The first Ryzen 7000 chips go on sale Sept. 27, and AMD just made official the initial specs and lots of new platform details. Here’s the latest on the new architecture, socket, and silicon, coming soon to power-user and gaming PCs near you.
At a press and analyst event in Austin, Texas, AMD unsealed its plans for its much-anticipated Ryzen 7000 desktop CPU line, code-named “Raphael.” First tipped at CES 2022, the Ryzen 7000 will be the first set of consumer chips to employ the new “Zen 4” microarchitecture with leading-edge process technology. The first chips in the new line will target power users and gamers.
The Zen 4 essentials were shared by AMD’s Chair and CEO Dr. Lisa Su, and CTO Mark Papermaster. Built on 5-nanometer process technology from TSMC for its CCDs (the I/O portion, home to the PCIe 5.0 and memory controllers, is on 6nm), the Ryzen 7000 is the first large-scale architectural advance from AMD on its desktop processors since 2020. November of that year saw the launch of “Zen 3” to consumers, in Ryzen 5000 desktop (“Vermeer”) chips like the Ryzen 9 5900X, and Zen 3 was expanded to the company’s workstation-grade Threadripper Pro chips in 2022. Zen 3 came to laptops after the initial Zen 3 launch, in the 5000-series-mobile “Cezanne” line (the first of which debuted in January 2021) and a later refresh in the form of Ryzen 6000 H- and U-series mobile CPUs (“Rembrandt”) in early 2022. (See our initial tests of laptop samples bearing Cezanne and Rembrandt mobile chips.)
Now at the end of 2022 we will get our first taste of something new from AMD after two long years of waiting. A Ryzen 6000 desktop-CPU family never materialized, though some integrated-graphics-equipped (IGP) Ryzen 5000 CPUs and enhanced Ryzen 4000 chips (formerly OEM-only) fleshed out the desktop line between 2020 and present. Thus the pent-up demand for more detail on AMD’s latest big chip update, coming on the heels of Intel’s own largely successful 12th Generation “Alder Lake” launch (and its upcoming 13th Gen Core).
2022’s Ryzen 7000 family is not just a long-awaited refresh to the Ryzen line, but AMD is bringing with it more new base-level technologies than any launch since the original, revolutionary Ryzen 1000 series in 2017. AMD broke it all down at its launch event. Strap in…it’s a lot!
First, the CPUs: Ryzen 7000 Desktop Specs and Models
Let’s first take a look at the first Ryzen 7000 chips AMD made official. Plenty of leaks have hinted at the identity of these chips in the run-up to today’s event, but here is the initial launch lineup. It comprises four distinct chip SKUs in the Ryzen 9 (two of those), Ryzen 7, and Ryzen 5 lines…
AMD Ryzen 7000 Desktop CPUs 2022: Spec Breakout
|AMD Ryzen 9 7950X||AMD Ryzen 9 7900X||AMD Ryzen 7 7700X||AMD Ryzen 5 7600X|
|Maximum Boost Clock||5.7GHz||5.6GHz||5.4GHz||5.3GHz|
|Cache||80MB (L2+L3)||76MB (L2+L3)||40MB (L2+L3)||38MB (L2+L3)|
|Onboard Graphics||RDNA 2-Based||RDNA 2-Based||RDNA 2-Based||RDNA 2-Based|
|CPU PCIe Lanes||24||24||24||24|
Now, these are clear enthusiast-grade processors with obvious parallels in the Ryzen 5000 series (the Ryzen 9 5950X and 5900X, Ryzen 7 5700X, and Ryzen 5 5600X). You will note the lack of chips ending in “G,” which has traditionally signified on-chip graphics for a select few Ryzens. Now, on-chip Radeon Graphics (based on “RDNA 2” architecture) is the norm, not the exception. (More about that in a bit.)
One thing that is not different is the classification of cores: They are all of one undifferentiated type. With its 12th Generation (“Alder Lake”) CPUs, Intel in most of its mainstream and higher-end desktop and laptop chips established two classes of core making up the die: Performance cores, and Efficient cores (P-cores and E-cores, respectively). The two types are designed for different classes of task; relegating easy or non-time-sensitive tasks to E-cores can save energy. (E-cores also do not support multithreading.) The P-cores and E-cores can also be deployed together for peak multithreaded muscle in applications designed to leverage them both. Here, the core types are one.
AMD uncorked some general, family-wide Zen 4 performance claims, which we’ll outline here; following that are some claims on the highest-end and lowest-end Ryzen 9 and Ryzen 5 CPUs. Overall, against the Ryzen 5000, AMD is claiming an approximate 13% IPC uplift vs. those Zen 3 processors on average, and a 29% uplift on single-threaded performance vs. same. (The IPC uplift figure is a few percentage points higher now than the company’s original claims from Computex 2022.) The top-end boost clock, on the Ryzen 9 7950X, also tops that of the 5000 series by a whopping 800MHz.
The chip maker also shared the following Geekbench 5.4 single-thread performance projection for the four Zen 4 chips vs the Intel Core i9-12900K, Team Blue’s Alder Lake flagship chip…
The single-threaded performance boost is the most interesting aspect of these claims, given that that has traditionally been an AMD sore point vs. equivalent Intel chips. The known effect of good single-threaded performance on games, all else being equal, is also an intriguing factor here.
AMD Ryzen 9 7950X: The New Flagship
The 7950X is the 16-core flagship. Here, AMD is claiming some pretty healthy upticks in selected games and content-creation applications vs. its own last-generation equivalent, the Ryzen 9 5950X…
Given that there is no change in core/thread count here, the content creator upticks raise some intrigued eyebrows. And the last shared claim around the 7950X, a V-Ray rendering test vs. the 16-core/24-thread Intel flagship, is especially eye-opening…
AMD Ryzen 5 7600X: A Gaming Teaser
Meanwhile, the “bottom” of AMD’s initial stack of Zen 4 chips is this Ryzen 5, for which AMD supplied its own comparisons with Intel’s Core i9-12900K (note, that’s not a mistake: that’s the Core i9, not a Core i5) in F1 2022 at 1080p, in which the 7600X came out the victor…
In a slightly expanded group of games (all also at 1080p), the Ryzen 5 7600X topped the Core i9-12900K in several other titles, notably Rainbow Six Siege by the largest margin…
Overall IPC Uplift
The company also shared a breakout of its generational IPC uplift claims. As noted above, from Zen 3 to Zen 4 across 22 representative desktop applications in content creation, gaming, and a smattering of compute benchmarks, the Zen 4 is showing a 13% uplift (a geomean of the 22 applications). The chips were tested at a 4GHz fixed frequency, with eight cores and 16 threads.
And here you can see a breakout of how, AMD claims, various aspects of the design contribute to the rated IPC uplift…
Ryzen 7000 Power Consumption
One questionable aspect of AMD’s new Ryzen 7000-series processors is their power consumption. The two Ryzen 9 SKUs at launch both have TDPs of 170W, which is markedly higher than the 105W TDP of the Ryzen 9 5950X. In spite of this increase, however, AMD reported significant improvements in energy efficiency.
AMD claimed that the Ryzen 7000-series processors will be able to operate with 62% less power while maintaining the same level of performance as Ryzen 5000-series processors. Alternatively, Ryzen 7000-series processors can deliver up to 49% better performance than Ryzen 5000-series processors while consuming the same amount of power.
AMD took this a step further and said that the Ryzen 9 7950X operating with a 170W TDP would provide an overall improvement of 35% over a Ryzen 9 5950X. Though some of these statements may seem contradictory, they could actually all be true. It’s perfectly possible for the CPU to use more power and deliver more performance while also being more energy efficient. If it is able to complete work faster, it will be able to slow down and drop its power draw and conceivably save more power overall.
The company also took some potshots at Intel Alder Lake on die area and energy efficiency…
To be certain, we will need to run some tests once we get one of these chips in hand to better understand its performance and efficiency. It’s also worth noting that AMD mentioned this increase in TDP was to support turbo clock speeds, which could indicate it will reach the 170W TDP for sporadic brief periods. Last, we also cannot ignore that Intel has also been raising the power ceiling on its processors in recent years, and as clock speeds continue to rise a bump in TDP may be a regular feature of new processors going forward.
The New AM5 Platform: Ryzen (Finally) Reboots Its Socket
AMD has long prided itself on the remarkable consistency of its AM4 platform for its desktop processors. As Intel continued its cadence of new, incompatible sockets every two generations (meaning no Intel motherboard could remain cutting-edge, or even upgrade-possible, for long), AMD kept its AM4 socket and chipsets viable to various degrees since the platform’s debut with the original Ryzen 1000s in 2017. Not every AM4 motherboard and every chip is interoperable up and down the line, but the upgrade viability of AM4 has long been one of its major strengths.
That long ride ends with AM5, but for what look like good reasons. Ryzen 7000’s new platform brings a bunch of wholesale-new elements to a variety of fronts. It’s not just a simple socket swap.
The most physical and obvious manifestation of AM5, though, is simple: AMD is moving away from a pins-on-the-CPU (“pin grid array,” or PGA) die design to what Intel has used for umpteen generations now for its mainstream desktop processors, since 2004: a land-grid-array (LGA) design with contacts on the CPU and the pins (“lands”) living inside the motherboard socket. The new AM5 socket will be a 1,718-pin design supporting up to 230 watts of power delivery.
This isn’t AMD’s first dance with an LGA design. Its high-end Ryzen Threadripper and Threadripper Pro families use pins in the socket for their two sockets, the older TR4 and the newer sTRX4 (along with an elaborate load mechanism to keep the 4,000-plus pins safe). And AMD has employed LGA sockets for its EPYC server designs. But with AM5, the Ryzen chips get a new integrated heat spreader (IHD) design and those smooth bottom surfaces that are long familiar to users of Intel desktop processors.
How this change to the chip IHD and the tweaked socket sizing plays out with existing CPU-cooling hardware is another matter. Some makers of coolers, such as Noctua and Gigabyte, have laid out explicit AM5 compatibility plans for their existing AM4 cooling gear, and AMD is claiming compatibility with older AM4 coolers in a general sense. But whether we see distinct AM5-optimized cooling designs remains to be seen, especially given the new and unique shape and design of the IHD.
AMD also did not detail whether any stock coolers will be shipped in the box with these chips at retail (classic AMD Wraiths, or otherwise). If precedent is any guide, though, these initial high-end X-class Zen 4 chips will come without a stock fan.
On-Chip Graphics: Is It APUs All Over Again?
Since the launch of Ryzen in 2017, AMD has always had a small subset of desktop Ryzen CPUs with on-chip graphics in its line, powered by Radeon Graphics and distinguished by a “G” at the end. (See for example our review of the Ryzen 5700G, one of the most recent such efforts.)
With the Zen 4 Ryzen 7000 desktop line, on-chip graphics now comes standard. Most mainstream Ryzens in AMD’s 1000 to 5000 series were designed to run with a dedicated video card, providing no graphics accelerator or display adapter on the die. That changes now.
The difference here, beyond the wide inclusion of graphics on the 7000 series CPUs, is the move to RDNA 2-based graphics. This architecture, which debuted on the Radeon side of the house with the Radeon RX 6000 (“Big Navi”) line of dedicated graphics cards, is AMD’s riposte to Intel’s introduction a few years back of its Iris Xe graphics silicon as part of its mobile (and some desktop) CPUs.
We don’t have any hard specs on the integrated graphics processors (IGPs) quite yet, though. That said, AMD characterizes the RDNA 2-based graphics as display-adapter-grade, not a gaming solution. It’s designed to act as a video output for content-creator PCs, or as a troubleshooting backup for systems packing discrete cards. But it’ll be standard issue on the first four Ryzen 7000 desktop chips, and presumably following ones. AMD did not disclose any plans to also bring to market binned versions of the chips with the IGP disabled, along the lines of the Intel F-series.
As disclosed earlier on with midyear reveals from AMD, the RDNA 2-based graphics should support up to four display outputs, in a mix of DisplayPort and HDMI 2.1. Individual motherboard I/O designs, though, will dictate the practical physical connectivity.
PCI Express 5.0: New Day, New Bus
We have yet to see the impact of PCI Express 4.0 on the core graphics card market (though both previous Intel and AMD platforms support it); where it has evolved faster, in the short term, is of course storage, where M.2 SSDs supporting the PCIe 4.0 spec can attain peak speeds of 3,500MB to 7,000MB per second, depending on the drive. With the introduction of PCIe 5.0 to varying degrees on the first set of supporting chipsets for the platform (B650, B650 Extreme, X670, and X670 Extreme), this should open the door for even faster storage solutions.
PCIe 5.0 support will be governed primarily from the chipset side of things, not the CPU. Expect the most robust PCI 5.0 support from X670 Extreme and B650 Extreme motherboards, with X670 and B650 offering lesser support for PCIe 5.0. In short, the Extreme versions of the chipsets will support PCIe 5.0 for both the GPU slot and for PCIe-based storage, while the non-Extremes will be 5.0 only for storage. (See our first look at flagship AM5 motherboards.) Note that the B650 Extreme had not been disclosed until this point in time.
Admittedly, these will likely be niche concerns for the near future, as PCIe 5.0 drives aren’t yet on the consumer market (a few have appeared in the enterprise space). But that is changing soon, too. AMD disclosed availability of PCIe 5.0 SSD from a host of storage makers, starting in November. The list of SSD makers, shown below, will include stalwarts in the US market such as Asus, Corsair, Crucial/Micron, Gigabyte, MSI, PNY, Seagate, and Sabrent.
New Memory, New Profiles: DDR5 and ‘AMD EXPO’
Intel was the first mover in the transition to desktop DDR5, with the new form of memory debuting with its 12th Generation (“Alder Lake”).
AMD’s Zen 4 desktops chips aren’t its first DDR5 processors, though: That honor goes to the Ryzen 6000 Rembrandt series of mobile chips, which brought DDR5 to the AMD fold at the start of 2022. (It was announced at CES 2022.) However, the Zen 4 desktop CPUs represent the transition of the desktop CPUs to wholesale DDR5. Unlike Intel Alder Lake, where motherboard makers had the option for DDR4 or DDR5 implementations, with Zen 4 Ryzen 7000, it’s all DDR5, all the time.
With the DDR5 transition, AMD is introducing a new technology it is dubbing AMD EXPO. EXPO stands for Extended Profiles for Overlocking, and you can think of it as roughly analogous to XMP, in which set profiles allow for easy memory speedups without manual configuration. We’ll have a followup story on AMD EXPO as soon as we can rustle one up, after this initial summary.
Near term, AMD expects 15 discrete SKUs of AMD EXPO memory kits to debut alongside the first Zen 4 CPUs. You can see some samples of the memory makers who will be contributing to the initial EXPO launch.