increase in gaming performance compared to core number and performance-frequency CPUs. It includes the transition from DDR3 to DDR4.
so far, the biggest increase in Intel's performance by increasing the capacity of the L3 cache, increasing the performance iteration near the maximum, and of course, adding more cores in principle, it can have Intel 10 Generations 2020 demonstrated their existence five years ago with Skylake, but due to the lack of competitive pressure, they didn't need it, and so they made up for the smaller 122mm square blocks.
Speaking of competition, today's review is about AMD's progress using the same testing method. We compare the index components of each generation with only 4 active cores at a fixed frequency and some components of the current generation where all cores are active.
Even without looking at any new numbers, we acknowledge that this could show AMD in a more positive light because we have to see the huge accomplishments the company has made. Highlighting the days of FX and the latest generation gained by Ryzen. However, the reality is that the company's CPUs are miles behind Intel because AMD was largely irrelevant during the FX era.
Let's talk about exactly what we're looking for here...
We're going back to 2012 when AMD released the FX-8350 based on the Piledriver kernel. Using a 32nm problem process from GlobalFoundries.” Technically, the first FX-series processors came on the market a year ago and were named Bulldozer, but we're skipping the FX-8150 and the entire series. We really only need one FX processor, maybe the best.” “And I love using that term in discussing forex processors. There was also the FX-8370, which came two years later and had the same CPU, as well as the FX-9370 and 9590, which were basically the same parts again but with a 220W TDP increase, or about 80 higher over the 8350(!). These were just overclocked models, and in the case of the FX-9370 and FX-9590, the base and subwoofer frequencies were not improved by 20%.
Since we all test components at 4.2 GHz, we chose the FX-8350 and installed it at that frequency. We're testing with only 4 cores enabled, but since the FX series only hit 4 cores with 8 threads, we don't have to make any changes here. Some might still argue that the FX-8350 is an eight-core CPU, but that's not the case and certainly doesn't comply with California consumer law. In 2015, AMD successfully claimed that it misrepresented its dozer chips, and in August 2019, AMD agreed to pay $12.1 million.
The latest motherboard BIOS still claims it's an octa-core FX-8350 processor, while Windows says it's a quad-core processor with 8 logical processors, so it's basically a quad-core with SMT support. For the FX-8350 test, we are using the Asus M5A99FX Pro R2.0 with DDR3-2400 CL11-13-13-31 memory.
Then for comparison we have the Ryzen 7 1800X, 2700X, 3800X and 5800X except for the 1800X, we limit all CPUs to 4 cores at 4.2 GHz. The 1800X was active with only 4 active cores, but its frequency was 4.1GHz, because it was the highest fixed frequency I could achieve with this part.
We've also included the Ryzen 5 5600X and Ryzen 9.5950X for 6-16 cores and are directly comparable to the FX-8350, although it's weaker in terms of clock speed since it's weak at 4.2GHz.
< Ryzen 7-Series tested from 2700X to 5000 on the Gigabyte X570S Aorus Master and 1800X on the MSI B450 Tomahawk Max.
All Ryzen processors with dual-channel DDR4-3200 CL14 memory are manually configured using all primary, secondary, and tertiary scheduling operations. Then all CPU configurations were finally tested with a Radeon RX 6900 XT. Let's take a look at the results...Criteria
Starting with Rainbow Six Siege, we see that the FX-8350 was good at an average of 211 fps, which obviously has a lot of performance, which is Vulkan not Really, it requires a lot of CPU performance. However, it was still significantly slower than the first generation Ryzen, the 1800X delivered about 40% more performance, and remember that only half of the Ryzen CPU is active, despite the significantly overclocked.
The 2700X only offers a small upgrade from the 1800X, and here we're looking at a 9% increase in performance. Then as we know Zen 2 delivered significant performance from Zen+ and in this example we're looking at a 23% increase in frame rate and we've upped the Radeon 6900 XT to 390 fps, which is a whopping 84% increase over the FX. 8350.
Then with the introduction of Zen 3 we see the biggest step forward, here the 5800X with 135 active cores is 135% faster than the FX-8350, and that shows just how bad the FX series was, limiting the performance of the 6900 XT by more than half. For those surprised, the 5800X tested in a 4+0 configuration because AMD switched from Zen 3 to 8-core CCX, where Zen 2, Zen+ and Zen per CCX only have 4 cores. Finally, it is interesting to note that when trying out Rainbow Six Siege, you will only get a 9% performance increase when you increase the number of Zen 3 cores from 4 to 6 and then a 10% increase when you increase from 4 to 16 cores.
This title obviously doesn't use a lot of CPU, which is why the FX-8350 is capable of delivering highly capable gaming performance.
The Assassin's Creed Valhalla experience is very different, at least from that of FX. 8350. While everything was very smooth in Rainbow Six Siege, the constant stuttering when playing with Valhalla with a low score of just 1% 39 fps was very noticeable. The 1800X increased the low 1% performance by about 60%, giving it a much better gaming experience. The 2700X was a lot faster again, boosting low performance by 21% and then very close to the 6900 XT's max with the 3800X. In fact, the 5800X 4-core wasn't even faster, while the full 5950X only improved the frame rate by up to 7%.
< Battlefield V is a great example of how bad the AMD FX series can be, and of course it still seems like the more FX processors you use, the more it disintegrates. Low-level APIs like Vulkan, which help bypass the CPU, and thus significantly reduce CPU overhead, often play well with very low-power CPUs, like the FX-8350, and we did just that a moment ago. We saw it in Rainbow Six Siege.
But Battlefield V crashes using the DX11 API dedicated to this title, FX-8350. Averaged 47 fps and 29 fps 1 1 the game was completely unplayable by my standards and was completely useless even for semi-competitive play, remember we got a Radeon RX using a 6900 XT for this test.
On the other hand, the Ryzen 7 1800X was 160 faster, a 160 percent increase in performance from generation to generation for AMD, a failed generation to a somewhat successful one. AMD was able to improve its performance by up to 11% with the Zen+ refresh and then by another 16% with the Zen 2, where the 3800X allows for 158 frames per second.
Then we see more with Zen 3. 21% performance increase at 191 fps, 306% increase in FX-8350.
With a quad-core configuration using the 5800X, the performance at this address was increased by 6 cores and 10%. Then 27% with 16 cores. That means the Ryzen 9 5950X is 415% more efficient than the FX-8350, great.
F1 2020 is not the most popular title and also supports DirectX 12. So here FX-8350 works logically and definitely provides a very interesting and playable gaming experience. However, we're still looking for a 46% increase in performance with the 1800X, where only half of the cores are enabled to compare apples and more apple bodies.
Once again, AMD is a small step forward compared to the second generation Ryzen, here the 2700X offers a 9% performance increase, then another 15% increase with the 3rd gen. When we see a noticeable 24% increase from 3800X to 5800X and up to 266 fps.
Then when activating more cores for the Zen 3 architecture, we see a 14-15% increase from 4. Cores up to 16 cores In this example, the 5950X was 162% faster than the FX-8350, and we're looking forward to Same margin for the hexa-core version.
< Hitman 2 CPU is much faster than F1 2020 and this is causing problems for FX-8350. The game was more playable than Battlefield, but the 6900 XT experience was by no means perfect. The 1800X provided a 67% increase in average frame rate and a 1% improvement in low performance by 82%, so the first-generation Ryzen processor was miles faster.
We see similar performance improvements in the Ryzen generations. And when we got to Zen 3, AMD improved the performance of the FX-8350 by 147% when comparing base and normal frequency. That time was 10% faster than the quad-core 5800X, 10% faster than the hexa-core 5600X, and 21% faster than the 1650 5950X. We increased the performance of the FX-8350 by almost 200% to the 5950X, and both CPUs are running at 4.2GHz.
Horizon Zero Dawn plays well on the FX-8350, and what we have here is a game with a low-level API that isn't particularly CPU-required again. However, the 1800X offers 35% more performance with half of its cores, and when we get to the 5800X, where half the cores are still active, we're looking at a 105% performance improvement in the FX processor.
Interestingly, adding more cores doesn't do the game much good, so if there was a quad-core Zen 3 processor, it might have made the most of the 6900 XT for the game. .
Cyberpunk 2077 is another game that uses a low-level API, but these games have a lot of CPU and as a result, the FX-8350 tanks, a 1% low score at only 32 fps, let May I tell you, compared to other processors tested, the FX processor experience was horrible. Even 1800X with half its cores was able to improve low performance by 1% to 94%, which is incredible on every level. Upgrading the number from 1800X to 5800X AMD was able to improve its base performance by up to 53%, when we look at the average frame rate. Then by activating the other two cores, a 1% lower performance increases the improvement from 86fps to 101fps, so a 17% performance increase is good.
The 5600X and 5950X offer roughly the same performance. At this title that means the 5600X was 216% faster than the FX-8350 compared to a 1% lower performance.
Shadow of the Tomb Raider is another DX12 title that is highly demanded by the CPU. For testing we don't use the internal benchmark because it's more of a GPU test, but in a big village it increases the CPU load significantly.
Here the FX-8350 offers playable performance. Depending on your benchmark for a single player game, it can be said that it is playable but not desirable because we are using a very fast GPU like the Radeon RX 6900 XT. This means that the Ryzen 7 1800X is able to improve a low 1% performance by up to 51%, which is a huge improvement over the generation, because the FX series was such an unwanted fire that AMD took 6 years to shut down, and even then the stench became possible as well. With the advent of AMD's Zen 2 it's already gone, resulting in a roughly 30% performance increase by about 1800X over the same core and clock frequency, making TSMC's 7nm process much better. At the time of Zen 3's debut, AMD outperformed the FX-8350 when matching the number of cores and clock frequency 126.
But with 6 cores activated, we're looking at another 30% increase and then an increase of 38 by 5950x. The 5950X is therefore 213% faster than the FX-8350 than the average frame rate and 250% faster than the 1% lower result. So yeah, it's a little faster.
< Finally, we have Watch Dogs Legion, which also requires a CPU, so it rips the FX-8350 to the point where I can tell it can't run at full 1 ratio, low ratio, only 28 fps, 1800X at half fps, resulting in an 81% performance increase.
Then we see a 126% increase in performance with the 3800X and a 152% increase with the 5800X, which is active again using only 4 cores. Hopping from four Zen cores to 3 to 6 improves the average frame rate by 18% or decreases by 29% versus 1%. Jumping from 6 to 16 centers in this title doesn't work much and we're far from the processing power required for the game.
As expected from AMD's dark days in the FX series, the company made major improvements with Ryzen and eventually overtook Intel with Zen 3, although Intel is expected to come soon to Alder Lake (the Core series of the twelfth generation).
Before doing anything in the rest of this discussion, let's take a look at the average performance seen in the 9 games we tested.
< Running on average, the FX-8350 did well at 70 fps, down 47% at 47 fps. Compared to Zen architecture with 1800X with half of the non-active cores, but with a noticeable increase in frequency, the clock improves clock performance by a whopping 60%, or if we look at results by 1% less than 70%. This is a huge achievement for a generation, and while the 1800X didn't run at 4.1GHz and only the best silicon reached that frequency, it's still a huge architectural feat if adapted to the clock speed.
Even if we want to increase 1800X to 3.6GHz, which is a 12% decrease in frequency, you can only expect a 10% decrease in performance. However, the first generation Ryzen was a huge upgrade for AMD.
Then from Zen to Zen + we see another 10% increase. This time, the 4.2GHz frequency was more realistic from the point of view of the off-the-shelf product. AMD increased 15% with Zen 2 and then 17% with Zen 3. These clocks are good development clocks for the Ryzen generations.Incredible Performance
Seeing AMD's lead in the past is insane in the past decade and how long it took to finally get to Intel's door with 3rd generation Ryzen. At the time Intel was ditching Coffee Lake, which was architecturally just Skylake a couple of years ago.
In 2016, a year before the first Ryzen processors were released, AMD relied on components like the FX-8350 to battle Skylake; Processors like the 4C/4T 6600K and 4C/8T 6700K, AMD tried to keep up with even Core i3 cores at the time because most games only needed an Intel 2C/4T processor.
In short, the i7-6700K is equivalent to a 3800X, locked at the same frequency and count. This means that there was a time when Intel processors delivered 100% more performance if you removed the GPU limit. That is why the FX-8350, which was already the flagship of the series, was cheaper than the fastest Core i5 from Intel, and at that price it was still a terrible buy. Since then, AMD has made serious headway with Ryzen, increasing performance by 50% from the 1000 series to the 5000 series when matched at the same frequency. Given the higher Zen 3 clock and the 12 and 16 base models, the real performance boost is much bigger.
If you ask yourself 'where are the power consumption figures', we'd prefer not to put them in the 'for science' type criteria because they are terribly imprecise and no real conclusions can be drawn from them. This is because components like the 1800X are significantly overclocked and thus provide a large amount of voltage to achieve power stability, while the 5800X is fundamentally weak. The power meter is the best in stock, it allows us to perform in watts comparison, and you can find this information on the first day of checking these components.
You certainly can't compare the 10 Intels architecture achievements in the past six years to a 50% increase for AMD in the past four years, because they were different from my two starting points and it took a long time to be able to reach their peak. AMD Ryzen 7 5800X on Amazon AMD Ryzen 5 5600X on Amazon AMD Ryzen 9 5900X on Amazon AMD Radeon RX 6900 XT on Amazon AMD Radeon RX 6800 XT on Amazon AMD Radeon RX 6600 XT on Amazon
9 Years of AMD Processors: From AMD FX to Ryzen 5000 Series, Tested