After a long period of anticipation, on July 15, 2024, Mr. David McAfee, Vice President and General Manager of Client Channel Business at AMD, officially unveiled the mystery of the Ryzen 9000 series desktop processors. The new processors bring a brand-new Zen 5 architecture, the latest process technology, and at least a 16% IPC improvement compared to the previous generation. It is no exaggeration to say that the most anticipated and exciting product of this summer is the AMD Ryzen 9000 series. Compared to its competitors, it has achieved transcendence in process technology, technology, and performance, and has excellent stability and reliability, especially in the top-tier models. Without further ado, let's take a quick look at the specific content of this disclosure.

Mr. David McAfee, Vice President and General Manager of Client Channel Business at AMD, unveiled the mystery of the Ryzen 9000 series desktop processors.

The union of top-tier performance and extremely high energy efficiency ratio: Ryzen 9000 series desktop processors

AMD's Zen series microarchitecture and Ryzen series processors are currently the most important and mainstream desktop processors on the market. Starting with the first-generation Zen architecture's IPC improvement of over 50%, all the way to the multi-core capabilities of Zen 2, the powerful gaming performance of Zen 3, and the high-frequency, high-performance of Zen 4, AMD has been steadily enhancing the performance of desktop processors, while providing users with an even better experience. At ComputeX 2024, AMD officially announced information about the Zen 5 architecture and the Ryzen 9000 series processors, which has been eagerly awaited by countless users in the industry. Now, the detailed information of this brand-new processor is finally officially announced.

Advertisement

AMD's positioning for the Ryzen 9000 series desktop processors is to have ultimate performance and energy efficiency ratio. AMD believes that the Ryzen 9000 series brings powerful performance, excellent energy efficiency ratio, outstanding overclocking experience, and a long-life platform. The first two points come from the Zen 5 architecture and the new process technology. In terms of overclocking experience, AMD has introduced new features that are more finely tuned and practical this time. In terms of long-life, in addition to the newly released AMD 800 series chipset, the Ryzen 9000 series can still be used on the AMD 600 series chipset. Users only need to update the BIOS to use the new processor, which is very convenient.

The Ryzen 9000 series desktop processors released by AMD this time include four models: the Ryzen 9 9950X, Ryzen 9 9900X, Ryzen 7 9700X, and Ryzen 5 9600X. Compared to the previous generation of the Ryzen 7000 series, the new processors do not seem to have significant changes in frequency, core count, thread count, and cache at first glance. However, in terms of TDP power consumption, except for the top-tier Ryzen 9 9950X, the other three models have significantly reduced TDP power consumption compared to the corresponding products of the previous generation of the Ryzen 7000 series. For example, the TDP of the Ryzen 9 9900X has been reduced from 170W of the Ryzen 7 7900X to 120W, and the TDP of the Ryzen 7 9700X has been reduced from 105W of the Ryzen 7 7700X to 65W. At the same time, the maximum boost frequency has slightly increased, which means that there are significant innovations inside the Ryzen 9000 series desktop processors.

As of our press time, AMD has updated the specifications of the Ryzen 9000 series desktop processors on its official website. Interested readers can visit the AMD official website for details.

Zen 5 to the rescue: A brand-new microarchitecture design brings a 16% IPC improvement

In terms of microarchitecture, the Ryzen 9000 series desktop processors adopt the latest Zen 5 microarchitecture, which has undergone significant improvements compared to the previous generation Zen 4, from the front end to the back end. Zen 5 is wider, larger, and has stronger execution capabilities. Let's briefly introduce it here.

The four major design goals of Zen 5 include executing more instructions per cycle, wider overall width, doubling the cache data bandwidth, and AI acceleration. Although it looks like a simple four sentences, it reflects a revolutionary change in the architecture.On the frontend, the focus of Zen 5's improvements lies in the decoder being changed to two 4-width decoders, which can achieve 8-width decoding at the same time. Since the Zen architecture, AMD has been using a single 4-width decoder for several generations, and it has finally been thoroughly improved in Zen 5. The corresponding branch prediction has also been greatly enhanced, including less latency, higher accuracy, and greater throughput. The instruction cache has also become faster and wider due to the adjustment of the decoder. Overall, Zen 5 now has a larger frontend, which brings more optimization and adjustment work, especially in branch prediction. We predict that AMD will make significant improvements and enhancements to the frontend in the Zen 6 architecture.

The changes on the frontend have brought in more instructions, so AMD's execution part is larger and wider. This includes a wider instruction fetch unit, dispatch, and retirement units, more ALU units, and a larger execution window.

In terms of cache and data processing, AMD has introduced a new 48KB L1 12-way data cache with a latency of 4 cycles. In terms of bandwidth, the L1 cache bandwidth has doubled, and the floating-point unit bandwidth has doubled (corresponding to the adjustment of the floating-point execution part). In addition, data prefetch performance has also been strengthened.

In terms of floating-point execution units, AMD has introduced a powerful SIMD unit that supports AVX-512, with a queue depth of 384, and six FADD units with a 2-cycle delay, significantly enhancing the overall FP instruction execution capability. The improvement in floating-point capabilities will bring performance gains for AMD in AI and gaming computing, as well as some scientific computing, which aligns with AMD's design goals.

In terms of IPC improvement, due to the improvements in the Zen 5 microarchitecture, the overall IPC has increased by an average of 16% compared to the previous generation, fulfilling AMD's promise of a "double-digit" IPC increase, with AES XTS increasing by up to 35%, which may be related to the greater bandwidth and more powerful floating-point computing capabilities.

In the overall architectural performance improvement of Zen 5, the performance increase brought by the decoding and execution parts is the greatest, followed by the increase in data bandwidth, and the instruction fetch and branch parts are the least, which may be the focus of improvement for AMD's next-generation microarchitecture.

In terms of manufacturing process, products represented by the Ryzen 9000 series processors, based on the Zen 5 architecture core, will use TSMC's 4nm process, while processors using the Zen 5c small core will be manufactured using TSMC's 3nm process. Overall, the performance improvement of the Zen 5 microarchitecture is evident. Judging from the current information, compared to the "minor changes" from Zen 3 to Zen 4, the changes in Zen 5 are earth-shaking, basically resetting most of the architecture. Wider and larger decoding and frontend, wider and faster execution parts, and significantly strengthened storage parts all contribute to a more excellent comprehensive performance.

Powerful performance: up to 94% productivity advantage, up to 31% gaming advantage.

So how does the performance of the Zen 5 processor perform? Let's take a look at AMD's official test data next.

AMD has provided a large amount of performance comparison data. The Ryzen 9 9900X compared to the Intel Core i9 14900K, AMD data shows that the creator and productivity performance leads the competitor by 2% to 41%, and the gaming performance leads by 4% to 22%. The parameters used by AMD in the comparison are the default Intel settings and default AMD settings, without any special adjustments. Considering the high power consumption of up to 300W under PL2 for Intel processors, the 12-core, 24-thread design of the AMD Ryzen 9 9900X can defeat the 24-core, 32-thread design of the Core i9-14900K in both performance and energy efficiency.Additionally, AMD has also provided performance test results comparing the Ryzen 7 9700X with the Intel Core i7-14700K, which has more cores and threads and higher power consumption. The performance is similar to that of the Ryzen 9 9900X. AMD demonstrated an astonishing lead in the Handbrake test, a video transcoding software, which may be related to AMD's support for the AVX-512 instruction set. Currently, the 13th and 14th generation Core series processors do not provide support for AVX-512. Similarly, whether in gaming or productivity and content creation tests, the Ryzen 7 7700X with an 8-core, 16-thread design outperforms the Core i7-14700K with a 20-core, 28-thread design.

In the comparison of mid-range processors, the overall performance gap is even more significant. We noticed that the Ryzen 5 9600X is merely a 6-core, 12-thread processor, while the compared Core i5-14600K boasts up to 14 cores and 20 threads. However, the transcoding speed of the Ryzen 5 9600X leads the Core i5-14600K by as much as 94%, and it also has a 22% lead in the 7-Zip file compression and decompression test. Additionally, it has a lead of over 20% in gaming with titles like F1 2023 and "Horizon Zero Dawn," which means that AMD processors have significantly better single-core performance, allowing for such performance despite a considerable difference in core counts.

For gaming users, AMD's tests show that the Ryzen 9 9700X has finally surpassed the first-generation processor with 3D V-Cache, namely the Ryzen 7 5800X3D, in gaming performance, with an overall lead of about 12%. Generally, 3D V-Cache has a very significant advantage in games, as it reduces the inherent memory latency of the von Neumann computer architecture, allowing more fragmented data generated in games to be stored and accessed directly by the high-speed 3D V-Cache without having to query the memory. However, the Ryzen 9000 series desktop processors, through improvements in communication performance between the I/O Die (IOD) and the Compute Die (CCD), reducing memory to IOD latency, and enhancements in the CCD microarchitecture, can achieve equivalent results, which is quite exciting.

Exceptional Energy Efficiency: 65W TDP is also powerful

In terms of energy efficiency, since desktop processors are positioned for high performance, both AMD and Intel often introduce very powerful "X," "K," "KS" series, followed by products without suffixes that reduce power consumption. Under such circumstances, the TDP power consumption of desktop processors that ordinary users most frequently encounter is generally above 100W, and some Intel products even soar to around 250W to 300W, which is actually abnormal.

Essentially, it is the competition between manufacturers that has caused the processor frequency and voltage settings to deviate from the "optimal energy efficiency range," leading to a surge in power consumption and even a series of instability events due to high voltage and high frequency. AMD has always been restrained in this regard, with the highest TDP of the previous generation Ryzen 7000 series being limited to 170W. This time, due to new architecture and new processes, the overall performance-to-power ratio has been significantly improved, and even a 65W TDP version has been directly introduced.

In this regard, AMD first introduced that by improving energy efficiency, it has achieved a lower overall platform power consumption, lower system temperature, and a quieter operational experience.

In terms of processor packaging materials, the use of materials with lower thermal resistance has reduced the thermal resistance by 15%, resulting in a 7-degree Celsius temperature reduction at the same TDP.

More specifically, at the same TDP or lower TDP, the Ryzen 9000 series has lower overall power consumption compared to the Ryzen 7000 series, but the performance improvement ranges from 11% to 22%. Especially for the Ryzen 7 and Ryzen 5, the TDP power consumption of the Ryzen 7 9700X and Ryzen 5 9600X is now only 65W, which is also one of the lowest TDP power consumption values for high-performance processors in recent years (excluding special ECO version processors). It is worth noting that the corresponding Intel processors, even the lower-end Core i5-14600K, have a TDP of 125W, and its PL2 maximum power consumption can even reach 181W.

Overclocking is more interesting: Support for DDR5 8000, smarter PBO, and the new Curve Shaper for a cooler experience.For desktop-oriented processor products, overclocking is a topic that cannot be avoided. This time, AMD has introduced new overclocking features for both processors and memory.

On the memory front, AMD has officially supported DDR5-8000 memory and has introduced new memory optimization and overclocking features. All motherboards with the AM5 socket can unlock the memory overclocking capabilities. For enterprise-level users and those who pursue stability, the JEDEC memory specification supports DDR5-5600, which is also a certain improvement over the previous generation.

In terms of processor overclocking, in addition to the advantages such as unlocked multipliers, AMD has also optimized Precision Boost Overdrive (PBO) overclocking. If users have excellent cooling configurations and prioritize performance, especially multi-threading performance, then enabling the PBO overclocking feature on the Ryzen 9000 series desktop processors can bring up to a 15% performance improvement. AMD mentions that if the TDP is lower, the performance gain through PBO will be more significant, such as the Ryzen 7 9700X with a TDP of 65W, 8 cores, 16 threads, and a default frequency of 5.5GHz. PBO should allow this processor to run at high frequencies for a longer period of time or achieve even higher frequencies. We will also conduct a detailed test on this feature.

In addition to the traditional PBO updates, AMD has also brought an updated version of the Curve Optimizer and introduced a new feature called Curve Shaper overclocking adjustment. The Curve Optimizer is used to adjust the AVFS curve of specific processor cores, in simple terms, by under-voltage overclocking to achieve better performance, and it currently supports the Ryzen 9000 series desktop processors.

The latter, Curve Shaper, is mainly used to address issues related to power consumption, temperature, and frequency of the processor under different conditions. Curve Shaper allows users to set different offset voltages and frequencies based on the processor's five working states: minimum frequency, low frequency, medium frequency, high frequency, and ultra-high frequency. This is to ensure that the processor reduces frequency and power consumption as much as possible when not under full load during gaming or daily application use, thereby achieving a higher performance-to-power ratio. However, Curve Shaper adjustments are relatively complex and require more experienced DIY enthusiasts to operate.

AMD 800 Series Chipsets: The choice to change or not change the motherboard is entirely up to you.

Apart from the processors, AMD has also released new 800 series chipset products this time. The reason is not due to processor interface changes or the inability of old motherboards to use new processors as an "excuse," but simply a product specification update and replacement. The previous 600 series motherboards can still use Zen 5 processors.

According to AMD's introduction, AMD's performance in platform stability is very impressive. The AMD 600 series chipset can still use the Ryzen 9000 series desktop processors, and even AMD previously claimed that the interface would not be updated until 2027+, which means users can buy AMD products with more confidence. After all, a processor change can achieve a comprehensive upgrade, which can save users a lot of costs.

The latest AMD 800 series chipsets mainly include the X870 series, B850, and B840. The release of the B850 and B840 has greatly enriched the product offerings in the mid-to-low-end market of AMD motherboards, as the current B650 series motherboards are also relatively expensive, and the A620 is weaker. In terms of specifications, the AMD B850, which supports PCIe 5.0 and USB 3.2 Gen 2x2 technology with a bandwidth of 20Gbps, is expected to become a mainstream choice in the mid-range market. The advantages of the X870E and X870 lie in their support not only for PCIe 5.0 graphics cards and SSDs but also for USB4 technology. This technology is based on the Thunderbolt 3 protocol specification, has a large bandwidth (40Gbps) data transfer capability, and supports video expansion capabilities and PD charging functions equivalent to Thunderbolt 3.

AI Computing Power Enhancement: AVX-512 Support Plays a Major RoleLet's take a final look at the AI support of the Ryzen 9000 series desktop processors. The main focus is on the support for AI-related instruction sets such as AVX-512 and VNNI.

AMD mentioned that the current AM5 platform can achieve synchronized support for multi-GPU and SSD with PCIe 5.0, something that competitor products cannot achieve. The processors on Intel's Z790 platform only support PCIe 5.0 x16 and PCIe 4.0 x4, so SSDs or GPUs cannot all utilize the PCIe 5.0 specification. In practical applications, due to the need for more bandwidth in AI computing, support for GPUs and SSDs with PCIe 5.0 is necessary. AMD's test data shows that after full implementation of PCIe 5.0, there will be approximately a 100% increase in AI performance and a 58% increase in SSD performance.

In terms of AI computing, because the Ryzen 9000 series desktop processors support AVX-512 and VNNI, running large language models like Llama and Mistral on the processor can lead to a 17% to 20% advantage over Intel products. It is worth noting that AMD did not use the most powerful Ryzen 9 9950X, but compared the 12-core 24-thread Ryzen 9 9900X with Intel Core i9-14900K. If a more powerful processor is used, there should still be room for overall performance improvement.

Achieving more with less, the Ryzen 9000 series desktop processors are worth looking forward to.

Well, that concludes the introduction to the Ryzen 9000 series desktop processors. Based on the content and data revealed by AMD this time, we believe the most impressive thing is that the new generation Zen 5 processors can achieve more with less. A 12-core processor can defeat a competitor's 24-core product, an 8-core processor can surpass a competitor's 20-core product, and a 6-core processor can defeat the 14-core Core i5-14600K. Of course, whether this is the case or not? Please continue to follow the "Mini Computer" public account, and we will announce our detailed review around the launch date of the Ryzen 9000 series processors, which is around July 31st. Stay tuned.