The Ryzen 9 5900X’s performance and AMD’s latest Zen 3 architecture are truly impressive. They have successfully met the expectations set during the chip reveal on October 8, which could be concerning for Intel. This is great news for gamers as it provides a real option for the best CPU for gaming in their upcoming builds.
Intel has maintained its competitiveness against AMD’s continuous performance enhancements with Zen, largely due to its strong position in gaming. Despite AMD’s advancements in other areas of computing, Intel still had a stronghold in gaming. However, with the arrival of Zen 3, Intel no longer has this advantage. While AMD may not have completely surpassed Intel in gaming as initially anticipated, the difference between the two is now minimal. The fact that this chip can handle various processor-intensive tasks and perform just as well in gaming as Intel’s top processors makes Intel a less appealing choice for those seeking high performance.
There are, however, some considerations to keep in mind. It is a reality that high performance often comes with a high price tag. AMD has transitioned from being the underdog, and this is evident. The Ryzen 9 5900X is priced at, which is comparable to Intel’s leading chip, the Core i9 10900K (originally launched at, but has since increased in price). While it is true that this AMD chip offers more cores than Intel’s 10-core processor and excels in demanding tasks, the fact remains that it is a costly investment.
However, this is not the complete narrative, and there is much more to AMD’s most recent microarchitecture than what has been mentioned so far. Therefore, without any more delay, let us explore AMD’s most impressive architecture to date.
AMD emphasizes that Zen 3 architecture is not just a simple evolution of Zen 2 on paper. Instead, it is a significant redesign from the ground up. The performance improvements it brings further support this claim. AMD boasts a 19 percent IPC improvement over Zen 2, a notable achievement that is not easily attained, as evidenced by Intel’s more modest improvements in recent generations.
The enhancements made to the Zen microarchitecture in the AMD Ryzen 5000 processors do not stem from a new process node. In fact, these chips utilize the same production process as the XT chips released earlier this year, such as the 3900XT, 3800XT, and 3600XT. TSMC’s 7nm node continues to be the technology used for these new processors. While there have been optimizations implemented by both AMD and TSMC to maximize the efficiency of the process, it is important to note that Zen 3 does not incorporate TSMC’s upgraded 7nm+ production process.
Zen 3 brings about a significant change in the configuration of cores compared to Zen 2. In Zen 2, each core complex (CCX) can have up to four cores, allowing each four-core cluster to access 16MB of L3 cache. However, if a core from one cluster needs to access the L3 cache from another cluster, it must communicate with that cluster through the I/O die using the Infinity Fabric, resulting in much slower access to the local L3 cache.
In Zen 3, AMD has transitioned to an eight-core design. All eight cores now have direct access to 32MB of L3 cache, eliminating the need to pass through the I/O die. This enhancement results in faster core-to-core access, leading to potential performance boosts. While dual chiplet CPUs may still require one cluster to access the L3 cache or communicate with cores from another cluster through the Infinity Fabric, such scenarios should occur less frequently during regular usage.
This information holds significance for us as games frequently depend greatly on fast cache and quick memory access. Therefore, having more cores with access to a larger L3 cache can lead to a noticeable increase in performance. Although transitioning from two four-core CCXs with 16MB of L3 cache each to a single eight-core CCX with 32MB of L3 cache may not appear to be a significant enhancement at first glance, it is actually crucial for gaming purposes.
This data is crucial for us because games heavily rely on fast cache and quick memory access, so having more cores with access to a larger L3 cache can result in improved performance. While the shift from two four-core CCXs with 16MB of L3 cache each to a single eight-core CCX with 32MB of L3 cache may not seem like a major upgrade initially, it is actually highly important for gaming.
This adjustment enables AMD to manufacture CPUs with up to eight cores using just a single CCX and the corresponding I/O die. The Ryzen 7 5800X is a prime example of this configuration. On the other hand, the Ryzen 9 5950X, positioned at the top of the lineup, features two eight-core chiplets in addition to the same I/O die. Meanwhile, the Ryzen 9 5900X is equipped with a pair of chiplets, each containing six active cores. This flexible design allows AMD to maximize the efficiency of its operational chiplets. As demonstrated with Zen 2, this strategy enables AMD to achieve significant cost savings per processor.
The 19 percent IPC improvement claim is not solely attributed to the reworking of core layout. The upgraded performance of Zen 3 also stems from enhancements in the processor front-end, including faster fetching due to a doubled L1 buffer size, increased bandwidth for handling branch predictions, and quicker recovery for mispredictions.
The execution engines have undergone a significant upgrade, resulting in decreased latency and increased silicon capacity for improved performance. Additionally, there has been an enhancement in load/store bandwidth to accommodate larger structures, leading to more efficient code handling and faster completion.
The significance of these IPC enhancements cannot be understated, as many PC games continue to favor CPUs that excel in processing numerous instructions per clock cycle rather than simply having a high core count. Intel’s historical advantage in gaming has been attributed to its superior IPC and clock speeds, but this may no longer be the case. AMD has made significant strides in improving its IPC to the point where it can now deliver comparable performance to Intel, even at lower clock speeds.
Intel is capable of achieving higher clock speeds compared to AMD, however, AMD currently possesses a more efficient architecture. One contributing factor to this is AMD’s decision to adhere to the 142W power limit set by the AM4 platform. With the improvement in IPC, AMD has managed to lower the base clock by 100MHz across its Ryzen 5000 CPUs in comparison to the previous generation. Nevertheless, this adjustment has led to an increase in the boost clock, enabling the core, or cores, requiring higher frequencies to operate accordingly.
It is important to highlight the origin of AMD’s 19 percent IPC improvement statistic. This figure is derived from the average performance gains observed across various benchmarks when comparing 8-core chips running at the identical 4GHz frequency. However, when specifically considering gaming performance, AMD claims a 26 percent improvement when transitioning from a Ryzen 9 3900X to the latest Ryzen 9 5900X. Games such as PlayerUnknown’s Battlegrounds may even experience up to a 50% boost in performance.
AMD has opted to utilize the identical 12nm I/O die manufactured by GlobalFoundries for these latest processors. This decision is logical given the minimal impact on system connectivity. Moreover, the Infinity Fabric now has the capability to support clock speeds of up to 2,000MHz, when paired with DDR4-4000, this combination signifies the new optimal level for memory performance.
The approved standard for memory support remains DDR4-3200, despite successfully testing the new chips with DDR4-3600 and DDR4-4000 RAM without any issues. It is important to highlight that AMD is currently in the process of optimizing the Infinity Fabric to operate smoothly at 1,900MHz and 2,000MHz on all platforms, although this feature will not be accessible upon release.
It is important to emphasize that Ryzen 5000 CPUs are compatible with 400- and 500-series motherboards. However, older B450s and X470s will not receive the required BIOS updates until January at the earliest, and even then, it will be for the first beta BIOSes. Those with older motherboards will likely have to wait until February or March for full Ryzen 5000 compatibility. If you are planning to build a new system with these new chips, it is recommended to choose a motherboard with A520, B550, or X570 chipsets.
The Ryzen 9 5900X is AMD’s high-end mainstream chip, equipped with 12 cores and 24 threads, a base clock of 3.7GHz, and a potential boost clock of 4.8GHz. It features 64MB of L3 cache, support for PCIe 4.0, and maintains a 105W TDP like its predecessor. Unlike the 3900X, it does not include a cooler.
The core and thread count of this chip is identical to that of the 3900X, which was launched slightly over a year ago. At that time, it was reasonable to conclude that while this setup was excellent for demanding tasks, games were not yet optimized to fully utilize the abundance of threads. This remains true today, and anything beyond an 8-core, 16-thread setup is still excessive for current gaming needs. This chip is tailored for professional work, but it is also capable of meeting your gaming requirements, but we will discuss that in more detail later.
The frequencies observed in operation are heavily influenced by the specific application being utilized and the areas where power demand is highest. For instance, in a thread-intensive application such as 3D rendering, all cores may be utilized equally, resulting in a stable frequency of 4.2GHz. On the other hand, single-threaded applications are expected to reach a maximum frequency of 4.8GHz on one core, but in practice, some cores have been observed to exceed this limit, reaching speeds as high as 4.95GHz during testing.
To comprehend the current situation, it is important to briefly review Precision Boost 2, the technology utilized by AMD since the release of its Ryzen 2000 processors to enhance frequencies in real-time. This feature is not related to overclocking and cannot be disabled; rather, it is a mechanism designed to increase clock speed when there is available capacity to do so, ensuring optimal performance of the chips. It is not a fixed setting, but rather enables the chips to adjust dynamically to different workloads. The chip monitors various activity levels to determine if boosting is possible, and it can make rapid adjustments in milliseconds.
In order to grasp the ongoing situation, it is essential to briefly examine Precision Boost 2, the technology employed by AMD since the debut of its Ryzen 2000 processors to elevate frequencies on-the-fly. This is distinct from overclocking and cannot be deactivated; instead, it is a system that boosts clock speed when there is room for improvement, guaranteeing peak performance of the chips. It is not a static configuration, but rather enables the chips to adapt dynamically to varying workloads. The chip monitors multiple activity indicators to ascertain the feasibility of boosting, and it can swiftly make adjustments in milliseconds.
One crucial aspect to consider is that Ryzen 5000 chips exhibit an assertive approach towards utilizing sleep states to minimize power consumption. Inactive cores are not downclocked or undervolted; rather, they are effectively disabled, drawing no power and generating no heat. Consequently, the active cores have additional power and thermal capacity to operate at higher clock speeds in critical tasks.
It is encouraging to note that the 5900X has exceeded the maximum boost clock on multiple occasions, indicating that AMD is not exaggerating the capabilities of these new chips. This is a significant improvement from Zen 2, where it was difficult to observe the top boost clock being reached. As a result, users can expect to see the boost clocks being consistently achieved.
The crucial aspect is performance. A processor may possess the most intelligent architecture, but if it doesn’t perform, what’s the use? Fortunately for AMD and potential buyers of the Ryzen 9 5900X, it performs exceptionally well. Whether it’s handling intensive tasks like video encoding and 3D rendering or gaming, the Ryzen 9 5900X truly delivers. It’s a powerhouse.
Cinebench R20 serves as an excellent tool for evaluating the sheer processing capability of a chip, and it thrives on utilizing numerous high-performance threads. In comparison to the previous generation 3900XT, the 5900X demonstrates a substantial 16 percent improvement in multithreaded performance, while also showcasing a noteworthy 20 percent increase in single-core performance.
I have added the 16-core, 32-thread 3950X to provide a point of reference, and the recent processor performs nearly as well. In contrast, Intel’s 10-core, 24-thread 10900K falls short, as AMD’s newest chip boasts a significant 30 percent increase in speed.
The 16-core, 32-thread 3950X has been included for comparison purposes, and the latest chip from AMD is almost as speedy. In contrast, Intel’s 10-core, 24-thread 10900K lags behind significantly, with AMD’s most recent chip outperforming it by a solid 30 percent.
When examining video encoding, the X264 v5.0 benchmark shows a significant difference between AMD’s latest chip and the 10900K. The AMD chip achieves an impressive 75 frames per second compared to the 52 frames per second of the 10900K, giving AMD a 45 percent lead. Additionally, the 5900X slightly outperforms the 3950X, highlighting the impressive capabilities of this new chip.
In terms of temperatures and power consumption, there are no concerning issues to mention in this case. AMD has maintained the 142W socket power, as previously stated, and this was evident during intense usage. During idle periods, when cores are deactivated and activity is minimal, the power draw decreases to 25W, showcasing impressive efficiency. The highest temperature recorded was 81°C, a figure similar to that of the 3900XT.
Regarding the games, there are several key points to consider. One notable aspect is the significant improvement in overall performance in certain titles, as evidenced by the score of F1 2019. When compared to the 3900X, there is a 20 percent boost in performance. However, what is even more significant than the advancements from the previous generation is the fact that there is now virtually no difference between the 5900X and Intel’s top Core i9 10900K.
In some games, there is not an equal performance between Intel and AMD processors. For example, in Assassin’s Creed Odyssey at 1440p, Intel still maintains a 7fps advantage. However, in Total War: Three Kingdoms at 1080p, the 5900X surpasses Intel’s top processor. When gaming at 4K, the performance gap between Intel and AMD is negligible, with frame rates falling within the margin of error.
If you are looking to purchase a processor and are mainly a gamer, there is virtually no difference between the two options. While AMD may not have achieved the level of dominance it initially promised at launch, our testing shows that the performance gap between these chips is now minimal. While there may be a few games that run slightly better on one platform compared to the other, the overall gaming performance is essentially identical.
Overclocking AMD processors has seemed rather futile for several generations, and Zen 3 doesn’t alter that perception significantly. While it is possible to overclock and witness an increase in multi-core performance, this usually comes at the expense of single-core performance. Consequently, the overall benefit is often not substantial, especially considering that certain games may actually run worse when overclocked, despite a slight improvement in a few cases. In essence, the key point to grasp is that AMD excels at maximizing performance potential more effectively than individual users, and recognizing this sooner rather than later is advisable.
We are compelled to pursue better performance, so we experimented with BIOS and Ryzen Master overclocking to explore the potential. The simplest method, Precision Boost Overdrive, resulted in a slight increase in Cinebench R20, reaching 4,374MHz and achieving an overall score of 8,553 (compared to the stock score of 8,348). However, auto overclocking did not significantly improve upon this, and we do not see much advantage in using it.
The Cinebench R20 score of 9,062 was achieved by manually overclocking to 4.6GHz at 1.3V, resulting in the biggest performance boost. This surpasses the speed of the 16-core, 32-thread 3950X. The enhancements made by AMD to this chip compensate for the reduction of four cores compared to the previous generation, which is quite remarkable. However, the single core performance is still affected by this overclocking method, resulting in a seven percent decrease in performance to 594.
Maybe overclocking on a per CCX or per core basis will make a difference in the future. However, at the moment, it is not beneficial to overclock these processors.
One last question remains: Is it advisable to purchase the Ryzen 9 5900X? The response, as frequently happens, is: it varies. If AMD had introduced this processor at the identical price as the 3900X’s, it would have been a straightforward decision. However, the additional does make one stop and think, particularly when considering the necessity of acquiring a good aftermarket cooler. While it is a superb processor, it comes at a high cost.
It seems evident that this chip was designed with longevity in mind. The inclusion of PCIe 4.0 support is a significant advantage for AMD’s processors. Despite the fact that support for this technology has been around since Zen 2’s launch, its importance remains high. With the increasing demand for fast SSDs, combining this chip with high-capacity storage will ensure optimal performance as we enter an era that fully utilizes storage technology. This alone gives AMD’s processors an edge over Intel’s PCIe 3.0 supporting counterparts.
If I were to construct a high-end PC at this moment, would this CPU be the ideal choice? Indeed, it most likely would. Coupled with an Nvidia GeForce RTX 3080, it forms a powerful core capable of effortlessly handling 4K tasks and excelling in productivity. Additionally, it is expected to work seamlessly with the upcoming RX 6000-series GPUs. By adding a top-notch next-gen SSD, a reliable motherboard, and high-quality RAM, you will have a setup that surpasses the longevity of both the PS5 and the Xbox Series X. While it may come at a higher cost, the capabilities it offers far outweigh the price.
AMD has successfully claimed the top spot in gaming performance from Intel. While it may not be a complete victory, more of a narrow win, the disappearance of diehard Intel supporters speaks volumes. Whether you prefer Intel or AMD for gaming, both are equally competent. With AMD’s comprehensive package, it’s clear that this chip is the superior choice.
No account yet?
Create an Account
