Article AMD Zen 6 – The Next Big Leap in Desktop Processors

[gOJDO_n]

Thread is closed. There is already a thread on the topic - Mod DAPUNISHER

AMD Zen 6 – The Next Big Leap in Desktop Processors

​

In the sea of tech news, special attention is drawn to AMD’s upcoming architecture — Zen 6, codenamed Magnus. It is the successor to the current AMD Zen 5 architecture and is designed for the existing AM5 (DDR5) platform. The first desktop processors based on Zen 6 are expected to hit the market at the end of 2026 or early 2027.

What sets this architecture apart is that, after more than seven years, we will finally get AMD desktop processors with more than 16 cores — up to 24 cores, representing a historic leap for AMD’s desktop platform.

​

Retrospective: From Zen 1 to Zen 5​

The Birth of Ryzen: The 2017 Revolution​

Until the appearance of Zen (Ryzen 1000 series), multi-core processors were a luxury reserved for expensive Intel workstation solutions.

At that point, “ordinary mortals” could finally afford a powerful render station at just a third of the cost of Intel’s current workstation solutions. This move drastically changed market dynamics and shook the entire industry. Intel, which had absolutely dominated the market for over a decade, was no longer the only option — in almost every segment. In its lethargic style, Intel failed to take timely and concrete steps to maintain its leading position. After ten years, real competition finally arrived — from a direction they didn’t expect.

Buoyed by its past success and stuck in its comfort zone, Intel underestimated the threat that Zen posed and had no idea what was coming next. First came Zen+ (Ryzen 2000 series) — an evolutionary and relatively minor upgrade meant to keep the hardware scene alive while AMD tirelessly prepared its next big blow — Zen 2.

Zen 2: The Introduction of the Chiplet Design​

Ryzen 3000, Zen 2, 2x CCD, 4x CCX​

With Zen 2 (Ryzen 3000 series), AMD made a huge technological leap forward: implementing a chiplet design where CCDs (with cores) were physically separated from the IOD (I/O die). The Ryzen 3950X in 2019 brought 16 cores and a huge productivity boost.

From that point onward, the fates of the two companies began to shift — but in opposite directions. Today, Intel and AMD have positions that are the exact opposite of what they were a decade ago. Intel, once worth up to ten times more than AMD, now lags behind — AMD’s market cap is now almost three times higher than Intel’s.

Over the last decade, AMD’s processors have seen continuous technological evolution, while Intel, like a “sleeping beauty,” found itself in an unfavorable market and economic situation. The company kept losing market share, making poor strategic and technological decisions — some fatal and difficult to recover from. Among them was the infamous 10nm production process, which was a dead end from the start. The inert and inflexible “giant,” carried away by its vanity (“we manufacture our own CPUs”), didn’t just lose money and precious time — it lost all the technological advantage it had built up over decades.

With no real competition, a new reality emerged where AMD found itself in a very comfortable position, holding the strings. Much like Intel when it had no competition, AMD slowed down Zen development, opting for smaller, incremental upgrades rather than revolutionary steps. Zen 3, Zen 4, and Zen 5 brought just enough improvements to keep the market interested — similar to Intel’s 6th through 10th generations.

Zen 3 and Zen 4: Evolution, Not Revolution​

Zen 2 CCD vs Zen3 CCD ​

The key change in Zen 3 was fusing the two 4-core CCXs with 16MB of L3 cache each into one 8-core CCX with a unified 32MB L3 cache. This, along with a long list of microarchitectural tweaks, led to significant performance gains in games and software that used more than 4 cores and relied on inter-core communication. Zen 3 also introduced PCIe 4.0 support, keeping it relevant even today.

Zen 4 brought modest micro-architectural improvements but fair performance leap. The most significant change was the transition to the new AM5 platform with DDR5 memory. Other notable upgrades included the simple but effective implementation of AVX-512 instruction support, doubled L2 cache, a more advanced manufacturing process enabling 10–15% higher clock speeds, and PCIe 5.0 support.

Zen 5: A Cosmetic Upgrade​

Zen 5, was built on a more advanced production processes (CCD and IOD chiplets), but didn’t bring any major architectural changes:

- Same number of cores as Zen 2/3/4
- Same cache sizes as Zen 2/3/4
- Similar frequencies as Zen 4
- Microarchitectural refinements, including 1T AVX-512 execution

Overall, the Zen 5 flagship offered only ~5% better performance than the Zen 4 flagship, with 10% lower power consumption but higher operating temperatures.

However, the story is different for Ryzen X3D chips. In Zen 5, the X3D L3 cache is second-generation and placed on the underside of the chiplet, unlike previous generations where it was on top of the CCD (causing heat buildup). This change allows for much better cooling, enabling Ryzen 9000 X3D chips to hit higher clocks and better thermal efficiency. Unlike the Ryzen 5000/7000 X3D chips, the new 9000 X3D models are fully unlocked for overclocking — a welcome change for enthusiasts.

This makes the Ryzen 9000 X3D lineup a superior choice compared to their non-X3D counterparts in almost every scenario — without the usual compromises X3D users had to deal with before.

TL;DR:​

Since Zen 2, each new generation — Zen 3, Zen 4, and Zen 5 — has been an incremental evolution:

✅ Ryzen 9 5950X: ~+20% over Ryzen 9 3950X
✅ Ryzen 9 7950X: ~+30% over Ryzen 9 5950X
✅ Ryzen 9 9950X: ~+5% over Ryzen 9 7950X
✅ Ryzen 9 9950X3D: ~+14% over Ryzen 9 7950X3D

For comparison, the Zen 2 flagship Ryzen 9 3950X was ~120% faster in multicore workloads compared to the previous flagship Ryzen 7 2700X (Zen+).

Innovation Rate: Is AMD Losing Motivation?​

With the last three Ryzen generations (5000, 7000, 9000), AMD — positioned comfortably without serious competition — didn’t feel the need to make dramatic improvements with every new Zen. But the competition is working tirelessly.

Intel’s Aggressive Plans to Strike Back

Despite financial struggles, Intel is working on new, more efficient architectures with a clear goal: to regain lost market share and be relevant again.

AMD is well aware of this and, knowing how hard it was to get to the top, won’t allow itself to be passive. Zen 6 will implement advanced technologies and solutions and will be manufactured on the most sophisticated process to date — all to counter and neutralize Intel’s comeback.

Zen 6: Detailed Overview of Architectural Improvements​

1. Larger CCX Modules – 12 Cores per Block


Zen 6 CCX modules will feature 12 cores instead of 8, with 48MB of L3 cache instead of 32MB. This will drastically improve gaming performance for non-X3D CPUs and bring them closer to X3D chips.



2. Interposer Technology – A Communication Revolution

​

The biggest change is abandoning traditional IF communication via socket pins and motherboard traces. Zen 6 will use an interposer, enabling direct, fast, and energy-efficient chiplet-to-chiplet communication, similar to monolithic designs.

Benefits of the interposer:
- Greatly reduced latency
- Dramatically increased bandwidth
- Much lower power consumption
- Much lower heat dissipation



3. Memory Support: Two Independent Controllers


Zen 6 replaces the dual-channel memory controller with two independent single-channel controllers, enabling:
- Support for much faster DDR5 EXPO 2.0 modules (8GHz+)
- Improved stability
- Greater flexibility and bandwidth
- Potential support for very high-frequency CU DIMM memory

However, 8GHz+ speeds will only be supported on EXPO 2.0-compatible motherboards — i.e., next-gen AM5 boards with AMD 900 chipsets (X970/B950). Current AM5 boards with 600/800 chipsets will be limited to 6400MHz, which will be the native Zen 6-supported speed.



4. Manufacturing Process: State of the Art


- CCD: TSMC 2nm (2NP)
- IOD: 2nd-gen TSMC 3nm (3NP)

Compared to Zen 5 (3nm CCD, 4nm IOD), Zen 6’s processes:
- Occupy much less area for the same transistor count
- Are far more power efficient
- Allow much higher frequencies

Rumors suggest a boost up to 7GHz, but even 6.5GHz would be a major step forward.



5. Improved Microarchitecture


Based on Zen 5 but with many optimizations, including:
- Improved front-end & instruction fetch
- Better branch prediction
- Deeper out-of-order execution
- 50% larger L1 instruction cache (48KB)
- Larger, optimized L2/L3 caches
- More Load/Store units
- Improved NPU for AI workloads

These improvements are expected to deliver ~10% higher IPC compared to Zen 5.

Expected Zen 6 Performance​

Single-core performance:
- +10% IPC
- +15% frequency
➡️ Total: 20–30% improvement

Multi-core performance:
- +50% more cores
- +10% IPC
- +15% higher all-core frequency
➡️ 50–90% improvement

​

​

Conclusion: Zen 6 – A Historic Turning Point​

​

With Zen 6, AMD will not only continue the evolution of Zen architecture but finally take a revolutionary step forward. Thanks to all the technological upgrades, architectural improvements, and cutting-edge manufacturing, Zen 6 is set to be AMD’s biggest generational leap since Zen 2.

No matter what Intel brings, Zen 6 will inevitably become a reference point in hardware discussions for years to come.

Once upon a time, no one believed a “dwarf” like AMD could beat the “giant” Intel. But the prey became the predator, and the predator — the prey. The new AMD became the old Intel, and the new Intel — the old AMD.

History teaches us that no one stays on top forever. But this fight — this constant switching of roles — is what drives the industry forward. As long as the duel between AMD and Intel continues, we, the end users, will be the biggest winners, getting new processors with solid performance improvements at better and better prices.