What you need to know about Zen 3 and Ryzen 5000 series Processors

 

What is IPC? 

IPC stands for Instructions Per Cycle, and before we talk about Zen 3’s architectural enhancements, we need to discuss what IPC gain are and what they aren’t. 

When AMD claims that Zen 3 offers 19% IPC gains, that doesn’t mean that Zen 3 will offer a 19% performance gain in all CPU-limited applications. When discussing IPC changes, it must be noted that manufacturer-provided IPC gains are calculated and averaged out across a variety of workloads. This means that not all CPU-limited applications will see 19% performance benefits, with some seeing larger gains while others see shallower performance enhancements. 

If an application benefits heavily from specific architectural change, performance gains of more than 19% will be possible with Zen 3, while some other applications will see smaller performance benefits. Different applications will react to Zen 3’s design changes in different ways, which means that Zen 3 will not automatically deliver 19% performance gains under all workloads. 

AMD’s 19% IPC gains are calculated as an average across a range of 25 workloads, setting consumer expectations for what enthusiasts should expect from Zen 3. 

Zen 3 – It’s all about the Cache
 
With Zen 3, one of AMD’s largest innovations is their redesigned cache topology and core layout, which sees Zen 3 core complexes (CCXs) which are comprised of eight cores. With Zen 2, each eight-core chiplet was constructed using two quad-core complexes, with communications between these two core clusters receiving inter-CCX latencies.     

The benefits of AMD redesigning Zen’s core complexes are twofold. First, AMD has enabled faster inter-core communications through the elimination of inter-CCX latencies when communicating with cores in the same chiplet. The second is that each CCX’s 16MB of L3 Cache can now be combined into a larger unified L3 cache. 

 

With Zen 3, each CPU core has direct access to 32MB of L3 cache. With Zen 2, each core on a module had direct access to 16MB of L3 cache and access to other caches with an inter-CCX latency penalty. These changes allow Zen 3 to offer reduced average L3 cache latencies, allowing Zen 3 cores to less time waiting and more time working. Think of decreased latencies as lowered waiting times; the less time your CPU is waiting, the more time it can spend working. 

Some workloads are more latency-sensitive than others, with gaming being especially sensitive to cache latencies. AMD’s new cache layouts are a primary cause of Zen 3’s performance enhancements in gaming workloads, especially for titles like CS:GO.  

AMD’s 8-core CCXs also eliminate other inter-core communication latencies, which will be beneficial for some multi-threaded workloads. 

  

 
End-to-End architectural enhancements

AMD’s core topology redesign isn’t the cause of all of Zen 3’s performance enhancements. With Zen 3, AMD claims that they have enhanced their Branch Predictor, Execution Engines, Front End and Cache Prefetching mechanisms. 

Zen 3 won’t just see performance benefits when cache latencies are involved. Zen 3 will see performance gains across a wide range of workloads. You don’t get a averaged 19% IPC increase for fixing one problem, which is great news for Ryzen fans. 

With so many changes coming with Zen 3, there is a reason why AMD calls Zen 3 an all-new processor design. The fundamentals of Zen 2 have been altered, and the results of these changes are mindblowing. A 19% IPC gain is no small feat, especially without the benefit of process node advancements. AMD’s engineers should be proud of their work. 

Another major aspect of Zen 3 is the fact that the architecture doesn’t consume more power than Zen 2 to achieve its 19% IPC improvement. AMD has not just made its latest Zen processors more powerful; they are also more power-efficient. AMD has claimed that their Zen 3 architecture is 20% more power-efficient than Zen 2 and that their Zen 3 products are up to 2.8x as efficient as Intel’s i9-10900K. 

Specifications – It’s All About the Cash!

Cinebench – Single-Threaded Performance Leadership

With Zen 3, AMD has claimed performance leadership over Intel within the desktop market, highlighting the performance of their Ryzen 9 5900X in Cinebench. Where Intel’s i9-10900K offered a score of 544, AMD’s processor offered a score of 631. That represents a 15.99% performance advantage in single-threaded performance. 

With AMD’s Ryzen 9 5950X, AMD’s single-threaded performance rose further to 640 points, representing a 17.6% performance gain over Intel’s i9-10900K. AMD offers consumers more cores and more single-threaded performance in the desktop market. Intel should be worried. 

Gaming – Zen 3 claims the performance crown

When compared to the Ryzen 9 3900XT, AMD has claimed that their Ryzen 9 5900X can deliver a 26% performance boost in CPU-limited games on average, showcasing a 5-50% performance boost in the performance charts below. 

AMD’s architectural enhancements for Zen 3 deliver significant performance gains over today’s Ryzen 3000 series processors. These gains are enough to surpass Intel’s i9-10900K, assuming you believe AMD’s data. 

  
 

In summary

AMD has made some big promised with Zen 3, from claiming gaming leadership and single-threaded performance leadership to claims that they offer better value than Intel in seemingly all areas of the desktop PC market. 

Zen 3/Ryzen 5000 series processors will be released on November 5th. Expect to see a review from us at or before this date. 

You can join the discussion on AMD’s Ryzen 5000 series of processors on the OC3D Forums.