Intel Core i9-9980XE vs AMD Ryzen Threadripper

Today Intel is kicking off its newest High End Desktop processor, the Intel Core i9-9980XE Extreme Edition. This 14nm Skylake-X CPU boasts 18 Cores and 36 Threads and has an expected retail price of $1979. We compare the i9-9980XE to AMD's entire line of Threadripper CPUs to see where the 9980XE sits in the HEDT stack.

continued...

Multimedia Testing

Outside of gaming and encoding, there are few applications on the desktop that will push our systems to the limits, this especially becomes apparent when we start talking about multi-core processors that are now the norm. Some multi-thread aware encoding and content creation applications are starting to reach into available threads and truly utilize the processing power of these multi-core CPUs.

The benchmarks below all represent very real world situations just like you would run into at home while encoding video from your camcorder - or while using a picture or video editing program - or encoding a video for saving it to your hard drive or mobile device to allow you easier access to the content.

We have also included the synthetic Cinebench R15 and ray tracing benchmarks that should give you an idea about how 3D production programs will perform when rendering scenes. We have also added Cinebench R15 one thread performance metrics. The Blender application used below is the same file and version used by AMD when it previewed Ryzen last year, and we have added an actual scene from an animated movie feature We thought it would make for an interesting data point.

We have also added a Premiere Pro v13 "benchmark." This is actually not a benchmark at all, but rather an actual video that was produced with Premiere Pro v13. We simply timed the complete process to encode the video. The native video is 4K footage, and rendered out using the H.264 2160p 4K Youtube preset. Last but not least, we have updated our Handbrake benchmark to show a 4K video being encoded to 1080p.


Blender

Article Image

This is a very quick Blender scene render that does NOT heat-load our processors. So in terms of Turbo Boost and Precision Boost 2, it will show our CPUs in the best light. The 2990WX gives us a ~36% decrease in render time over the 9980XE, while the 9980XE comes in very close to our 2950X.

Article Image

Gooseberry uses Blender, the same as above except the scene rendered is from a real animated movie. As you can see our actual render time is about 10X as long so we do see the system heat-load. With our 2990WX heat-loading, it stops scaling as well as the 9980XE. Our decrease in render time is at ~8% now when looking at our two top CPUs, and the 9980XE shows an ~18% decrease in render time compared to the 2950X, when these were about tied above. Heat-loading these systems can make a big different in benchmarks.

POV-Ray

Article Image

POV-Ray is our open source ray tracing benchmark and utility and it does load all our cores and threads, however it does NOT heat-load our CPUs. Our scaling once again looks very much like it did above in our Blender Ryzen.

ChaosGroup V-Ray

Article Image

V-Ray is a professional ray tracing application used widely and it also fully loads our CPUs, however not heat-loading our CPUs. It returns nearly identical scaling as we saw with Pov-Ray ray tracing.

Adobe Premiere Pro CC

Article Image

Here again we are looking at a real-world workload as this a 10 minute 4K video encoded for Youtube. The original video was shot at 4K with a Canon XF405. This encode does in fact heat-load our CPUs. Do not think you are seeing incorrect data with Premiere Pro. This is the latest v13.0 version. We have seen "bad" results from our 2990WX CPU since its introduction, and we are seeing the same thing here with our 2970WX. Keep in mind both of those CPUs have four dies down on the substrate and two of those are not directly attached to memory. We have recently had communications with both AMD and Adobe about our workload and our results. We have now been able to make sure that our workload is not to blame for the 2990WX and 2970WX results lacking.

We have come to the conclusion that the 2990WX and 2970WX CPUs are "broken" when it comes to H.264 encode. We have another example below of H.264 having issues with the 4-die Threadripper processors.

Cinebench R15

Article Image

Single threaded Cinebench shows us that our 9980XE is boosting over our hand-overclocked 4.1GHz, and this is as it should be since Intel claims a 4.4GHz boost clock, or 4.5GHz boost clock, depending on what form of Turbo Boost you are running.

Article Image

Once we move to a fully threaded load we see our 2990WX once again distance itself in terms of performance over the 9980XE. We again see a ~39% increase in score, but this again is NOT a heat-loaded benchmark as it runs very quickly.

HandBrake

Article Image

We have changed up our Handbrake testing to give us a larger file size and be more realistic as to what you might be using it for at home. This is actually the 2.5GB 4K file that is 10 minutes in duration, encoded by Premiere Pro. We then are using the H.264 1080 Fast Handbrake preset to reduce our resolution and file size. Our 9980XE is on par with our 2950X and not incredibly faster than the 2920X.

We again see our 2990WX fall on its face with the 2970WX along with it. The 2950X gives us a ~33% decrease in encode over the 2990WX, and a ~23% decrease compared to the 2970X. This simply should not be happening.