AMD Ryzen 9 3900X in the test showed incredible performance

Conclusion of Test:

AMD’s Ryzen 9 3900X turns out to be a wonder CPU in the test. The twelve-core processor beats direct competition in many flying flag tests, is efficient and, at the same time, only slightly more expensive. This is also Intel’s last stronghold, the consumer high-end, fell. Whether you are a gamer or a high-end user, there is little reason not to resort to the 3900X.

Advantages

• Strong single and multi-core performance
• High efficiency
• Compatible with old motherboards

Disadvantage

• No integrated graphics unit
• A bit expensive

Test scores (compared to all tested products in this category)

Maximum power dissipation (TDP): 105 watts
Processor Clock: 3.80 GHz
Benchmark: x264: 150.26 fps
Number of CPU cores: 12
Base type: Socket AM4
Number of threads: 24
Level 2 cache: 12x 512 kByte
Level 3 cache: 65,536 kByte
Core Code Name: Matisse
Manufacturing process: 7 nm
Product: AMD Ryzen 9 3900X
Benchmark: PovRay, 1.280×1024, no AA: 6,174 pixels / s
Benchmark: TrueCrypt 7.2 AES Twofish Serpent: 979 MB / s
Maximum processor clock: 4.60 GHz
Integrated graphics unit: no onboard GPU
CPU performance: 1.4
Benchmark: HandBrake: 206.9 fps
Benchmark: Cinebench R15, max. Num. CPUs: 3,130 points
Graphic Benchmark: Bioshock Infinite: 0.00 fps
Benchmark: WinRAR: 27,823 MB / s
Benchmark: 3DMark Time Spy with GTX 1080: 8,143 points
GPU performance: 7.5
Benchmark: Excel 2016 – Monte Carlo Simulation: 0.45 seconds
Benchmark: PCMark 8: 4,153 points
Benchmark: 3DMark Firestrike with GTX 1080: 20,220 points
Graphics Benchmark: 3DMark Time Spy: 0 points
Graphic Benchmark: Metro Last Light: 0.00 fps
Graphics Benchmark: 3DMark Cloud Gate: 0 points
Graphics Benchmark: 3DMark Firestrike: 0 points

AMD Ryzen 9 3900X in Test: 

Performance

It’s moments like these when we look at test results with extremely mixed feelings. On the one hand, our eyes are glowing when we can experience a revolution live: The AMD Ryzen 9 3900X beats the direct competitor, the Intel i9-9900K, in our benchmarks by an average of 21 percent. And it is barely more expensive. On the other hand, the whole time the thought follows us: “That can not really be.”

Well, it can be and it is – we re-tested the i9-9900K with the latest updates, and in the end it has to be the short straw. We want to try to explain how AMD did that. But first a few words about the basic structure of the new Zen-2 microarchitecture and the benchmark results.

Turn two into Three

With Zen 2 aka “Matisse”, AMD says goodbye to the old Zeppelin-Die structure on the chip and splits the tasks into several parts: There are three components on the silicon of the R9 3900X. Two of these are so-called chiplets. Here live the cores of the Ryzen – maximum eight per chiplet, divided into four-cluster. The CPU-near cache is also located in the chiplets. Both chiplets communicate with the IO-Die via the “Infinity Fabric” data bus (“IO” stands for “in / out”). This in turn takes care of the data transfer to the rest of the PC, the memory management and also conveys information between the chiplets.

RYZEN 9 3900X VS CORE I9-9900K (BENCHMARK RESULTS)

Ryzen 9 is not Heavy on Power

A typical way to get more power out of the CPU is to increase power consumption. In our measurements with a 250-watt fan (TDP), however, there is hardly any difference between the top CPUs from AMD and Intel. In PCMark 10, the system power consumption comes to 234 and 350 watts depending on the test scenario. The Intel system, however, comes to marginally lower 233 and 348 watts. Even if one considers the different motherboards and their possibly different power consumption, the differences between the processors are negligible. So efficiency has not saved AMD.

The Secret is in the IPC

There is a big difference between AMD and Intel: clock speed. While Intel has cracked the 5 GHz, the Ryzen Boost only manages to reach 4.6 GHz. The stronger performance can only come from a monstrously improved IPC (instructions per cycle). AMD names here a few changes that in their entirety may contribute to the additional 15 percent IPC that the manufacturer specifies compared to the previous generation.

The most obvious is the enlarged L3 cache. 64 MB of CPU-near memory is now available. The improved AVX2 support is also exciting – the CPU now processes the data twice as fast. Furthermore, the chip improves branch prediction of instructions, gets a larger micro-op cache, and a more associative L1 cache.

A little more vivid are the last two improvements: on the one hand the so-called thread grouping. Processor threads, ie tasks of the executing programs, end up in Zen 2 in the same chiplet and there rather in the same computing cluster instead of at other ends of the processor. This should be a better solution especially for the spatially separated chiplets.

Subject Memory:

AMD has given the infinity fabric, so the CPU data bus, the clock more freedom. That should remove an old bottleneck – but there is, according to AMD, a “sweet spot” at DDR4-3733. If you want to save a bit of money without significant performance losses, should grab to DDR4-3600 (CL16). Unfortunately, we have not been able to test how different data rates affect performance.

Ultimately, you should not forget one thing in the CPU performance: AMD dispenses with an integrated graphics unit in the higher-end desktop processors. If Intel omitted these, more space would be available for CPU tasks. Where an integrated graphics unit can bring some significant advantages in benchmarks.