May 14, 2025
Trending News

Zen 5 in depth, Ryzen 9000 performance and new 800 series chipsets

  • July 16, 2024
  • 0

AMD sent us new information about Zen 5, the architecture that powers the new Ryzen 9000 processors, and this time it went a little deeper in technical details.

Zen 5 in depth, Ryzen 9000 performance and new 800 series chipsets

AMD sent us new information about Zen 5, the architecture that powers the new Ryzen 9000 processors, and this time it went a little deeper in technical details. He also touched on a very important topic that raised a lot of doubts, namely the new chipsets that could be used to design the new motherboards, and he did not forget to confirm the launch date.

In this article I want to tell you many interesting things, but Let’s start with the Zen 5 architecture, because that’s the basis of the new Ryzen 9000 processors, and that’s why it’s the most important thing. Next, we will continue with the final specifications, performance data of these processors and the release date, and we will finish with the specifications of the new 800 series chipsets.

Zen 5

Note that the Zen 5 architecture will also be used in AMD’s new Ryzen AI 300 APUs, though these They have some peculiaritiesWhereas the:

  • They will combine Zen 5 and Zen 5c cores.
  • They will use a monolithic core design.
  • They will come with a new generation NPU (neural processing unit) with a performance of 50 TOP.

Zen 5 on Ryzen 9000, an in-depth look

Zen 5

CPU chips are in green, I/O chiplets in red.

The Zen 5 architecture used in the Ryzen 9000 processors uses the same chiplet design which we have already seen with the Ryzen 7000. This means that we have two types of chiplets that are clearly differentiated and combined in the same package:

  • CPU chiplet or CCD unit, which includes 8 cores and 16 threads, 32 MB L3 cache and 8 MB L2 cache. It is manufactured on TSMC’s 4nm node.
  • Chiplet I/O, which houses the entire input and output subsystem, memory controllers, and also the integrated GPU, which in this case is a Radeon RDNA 2 with 128 shaders. It is manufactured on TSMC’s 6nm node.

Both chips communicate via Infinity Fabric System, and in this generation it is possible to combine a maximum of two CCD units, which means that the top processor Ryzen 9 9950X has 16 cores and 32 threads.

The jump from 5 nm to 4 nm was made possible by the node in the CCD chiplet improve transistor density, performance and efficiency. The architectural changes that AMD introduced with Zen 5 are remarkable and affect all the key areas of the processor, covering both the front end, the execution core and the back end.

Let’s look at them in order from one end to the other:

  1. Hop predictor: less latency, more accuracy and more output capacity.
  2. Cache bandwidth and latency improvements.
  3. Dual decoding system.
  4. Support for AVX-512 instructions with a full 512 bits per cycle data path.
  5. Six pipelines, two of them FADD with two cycle latency.
  6. A floating point unit can work with a larger number of instructions.
  7. More bandwidth for processing more data thanks to 48 KB of 12-address L1 cache.
  8. Doubles the maximum bandwidth of the L1 cache and floating point units.
  9. Improvements in data prefetching.

As I said, there are important changes from Zen 4. For example, in the mentioned architecture, AVX-512 instruction support was provided by two 256-bit FPUs, and the L1 cache was only 32 KB and 8 addresses. L2 and L3 caches remain at 1MB per core and 32MB per chiplet, but latency and bandwidth have improved.

All these changes will make it possible an increase in CPI, which according to AMD averages around 16%, although it can achieve up to 35% improvement depending on the specific test we use, as we can see in the attached graph.

Zen 5

AMD also managed to reduce the thermal resistance by up to 15% in Zen 5, which allowed the temperature of the Ryzen 9000 to drop by up to 7 degrees at the same TDP level. This means that The Ryzen 9 9950X should be 7 degrees cooler than the Ryzen 9 7950X running at the same TDP of 170 watts.

Ryzen 9000 final specs, performance and launch

AMD already has the final characteristics of this new generation of processors perfectly outlined. According to the plan, we will see a total of four new models and their launch will take place on the day 31st of July. That was the date we’ve been pointing to in all our latest articles, so we can say the rumors have come true.

Here are the final Ryzen 9000 specs:

Ryzen 9 9950X: 16 cores and 32 threads at 4.3 GHz, turbo mode up to 5.7 GHz, 64 MB L3 cache, 16 MB L2 cache and 170 watt TDP. It has 80 MB of total cache (L2 + L3).

In terms of performance, this processor promises to be AMD’s most powerful yet, thanks to its 16-core, 32-thread configuration, the improvements it brings at the IPC level, and a turbo mode of up to 5.7GHz compared to the Ryzen 9 7950X we can expect performance improvement of up to 22%.

Ryzen 9 9900X, Ryzen 7 9700X and Ryzen 5 9600X They will offer 16%, 11% and 17% higher performance than equivalent Ryzen 7000 series models although they will have much lower consumption. This also translates into a significant improvement in efficiency as we have more power per watt consumed.

Ryzen 9 9900X: 12 cores and 24 threads at 4.4 GHz, turbo mode up to 5.6 GHz, 64 MB L3 cache, 12 MB L2 cache and 120 watt TDP. It has 76 MB of total cache (L2 + L3).

This processor promises to outperform the Intel Core i9-14900K up to 41% in professional and synthetic applications and up to 22% in games.

Ryzen 7 9700X: 8 cores and 16 threads at 4.5 GHz, turbo mode up to 5.5 GHz, 32 MB L3 cache, 8 MB L2 cache and 65 watt TDP. It has 40 MB of total cache (L2 + L3).

One of the star models for the middle class and one of the most promising in terms of raw performance. According to AMD, it will be up to 42% more powerful than the Intel Core i7-14700K in professional and synthetic applications and Up to 31% more powerful in games.

Ryzen 5 9600X: 6 cores and 12 threads at 4.7 GHz, turbo mode up to 5.4 GHz, 32 MB L3 cache, 6 MB L2 cache and 65 watt TDP. It has 38 MB of total cache (L2 + L3).

Little of the family, but not potentially, that’s for sure. This processor can outperform the Intel Core i5-14600K by up to 96% in professional and synthetic benchmarks and up to 29% in games.

With the new Ryzen 9000, AMD has particularly tried to improve consumption and operating temperatures, but it hasn’t forgotten about performance lovers either. Way “Precision Boost Overdrive” It allows us to overclock with a simple click to get the maximum possible performance from the processor, at the cost of increasing the TDP.

Zen 5 overclocking

The Ryzen 7 9700X deserves a special mention, which with that up to 15% performance improvement In gaming, it could end up almost on par with the Ryzen 7 7800X3D. This processor proved to be so good that it is able to beat the Ryzen 7 5800X3D in games with a TDP of only 65 watts.

The selling price has not yet been confirmed, but all indications are that they will be cheaper than the Ryzen 7000. If confirmed, AMD would be launching a very interesting and very competitive generation.

New 800 series chipsets and faster memory

Ryzen 9000 processors will continue to use the AM5 socket, but AMD is going to use this launch to introduce a total four new chipsetswith which it will cover both the enthusiast band and the high, middle and lower bands.

In the attached image you can see these new chipsets perfectly explained but summarized:

  • AMD
  • AMD
  • AMD B850: mid-range, with support for PCIe Gen4 on the GPU and optional PCIe Gen5 on the NVMe drives, USB 3.2 at 20 Gb/s, CPU and memory overclocking, and additional PCIe lanes for multi-GPU configurations.
  • AMD B840: low-end, with support for PCIe Gen3, USB 3.2 at 10 Gbps and the possibility of memory overclocking.

These new motherboards will not be available at launch. If you want to buy it, you’ll have to wait at the end of this summeror maybe until September. This isn’t a problem as eventually the Ryzen 9000 will run smoothly on any current 600 series chipset motherboard that has been updated to the latest AGESA version available.

On the other hand, AMD confirmed DDR5 memory support at a maximum frequency of 8,000 MHz, optimized profiles for easier and more stable overclocking and support for the JEDEC DDR5-5600 standard. The best price-performance ratio will continue to be DDR5 memory at 6000 MHz with CL30 latency.

Source: Muy Computer

Leave a Reply

Your email address will not be published. Required fields are marked *