Core i9-9900K Encoding and Video Editing Performance
Moving on to video editing, we’ll kick this off with the free and popular HandBrake test. In this newer test, we use the latest version of HandBrake to convert the 4K open-source Tears of Steel video to 1080p and 30 fps, using the Matroska H.265 profile
We’re not entirely sure why, but it’s clear the Ryzen 7 2700X suffers using the H.265 profile. The Core i9 is about 23 percent faster than the Ryzen 7 2700X, and the 6-core Core i7-8700K actually edges past the AMD CPU as well.
But just to show that sometimes it’s the profile in HandBrake, we also encoded the file using the 1080p Apple preset. This preset is obviously far easier on the Ryzen, which closes the gap to about 7 percent with the Core i9.
Our next video encoding test uses the new Adobe Premiere Pro CC 2019 NLE to export a short video shot on a 4K Sony Alpha camera. It’s an actual video produced by our video team, so it’s about as real-world as you can get. For our first test, we export the file using the H.265 preset and switch on the maximum render option.
We also run the encode using the host CPU rather than the GPU. Video nerds, as we’ve been told, still believe CPU encodes produce the best video.
For this particular test, to ensure storage is not the bottleneck, we read and write to the same Plextor PCIe SSD, which was moved from system to system for the test.
As we saw with HandBrake in H.265 encodes, the Core i9 has a hefty advantage over the Ryzen 7 2700X. The 6-core Core i7 even runs about even with the 8-core Ryzen 7.
Why H.265 is such a problem in HandBrake and Premiere isn’t clear to us, but if that’s what you need to run, you’ll likely want Intel.
As we want to be fair, we did run our Blu-ray export. It’s basically the same project, but we export it using the Blu-ray preset in Premiere and check maximum render quality. In this H.264 encode, Ryzen is far more competitive than the Core i9 chip.
More information is better than less, so our last Premiere test switches the encode from CPU to GPU. Typically more cores still help, but it appears it doesn’t matter whether it’s Intel or AMD from our results.
Core i9-9900K Compression Performance
Moving on to compression performance, we used WinRAR 5.60’s built-in benchmark to measure how well each CPU fares. The result here is all smiles for the Intel CPU. It simply devastates the Core i7-8700K and blows the Ryzen 7 2700X out of the water.
The performance of the 8-core Ryzen 7 is no surprise though. We know from previous comparisons that WinRAR just doesn’t like the Ryzen microarchitectue. We’ve guessed it could have something to do with the fabric designs of the Zen, as WinRAR also performs poorly on Intel’s Core X chips that use a fabric-like mesh architecture.
WinRAR costs money, though, so many prefer the free 7-Zip. Besides being free, it outperforms the built-in Windows compression performance so emphatically, we can download 7-Zip, install it, and decompress a benchmark faster than using the built-in tool.
The free app has a built-in benchmark, which we use to measure both compression and decompression performance. It’s a multi-threaded test, and compression performance is mostly reliant on integer performance among other things. The Core i9 again rules the day, blowing both the Core i7 and the Ryzen 7 out of the water to the turn of almost 30 percent. In the case of the Core i7 with just 6 cores, this isn’t a surprise. What we can’t figure out is why the Ryzen 7 gets lumped so badly here.
7-Zip’s decompression performance is far more reliant on memory bandwidth, though, and Ryzen 7 closes up the gap nicely. The Core i9’s winning margin is still too close for comfort for a $488 CPU fighting a $329 rival.
Core i9-9900K Encryption Performance
Our last application performance test uses VeraCrypt to gauge AES decryption performance. The Core i9 again shines, while the 8-core Ryzen 7 is left slumming with the 6-core Core i7-8700K.
”World’s best gaming CPU?”
Hopefully, even ardent AMD fans would have to admit the results in applications put the Core i9-9900K at the front of the line for mainstream CPUs. But where Intel really planted its flag is in gaming.
For our tests, we tried to keep it fair by using some games AMD has already vetted to be OK on its CPUs. We also try to represent both sides of the resolution argument in today’s world. Some, for example, argue that 1080p should only be used for CPU testing, and at low settings, to remove the GPU as the bottle neck. Others argue that higher resolutions and game settings should be exclusively used to represent results that would actually be used.
We find validity in both sides. While we “only” used GeForce GTX 1080 cards for our testing, we’re working on procuring the current speed champ in GPUs and will rerun these tests to see if using a much faster GPU will give an advantage to the faster CPU.
Deus Ex: Mankind United Performance
First up is Deus Ex: Mankind United. Running this game at 1920x1080 (1080p) on its Ultra setting in DX12 mode, we can see that for the most part, it’s a wash between the two 8-core chips. This made us wonder whether our GeForce GTX 1080 card was holding us back.
Moving the game quality sliders to Low in Deus Ex lets us take the GPU out of the equation. The high clocks of both Intel CPUs take over, and we see that familiar “Ryzen 20- to 25-percent gaming deficit” appear.
Ashes of the Singularity: Escalation Performance
Moving on Ashes of the Singularity: Escalation, we decided to skip the high-quality test because Ashes is the game to measure DirectX12 CPU performance. For our test, we selected low quality to take the GPU out of the equation, and chose the CPU Focused benchmark which uses additional physics and units to push the CPU’s capability. No surprise: The Core i9 is a healthy 19 percent faster than the Ryzen 7 2700X.
F1 2016 Performance
Our next game is Code Master’s F1 2016. Let’s jump straight to the meat and see what happens when the workload is not GPU-bound. No surprise, the higher-clocked Intel CPUs are in front, with the Core i9 coming in 27 percent faster than the Ryzen 7 2700X.
Because we want to temper expectations, we also run F1 2016 at 2560x1600 resolution on Ultra. The results are exactly what you expect: the same. When it’s a GPU-limited game, we should expect this if the CPU is decently faster. We’ll definitely test out this theory with GeForce RTX 2080 Ti cards, which are a good clip faster than GeForce GTX 1080 cards.
Hitman Absolution Performance
Next up is Hitman Absolution which, honestly, doesn’t like Ryzen all that much. Running at 1920x1080 on Ultra, the Core i9-9900K is a crazy 37 percent faster. We’d guess this game is all about clock speeds, too, as the Core i7-8700K represents well.
Next we ran the game at 2560x1600 to see if we could make this more of GPU load, but Core i9 is still far in front. Some of this could be the same lack of CPU optimization that has vexed Ryzen since its launch, but some the monstrous 5GHz clock of the Core i9 is another likely factor.
Shadow of the Tomb Raider and Rise of the Tomb Raider Performance
We’ll close out game results with Lara Croft. First, we’ll run Rise of the Tomb Raider at 19x10 in DX12 mode. Even at this resolution, RoTR seems to be limited purely by GPU performance. We’ll be interested to see whether dropping in a GeForce RTX 2080 Ti will change this equation, but that’s for later.
Setting RoTR to a lower visual quality setting, we wanted to see what the results are once the GPU isn’t as much of a hindrance.
Rise of the Tomb Raider is a bit elderly at this point, so we moved on to the new Shadow of the Tomb Raider. This title is expected to roll in more multi-core support in its DirectX12 mode, and it doesn’t disappoint. The built-in benchmark reports overall frame rate as well as a CPU Average Frame Rate. The latter refers to performance when the GPU isn’t a factor.
Set to 1080p and Highest visual quality, we see the same puzzling pattern we saw with Deus Ex: The 8-core chips outperform the 6-core Core i7-8700K by a decent margin.
Once you lower the visual quality of the game, the Core i7-8700K jumps ahead of the Ryzen 7 2700X. And yes, the Core i9-9900K is simply cooking.
To close this out, we also ran the game on the Core i9 with MCE in Auto and On. This basically takes it from being a 4.3GHz all-core CPU, to a 4.7GHz all-core (MCE Auto) and 5GHz (MCE On) CPU. As you can see, the higher the clock speeds, the higher the CPU average result in the game.
Cinebench multi-thread benchmarking reveals gory details if you keep reading.