

The new Threadripper 3970X tested: High end in any case
Having already tested the Ryzen 3950X, I'm now holding the Threadripper 3970X in my hands. With so much computing power, I can no longer stop marvelling.
A Threadripper processor looks really powerful: In contrast to the Ryzen 3950X, it feels like it weighs a tonne. A high-end desktop processor (HEDT) has its weight. I can't marvel at it for too long. The processor has to go on the test bench, I only have a few hours to test it before the review embargo falls.

The chip in detail
Each Threadripper processor consists of 4 Zen 2 dies that are connected to an I/O die. This serves as the central hub of the processor. The part houses a total of 23.54 billion transistors.
The AMD Ryzen Threadripper 3970X is the most powerful high-end consumer desktop processor in 2019. It has 32 cores and 64 threads as well as 128 MB L3 cache. The processor is clocked at 3.7 GHz base and 4.5 GHz boost. 88 Gen 4 PCIe lanes are available and the TDP is 280 watts. The third generation Threadripper now has the sTRX4 socket, which requires a new TRX40 mainboard. Asus provided me with the ROG Zenith II Extreme for the review. According to the CPU-Z tool, it runs AGESA Castle Peak 1.0.0.2.
.
Test methodology
I base my test methodology on that of our graphics cards. I therefore carry out the same tests for the most part. For the Blender benchmark, however, I run the CPU benchmark, which only uses the CPU for rendering. I also run the Cinebench R20, the AIDA 64 stress test and transcode a 1080p film with Handbrake. I'm leaving out the VR benchmarks because this is where the graphics card comes into play.
The processor is tested on our DimasTech Easy V3.0 benchtable with the following components:
As I currently lack a basis for comparison - I have only just started benching processors and graphics cards - I am listing the results of this test and only comparing them with those of the 3950X. However, this comparison should be treated with caution. On the one hand, I have 64 GB of RAM on the testbench with the Threadripper, as opposed to the 16 GB on the graphics card testbench. On the other hand, some of you have questioned my results on the 3950X review. I am now doing night tests and will then deliver the results, of course also with 64 GB RAM.
Synthetic benchmarks and temperatures
Cinebench R20
With Cinebench from Maxon, you can test how your PC or processor performs when rendering Cinema 4D content. Here are the results of the 3970X compared to the 3950X.


In the single core, the 3970X does not come close to the 3950X. This is to be expected with the lower clock rate. However, it achieves a multi-core score of 16,596 points, which I have never seen before. It's a real pleasure to watch the render animation of Cinebench R20 in Multi Core. I usually do this with lame processors on notebooks.
Fire Strike and Time Spy
Here are the results of the Time Spy and Fire Strike benchmarks.

And here in detail:
Benchmark | Overall Score
3970X / 3950X | Physics / CPU Score
3970X / 3950X | Combined Score
3970X / 3950X |
---|---|---|---|
Fire Strike
(1080p, DirectX 11) | 22 815 / 22 467 | 25 197; 79.99 FPS / 31 526; 100.08 FPS | 15 272; 71.04 FPS / 12 549; 58.37 FPS |
Fire Strike Ultra
(2160p, DirectX 11) | 6596 / 6821 | 25 330; 80.42 FPS / 31 424; 99.76 FPS | 3619; 16.84 FPS 3676; 17.1 FPS |
Time Spy
(1440p, DirectX 12) | 11 142 / 11 556 | 11 598; 38.97 FPS / 11 435; 38.42 FPS | n/a |
Time Spy Extreme
(2160p, DirectX 12) | 5801 / 5590 | 15 537; 22.5 ms frametime / 8045; 43.5 ms frametime | n/a |
The 3970X performs about the same as the 3950X. Most of the differences are small, which can be attributed to the benchmark error rate. What stands out: In Fire Strike, the 3950X achieves the better result in the CPU-heavy calculations. In Time Spy it is exactly the opposite. At a clock rate of just over 4.2 GHz, the 3970X reached a maximum temperature of 67° Celsius during the benchmark.
To get an even better picture of the temperatures, I run the CPU stress test from AIDA64. I activate the options Stress CPU, Stress FPU, Stress Cache and Stress System Memory. I let the test run for two hours. During this time, I have at least 3.95 GHz on all cores and an average temperature of 62.5° Celsius. The temperature curve is flat. After just a quarter of an hour, I reach around 63° Celsius. The temperature does not rise any further.
Puget Systems Photoshop benchmark
The Photoshop benchmark uses the following reference workstation as the basis for calculating the scores:
- Intel Core i9 9900K 8 Core
- NVIDIA GeForce RTX 2080 8 GB
- 64 GB RAM
- Samsung 960 Pro 1 TB
The results of the reference workstation can be used to estimate how well other systems perform. The two processors achieve the following results

And here in detail:
Scores | Threadripper 3970X | Ryzen 9 3950X | Reference workstation |
---|---|---|---|
Overall Score | 1043.4 | 953.4 | 1000 |
General Score | 106.5 | 95.7 | 100 |
Filter Score | 101.9 | 92.5 | 100 |
Photomerge Score | 104.9 | 100.3 | 100 |
GPU Score | 108.5 | 94.5 | 100 |
The Threadripper achieves around 8 per cent more points than the Ryzen 3950X. Interestingly, the difference in the GPU score is large. It seems as if the 3970X works better with the graphics card than the 3950X.
Puget Systems Premiere Benchmark
Here, our test benchmark does not compete against a reference workstation. In the Puget Systems Premiere benchmark, the score is calculated relative to the frame rate of the test videos. If the test video has an FPS of 29.97 and the system renders it at 29.97 FPS, the score is 100 points. If it is only 14.98 FPS, it is only 50.
The benchmark runs media in the formats 4K H.264 with 150 Mbps in 8 bit (59.94 FPS), 4K ProRes 422 16 bit (59.94 FPS) and 4K RED (59.94 FPS). He tested live playback in Adobe Premiere Pro and the export. A value of 100 is the maximum for live playback, as Premiere cannot play back the media faster than specified. For export, on the other hand, over 100 points are feasible, as rendering is not limited to the FPS of the media.
In addition, effects are added to ten ProRes 422 clips, which place a heavy load on the graphics card. Puget Systems calls this "4K Heavy GPU Effects". These clips are then played back in Premiere and exported. The same happens with Heavy CPU Effects, with effects that place a heavy load on the CPU. The CPU value is particularly relevant for the Ryzen comparison. I have nevertheless listed the GPU values for the sake of completeness.
Here are the results of the 3970X:

And here in detail:
Test | Threadripper 3970X | Ryzen 9 3950X |
---|---|---|
Overall Score | 956 | 695.5 |
Live playback score | 80.6 | 71 |
Export Score | 110.6 | 68.1 |
4K H.264 with 150 Mbps in 8 bit (59.94 FPS) | 81 Live Playback Score
171 Export Score | 55 Live Playback Score
99 Export Score |
4K ProRes 422 16 bit (59.94 FPS) | 100 Live Playback Score
175 Export Score | 99 Live Playback Score
100 Export Score |
4K RED (59.94 FPS) | 82 Live Playback Score
121 Export Score | 66 Live Playback Score
85 Export Score |
4K Heavy GPU Effects | 78 Live Playback Score
50 Export Score | 79 Live Playback Score
48 Export Score |
4K Heavy CPU Effects | 62 Live Playback Score
36 Export Score | 55 Live Playback Score
20 Export Score |
The Threadripper virtually destroys the 3950X in the Premiere benchmark. Around 27 percentage points more is enormous. And this is despite the fact that the graphics card in combination with the 3970X is left behind compared to the 3950X.
Puget Systems after-effects benchmark
In the Puget Systems After Effects benchmark, the benchmark scores are structured similarly to Photoshop. The following reference workstation serves as the basis for calculating the scores:
- Intel Core i9 9900K
- 128 GB RAM
- NVIDIA GeForce RTX 2080 8GB
The results of the reference workstation can be used to estimate how well other systems perform. Our test benchmark achieves the following results:

And here in detail:
Scores | Threadripper 3970X | Ryzen 9 3950X | Reference workstation |
---|---|---|---|
Overall Score | 1070 | 1007 | 1000 |
Render Score | 113.5 | 103.1 | 100 |
Preview Score | 102.6 | 97.4 | 100 |
Tracking score | 104.9 | 101.6 | 100 |
The 3950X has already achieved brilliant results in the After Effects benchmark. At around 6 per cent, the Threadripper 3970X does even better.
Puget Systems Resolve benchmark
The scores of the Resolve benchmark from Puget Systems are also based on a reference workstation. The scores are determined by encoding in 4K. The reference workstation is based on the following components:
- Intel Core i9 9900K
- A minimum of 32 GB RAM (not mentioned by Puget Systems)
- NVIDIA Titan RTX 24 GB
The 3970X achieves the following results in the 4K benchmark
.

And here in detail:
Scores | Threadripper 3970X | Ryzen 9 3950X | Reference workstation |
---|---|---|---|
4K Average Results Overall Score | 1163 | 932 | 1000 |
4K H264 150 Mbps 8 bit Codec Average Score | 119.7 | 93.2 | 100 |
4K Cinema Raw Light | 109.7 | 95.7 | 100 |
4K ProRes 422 | 121.3 | 90.9 | 100 |
4K ProRes 4444 | 128.9 | 90.2 | 100 |
4K RED | 101.6 | 95.9 | 100 |
In Resolve, the difference to the Ryzen 3950X is striking. The Threadripper 3970X delivers around 23 per cent more performance. With this result, it also beats the reference workstation by far.
Handbrake
To compare the performance of the Ryzen 9 3900X and 3950X, I transcode the film "Drive". The source file is a 19 GB H.264 file. I select the "HQ 1080p30 Surround" preset in Handbrake and transcode the film.

The 3970X takes 20 minutes and 45 seconds to transcode. That's five minutes and 48 seconds less than the Ryzen 3950X needs.
Blender
The Blender benchmark renders six different scenes: Barbershop Interior, BMW27, Classroom, Fishy Cat, Koro and Pavilion Barcelona. The benchmark tests either the CPU or GPU. For the Ryzen comparison, I run the CPU benchmark. The times required per scene are recorded and totalled. Here are the results.

And here in detail:
**Rendered scene | Time required Threadripper 3970X | Required time Ryzen 9 3950X |
---|---|---|
Total | 17 minutes 44 seconds | 31 minutes 25 seconds |
Barbershop Interior | 5 minutes 57 seconds | 10 minutes 45 seconds |
BMW 27 | 1 minute 3 seconds | 2 minutes 1 second |
Classroom | 3 minutes 25 seconds | 6 minutes 39 seconds |
Fishy Cat | 1 minute 34 seconds | 2 minutes 59 seconds |
Koro | 2 minutes 19 seconds | 4 minutes 10 seconds |
Pavilion Barcelona | 2 minutes 27 seconds | 4 minutes 50 seconds |
What a difference: The 3970X is over 13 minutes faster than the 3950X. In Blender, the many cores are very noticeable. The Threadripper is even faster than the Sapphire Radeon RX 5700 XT Nitro+: It takes 24 minutes and 27 seconds to render the six scenes.
Games
Last but not least, our four game benchmarks. I don't yet have any values for "Deus Ex: Mankind Divided" on the 3900X. I'll provide those in the night tests. But that's it: The results of the 3950X should be taken with a grain of salt. In addition to the average, minimum and maximum FPS, I also provide you with the average frametime. As I now determine FPS and frametime with FRAPS, the whole thing is shown slightly differently.

The results in detail:
Games | Threadripper 3970X
FPS | Ryzen 9 3950X
FPS |
---|---|---|
"Deus Ex: Mankind Divided" (DX11)
FPS 1080p, highest presets | min. 29
average 116,657 max. 156 Frametime: 8.57 ms | n/a |
"Deus Ex: Mankind Divided" (DX11)
FPS 1440, highest presets | min. 29
average 84,453 max. 108 Frametime: 11.8 ms | n/a |
"Deus Ex: Mankind Divided" (DX11)
FPS 2160, highest presets | min. 27
average 47,614 max. 60 Frametime: 21 ms | n/a |
"Control" (DX11)
FPS 1080p, highest presets | min. 77
average 93,003 max. 114 Frametime: 10.7 ms | min. 59.1
average 113 max. 242.8 Frametime: 7-9 ms |
"Control (DX11)
FPS 1440p, highest presets | min. 53
average 63,795 max. 74 Frametime: 15.7 ms | min. 53.1
average 59.5 max. 60.6 Frametime: 16-18 ms |
"Control" (DX11)
FPS 2160p, highest presets | min. 26
average 32,031 max. 39 Frametime: 31.2 ms | min. 21.7
average 49.8 max. 242.7 Frametime: 31-34 ms |
"Control (DX12)
FPS 1080p, highest presets, raytracing maximum | min. 44
average 52,563 max. 66 Frametime: 19 ms | min. 31.3
average 51.7 max. 61.4 Frametime: 18-20 ms |
"Control (DX12)
FPS 1440p, highest presets, raytracing maximum | min. 27
average 35,888 max. 96 Frametime: 27.8 ms | min. 28.8
average 37.9 max. 60.1 Frametime: 28-30 ms |
"Control" (DX12)
FPS 2160p, highest presets, ray tracing maximum | min. 15
average 18,993 max. 25 Frametime: 52.6 ms | min. 16
average 19 max. 21.9 Frametime: 50-60 ms |
"Shadow of the Tomb Raider" (DX12)
FPS 1080p, highest presets | min. 2
average 134,192 max. 184 Frametime: 7.45 ms | min. 70.7
average 119.1 max. 186.4 Frametime: 6-8 ms |
"Shadow of Tomb Raider (DX12)
FPS 1440p, highest presets | min. 2
average 99,592 max. 344 Frametime: 10 ms | min. 67.5
average 99.1 max. 164.2 Frametime: 9-11 ms |
"Shadow of Tomb Raider" (DX12)
FPS 2160p, highest presets | min. 2
average 52,049 max. 140 Frametime: 19.2 ms | min. 24.9
average 54.4 max. 167.3 Frametime: 18-20 ms |
"Strange Brigade" (Vulkan)
FPS 1080p, highest presets | min. 155.3
average 216.5 max. 312.5 Frametime: 4.62 ms | min. 193.8
average 225.3 max. 267.9 Frametime: 6-8 ms |
"Strange Brigade" (Vulkan)
FPS 1440p, highest presets | min. 120
average 159.4 max. 217 Frametime: 6,274 ms | min. 165.7
average 182.2 max. 229 Frametime: 6-8 ms |
"Strange Brigade (Vulkan)
FPS 2160p, highest presets | min. 42.5
average 91.3 max. 269.6 Frametime: 10.952 ms | min. 79.2
average 91.6 max. 112.7 Frametime: 10-12 ms |
Next time, the frametime data will follow in graphical form.
The Threadripper 3970X and the Ryzen 3950X go head-to-head in the games. Sometimes the Threadripper is faster, sometimes the Ryzen. But the difference is never big.
Conclusion: Many cores for a lot of performance
The Threadripper 3970X is a real beast. It not only delivers plenty of performance in heavy-core applications, but also in games. Sure, just under 2200 francs for the processor is a proud price. But a HEDT processor costs money. If you have the money and need a lot of cores - and decent speed in single-core mode too - the Threadripper 3970X is definitely the right choice.
I'm now going to start testing the Ryzen 3950X and then compare the processors I've tested so far - 3900X, 3950X and 3970X - again.
30 people like this article


From big data to big brother, Cyborgs to Sci-Fi. All aspects of technology and society fascinate me.