AMD Ryzen 11,000 Zen 6 desktop CPUs tipped to launch in 2027

[Redaktion]

A report from Benchlife says AMD plans to showcase its next-generation desktop CPUs in 2027. They were originally slated to launch in 2026, two years after Zen 5 surfaced in 2024.

https://www.notebookcheck.net/AMD-Ryzen-11-000-Zen-6-desktop-CPUs-tipped-to-launch-in-2027.1231076.0.html

[HDPacks]

Isn't 10,000 series Zen 6 and 11,000 series Zen 7??

[Aluskinner]

On the laptop side, the Zen 6-based Medusa Point

If Medusa Point is based on RDNA4 and comes without transformer model based upscaling, which NVIDIA introduced some time ago already and NVIDIA's N1(X) APU is a thing this year already, I'm not sure I'd be interested in Medusa Point much. Not even talking about CUDA. Also, AMD needs to catch-up to NVIDIA' ray-tracing and path-tracing power efficiency:
pcgameshardware.de/Radeon-RX-9070-XT-Grafikkarte-281023/Tests/vs-RTX-5070-Ti-5080-Release-Benchmark-Overclocking-1467630/3:
Right now, depending on the game, NVIDIA is 1.3 to 2.88 times more energy efficient in path-tracing, the more path-tracing-heavy a game is, the more behind is AMD.
5070 Ti is 300 Watt TDP
RX 9070 XT is 304 Watt TDP
As such, one can compare the FPS/Watt on that table easily. RDNA4 improved nicely over RDNA3 in path-tracing power efficiency, but is still behind NVIDIA by quite a lot (several GPU generations or one could also say several full node jumps (a node jump is a 25% to 33% power efficiency improvement)). So, if you care about path-tracing (not to be confused with ray-tracing) (in ray-tracing AMD is still behind, tho, by like 1 GPU generation or 1.26 times on average, according to the table), the choice to get a NVIDIA GPU is clear.

[Reserved]

Isn't 10,000 series Zen 6 and 11,000 series Zen 7??

Maybe they're reserving 10k for something else. Sorta like how they went 7k -> 9k for desktop CPUs then released the 8k desktop APU series. Or how they went 3k -> 5k desktop CPUs and the 4k was released as a mobile laptop APU series.

Nobody can really explain the logic behind AMDs naming or marketing decisions. It's truly something special.

I think if they're releasing zen 6 desktop this late (2027), it's almost guaranteed that consoles will get delayed too. Consoles almost never get the latest cpu architecture when they just launch but wait a year or more so they can get cheaper pricing. So I'm expecting PS6 2028 now.

[context]

If Medusa Point is based on RDNA4

Not RDNA4, RDNA4m. There's a difference. RDNA4m seems to be more like a hybrid architecture of RDNA3.5 + with RDNA4's matrix ML units for hardware based AI upscaling.

Confusing, I know but it's AMD we're dealing with here. RDNA4m is a marketing term it really should be called RDNA3.6 or RDNA3.7 architecture.

NVIDIA's N1(X) APU is a thing this year already

I'm not sure what to believe anymore. It's been leaked several times over the years and then complete dead silence for many months repeatedly now. I'm beginning to get slightly worried and think the rumours are correct. That this is a troubled launch and the product has serious issues.

Maybe that's why they decided to partner with intel? Unfortunately, isn't looking like the year of arm laptops, just yet..

AMD needs to catch-up to NVIDIA' ray-tracing and path-tracing power efficiency
...
RDNA4 improved nicely over RDNA3 in path-tracing power efficiency

Apparently, RDNA5 will address this and be an even bigger leap for RT than 3 -> 4 was. But we'll see, need to keep the hype expectations in check. Seems somewhat believable as Sony is pushing hard for RT performance in their upcoming playstation console.

I don't think having good PT performance or efficiency matters too much *right now*, it's still incredibly heavy, unless you've a xx90 class dgpu which most people don't, so they just avoid it completely. Intel has good RT perf. and struggles to get playable fps in PT modes as well.