The Ryzen 9000 series, known during development by the codename “Granite Ridge,” is the first CPU series to adopt the Zen 5 architecture, succeeding the current Ryzen 7000 series and designed for the Socket AM5 platform, with shipments starting in July 2024.
According to AMD, the Zen 5 architecture boasts significant improvements, including enhanced branch prediction, expanded execution pipelines, and larger windows for parallel execution, emphasizing a “2x” increase in various areas compared to the previous Zen 4 architecture.
These improvements include doubled bandwidth for front-end instruction and data caches, and AVX-512 instruction throughput. The instructions per clock (IPC) performance is also up over Zen 4 by 16% on average, with notable gains in parallel-intensive applications like “CINEBENCH R23” and “Blender,” and even a 10% IPC increase in games like “Far Cry 6.”
Continuing the trend from previous generations, the AMD Ryzen 9000 series will debut with four primary models. Leading the lineup is the 16-core, 32-thread Ryzen 9 9950X. The series also includes 12-core, 8-core, and 6-core models, maintaining the same configurations as the preceding Ryzen 7000, Ryzen 5000, and Ryzen 3000 series.
While the L2 and L3 cache sizes remain unchanged from the Ryzen 7000 series, the lower two models in the Ryzen 9000 series see a 100MHz increase in maximum boost clock speeds. Interestingly, the 8-core Ryzen 7 9700X comes with a reduced TDP of 65 watts, similar to the 6-core Ryzen 5 9600X.
Previously, with the Ryzen 7000 series both the 8-core Ryzen 7 7700X and even the 6-core Ryzen 5 7600X both featured a higher TDP value of 105 watts, no doubt that these upcoming processors will be more energy efficient, there will likely be a larger discrepancy in terms of performance between them and higher SKUs.
This highlights that, like Intel, AMD are now offering lower-end hardware with significantly inferior binning, scrutinizing consumers who spend less on their products.
Another interesting detail is the complete absence of any announced base frequencies for the Ryzen 9000 series, which appears to be a first for AMD. Consumers only have the maximum boost frequencies to rely on, and these processors are unlikely to sustain those boost speeds across all cores for extended periods, similar to how Intel’s processors operate with significant differences between base and various boost frequencies.
This is something to be cautious about if you are considering an upgrade.
Despite the similarities with the Ryzen 7000 series, AMD claims substantial performance improvements in the Ryzen 9000 series. The company asserts that the Ryzen 9 9950X surpasses the gaming performance of the Core i9-14900K, the strongest in the current generation, and excels in creator-oriented applications due to the full utilization of PCI Express 5.0 bandwidth.
AMD’s comparisons between Zen 5 and the Intel Core i9-14900K were conducted with 6000MHz DDR5 memory paired with a Radeon RX 7900 XTX graphics card. Interestingly, AMD also used Intel’s recently imposed default setting, which has been shown to hinder gaming performance by a few percentage points.
It’s important to note that AMD’s gaming performance numbers seem skewed in their favor, as all companies use tactics to make their products appear superior to the competition.
For example, Intel CPUs benefit more from high-speed memory than Ryzen. Investing in 7200+ MHz DDR5 memory for an Intel processor will likely result in noticeably increased gaming performance.
However, on a level playing field with stock settings, or even with the Intel Core i9-14900K running with 6000MHz memory, which is above its official specification, the Ryzen 9 9950X still excels in content creation, boasting 55% performance gains in Handbrake and 56% in Blender. Gaming performance is mixed, with the Ryzen 9 9950X outperforming by 4% in Borderlands 3 and up to 23% in Horizon Zero Dawn.
If AMD had conducted these performance tests with a faster graphics card, such as NVIDIA’s GeForce RTX 4090, the figures for AMD versus Intel would likely be even higher.
Alongside the Ryzen 9000 series, AMD introduced new chipsets for Socket AM5: the AMD X870E and AMD X870. Despite the previous generation utilizing 600-series chipsets, AMD seems to be following Intel’s naming schemes this generation.
The X870 chipset will support USB4 with a maximum bandwidth of 40Gbps and the latest EXPO memory profiles. Unlike the previous X670E chipset, which required PCIe 5.0 only for GPU use, both the X870E and X870 support PCIe 5.0 for both GPUs and storage.
The Ryzen 9000 series will remains compatible with existing 600 series motherboards as the AM5 socket is once again designed for long-term support, with AMD extending the lifetime of the Socket AM5 platform from 2025 to 2027.
Unlike Intel, who will be abandoning their LGA 1700 socket after just two generations, the 12000-series Alder Lake and the 13000-series Raptor Lake processors.
I don’t see value in rebranding and re-releasing the same processors as a 14th generation as “extending” longevity for a platform, especially considering Intel scrapped plans to release a desktop variant of Meteor Lake.
To solidify their leadership in terms of platform longevity, AMD announced new CPUs for the previous-generation Socket AM4 platform: the Ryzen 5000XT series, set to launch in July 2024. This series includes the 16-core, 32-thread Ryzen 9 5900XT and the 8-core, 16-thread Ryzen 7 5800XT.
The Ryzen 9 5900XT offers a significant upgrade from the 12-core, 24-thread Ryzen 9 5900X, providing an affordable 16-core upgrade path for AM4 users. This decision might be influenced by AMD’s surplus stock, but regardless, it offers users an attractive option.
In the realm of affordable processors, the Ryzen 5000 series continues to reign supreme, especially with the addition of the Ryzen 7 5800X3D, celebrated as the budget gaming champion. However, with the introduction of revised 8-core and 16-core offerings, AMD aims to provide affordable alternatives that rival the gaming performance of Intel’s older 12th Generation CPUs.
Certainly, the current hardware landscape wouldn’t be complete without discussions of artificial intelligence, a buzzword that NVIDIA, Intel, and AMD are all investing in. Hence, the announcement of the “Ryzen AI 300 series,” high-performance APUs tailored for notebooks.
However, the naming scheme is perplexing, reminiscent of Intel’s revised Core Ultra branding, which is confusing. The Ryzen AI 300 series, dubbed “Strix Point,” stands out as one of their significant announcements. These SoCs fully support AI-based features integrated into the next generation of Windows, featuring an XDNA 2-based NPU.
Microsoft’s upcoming Windows features such as Copilot+, which aims to assist players mid-game alongside Microsoft’s invasive “Recall” function which acts like a keylogger on steroids alongside Live Captions & Real-time Translation, and Co-creator will necessitate NPUs with at least 40 TOPS performance.
The Ryzen AI 300 series, boasting an industry-leading 50 TOPS NPU based on XDNA 2, meets this requirement. This benchmark is also being met by next-generation processors from Intel, as well as ARM-based processors.
The series includes two models: the high-end Ryzen AI HX 370 and the standard Ryzen AI 9 365. The HX 370 features a powerful 12-core, 24-thread CPU based upon the Zen 5 architecture, though with different cache capacities compared to the standard Ryzen 9000 series which has been a common staple with these mobility Ryzen processors as they are monolithic in design, however it will also be featuring a combination of traditional Zen 5 cores and denser Zen 5C cores which have a decreased cache pool.
Both models support a wide TDP range from 15 to 54W, indicating flexibility for various notebook designs, like Microsoft’s Surface.
The Ryzen AI 300 series also boasts enhanced integrated GPUs, with a newly revised RDNA 3.5 architecture providing significantly improved performance per watt, the Radeon 890M and Radeon 880M integrated graphics options offer substantial increases in Compute Unit (CU) counts, promising impressive gaming capabilities even without discrete GPUs.
AMD’s emphasis on the advanced NPU capabilities of the Ryzen AI 300 series positions it strongly against competitors like Qualcomm’s Snapdragon X Elite and Intel’s Core Ultra processors. With support for major AI models and improved efficiency, the Ryzen AI 300 series aims to deliver high performance for next-generation AI applications on Windows.
Overall, AMD’s announcements at COMPUTEX 2024 signal significant advancements in both desktop and notebook computing, with a strong focus on integrating cutting-edge AI capabilities and improving overall performance and efficiency.
The initial batch of mobile devices featuring the Ryzen AI 300 series is set to debut in the upcoming quarter. However, it might be prudent to hold off on investing in Zen 5 processors for the time being. Without clear performance metrics for all four desktop SKUs, especially regarding base frequencies which appear to be potentially low, making a fully informed decision is challenging.
As it stands, the gaming performance enhancement compared to Intel’s Core i9-14900K, let alone Zen 4, isn’t particularly remarkable. However, for multithreaded applications, the scenario is quite different.
Many contemporary video games suffer from poor optimization, often relying heavily on upscaling technologies like NVIDIA’s DLSS, AMD’s FSR, and Intel’s XeSS to maintain acceptable frame rates. Coupled with an influx of political messaging within the games themselves, the overall gaming experience is compromised.
This diminishes the excitement surrounding new hardware releases, reducing the incentive to upgrade. Moreover, it’s highly likely that AMD will soon introduce X3D V-Cache versions of these CPUs. Intel’s upcoming 15th generation Arrow Lake CPUs, while not groundbreaking, are expected to arrive later this year, adding another layer of consideration.
For those primarily focused on gaming, it may be wise to delay purchasing decisions and wait for potential X3D-based Zen 5 processors later in the year. Personally, I find myself content with my current gaming setup and am only considering an upgrade to accommodate emulators, which heavily rely on AVX-512 instructions.
In this regard, Zen 5 is expected to deliver solid performance due to its increased AVX512 throughput. Additionally, for titles like Skyrim that heavily favor memory size and bandwidth, an expanded L3 cache pool offers a significant performance boost, serving as a viable alternative to insufficient system memory and VRAM.
