7 GHz? AMD aims high with Zen 6 clocks, but there’s a catch…

Rumors suggesting that the next generation of AMD CPUs based on the Zen 6 architecture could surpass 6 GHz have been out for quite some time (while today, Zen 5 cores top out at 5.7 GHz). However, the YouTube channel Moore’s Law is Dead has now thrown more fuel on the fire, claiming these CPUs could reach 7 GHz or even higher. But before you fall for the hype, it’s important to note that reality is likely to look a bit different.

According to one of Moore’s Law is Dead’s latest videos, AMD is allegedly targeting 7.0 GHz or more for Zen 6 cores—a figure he attributes to multiple sources (though, as always, the flavor wording of such videos should be taken with a grain of salt). Elsewhere in the same video, he states that 7 GHz is actually more of a middle of the target range with some margin for deviation, suggesting that the uppermost optimistic target could be even higher.

However, it’s almost certain that these goals don’t reflect what will end up in the official specs—actual clock speeds are likely to be significantly lower. AMD appears to be setting these targets aggressively, possibly even knowing they represent a best-case but unlikely scenario. In practice, it’s almost a rule that real-world clocks fall short. One of the most well-known examples of missed frequency expectations is the RDNA 3 GPU architecture. Its performance was expected to be higher, and there are official slides stating the architecture was “designed for clocks above 3 GHz” as an evidence.

Leaker Kepler_L2 also noted that AMD had internal targets that were loftier than the eventual reality for the clock speeds of the well-received Radeon RX 9000 GPUs based on RDNA 4 architecture. According to them, AMD aimed for a target clock of 3.46 GHz for the Navi 48 chip roughly a year before launch. In the Radeon RX 9070 XT, however, the GPU only reached around 3.1 GHz, and that’s based on factory-overclocked models. For the reference design, the boost clock is just 2.97 GHz—half a gigahertz below the intended target.

This serves as a good illustration of how to interpret “target” clock speeds. While RDNA 3 may have been more of an exception, falling slightly short of these targets seems to be the actual norm to be expected—even in otherwise successful products. That means the preliminary goals quoted by MLID are likely something we should understand as stretch goals rather than realistic expectations. That’s the lens through which the supposed 7 GHz for Zen 6 should be viewed.

Kepler_L2 also shared internal target figures for earlier Ryzen generations that reveal a similar pattern. Zen 3 allegedly had a goal of 5 GHz—which is the one case that nearly panned out, with the Ryzen 9 5950X officially boosting to 4.9 GHz but actually sometimes reaching up to 5050 MHz in real-world conditions. Zen 4 and Zen 5 cores supposedly aimed for 6 GHz, yet only achieved up to 5.7 GHz. Kepler_L2 agrees with Moore’s Law is Dead that 7 GHz is likely the stretch goal for Zen 6—but based on precedents, the actual clock the processors will reach as a product is almost certainly going to be lower.

6.4 GHz more likely?

It’s more reasonable to expect that even if Zen 6 breaks the 6 GHz barrier, a good result would be something like 6.2–6.3 GHz (a 10% gain over Zen 4 and 5), or perhaps 6.4 GHz in top-end desktop models. In fact, Moore’s Law is Dead does note that the CPUs will reach “at least 6.4 GHz” even should the 7GHz target not be fully met (which, given the context, seems like a near-certainty). So, if we are to take the video seriously at all, 6.4 GHz may be the most realistic takeaway from it.

• Tip: AMD to finally add more cores with Zen 6, up to 24 in desktop?
• Tip: AMD Zen 6 to increase cache size. X3D even better in games

Laptop models will, as usual, clock lower—and perhaps that’s what should concern us more, since the world continues to shift from desktops to laptops. In gaming PCs, where desktops still dominate, the key figure will in turn be the clock speeds of X3D processors—which is also a different matter entirely compared to the peak clocks of non-X3D flagships like the Ryzen 9 9950X.

Manufacturing processes used for Zen 6

One more factor should temper expectations: Moore’s Law is Dead still claims Zen 6 will be built on the N2X node. The thinking is that an “X-class” process must automatically enable higher clocks. That assumption may be mistaken though—in reality, AMD is expected to use the N2P process node or the baseline N2 variant, because N2X won’t be available until a year later (2027) and won’t be in mass production when Zen 6 is scheduled to launch.

• Tip: AMD boasts the first 2nm processor—First “HPC” one, that is

Kepler_L2 also referenced details about which versions of Zen 6 CPUs will use which manufacturing processes (there’s not just one). These details were apparently shared with AMD partners earlier and also appeared in previous Moore’s Law is Dead reports. Here’s a summary:

• Venice (Zen 6) and Venice Dense (Zen 6c) server processors will use dies made on the N2P process.
• Desktop Ryzen CPUs under the “Olympic Ridge” name will also use N2P for their CPU dies.
• High-performance laptop CPUs derived from them, the “Gator Range” series (likely the HX lineup), will use N2P-based dies—probably the same silicon as Olympic Ridge.
• High-end mobile APUs under the Medusa Point name will use N3P for IO and iGPU die, and N2P for CPU dies.
• Lower price-point mobile APUs will be monolithic and manufactured using the N3P process.

To be more precise, AMD may not be using the N2P process exactly as defined by TSMC. Much like Nvidia, AMD is likely to collaborate with TSMC to implement a customized version—the technology could possibly be classified as N2 but with additional tweaks, effectively making it something close to N2P. In theory, some experimental enhancements originally intended for the future N2X node could also be incorporated, but probably only to a limited extent.

Sources: Moore’s Law is Dead, Kepler_L2 (1, 2, 3, 4)

English translation and edit by Jozef Dudáš

⠀