Hey folks,

So recently I decided it's time for a CPU upgrade in my gaming rig, replacing my aging AMD FX 8350. It's now going to the spare parts bin in case I need to replace components for my main rig / streaming rig (which the FX does quite well).

I was mainly interested in a better VR experience (more on that later), but also to make sure I had enough room for the near future, and to improve some games that may have been bottle-necking with this CPU. See the list of components I changed below.

Specifications

Old specs:

CPU: AMD FX-8350 Unlocked
CPU Cooler: Noctua NH-C14S
RAM: 16 GB 1333 MHZ KINGSTON
Mainboard: ASRock 970M Pro3
PSU: RHOMBUTECH 800 W

New specs:

CPU: AMD Ryzen 1800X
CPU Cooler: Arctic Liquid Freezer 120
RAM: 16 GB 3200 MHZ Patriot Viper 4
Mainboard: ASRock Fatal1ty X370 Gaming-ITX/ac
PSU: Corsair RXḾ 750x 750 W

GPU remained the same, and is a MSI AMD Radeon RX 480 8 GB (reference card), running on AMDGPU Mesa 17.3.3.

Build stream on Twitch

For the interested, you can check the build stream here: (Two parts because the houses's power breakers tripped at one point :) )

https://www.twitch.tv/videos/224577204
https://www.twitch.tv/videos/224596517
(Headphone users warning: Due to my wireless mic disconnecting two times that stream, there will be loud noises in the break screen (just skip it), and a few seconds towards the end!)

Benchmarks

I benchmarked 3 games before and after (unfortunately, I couldn't get Unigine Superposition to run live on stream, so I only have a comparison to a while back): F1 2017 (Vulkan), Mad Max (OpenGL) and Tomb Raider (OpenGL).

F1 2017

Before (high settings):


After (same settings):


Settings now moved up to maximum:


This game was particularly interesting, since I now could move up to maximum settings with Vulkan, and still hit Vsync (60 Hz) more or less fine. On high settings, we have an increase of 87% on the minimum FPS value! Average FPS also experienced an increase of about 40%! Looking at maximum FPS, we have an increase of 17%, but keep in mind I used Vsync to 60 Hz here (since I play with Vsync, there is no point for me testing above this).

We can also see that the frame times have improved somewhat too, specially the maximum frame time has gone down significantly, leading to a smoother experience.

Tomb Raider

Before (max settings):


After (max settings):


Here we can see the minimum FPS going up by 28%. Rest of the values hit Vsync, so they look pretty similar.

Max Max

Now this game outputs the benchmark data in a couple of .XML files, read about it's capabilities here as reported by Liam: https://www.gamingonlinux.com/articles/mad-max-meets-vulkan-in-a-new-fully-public-beta-for-linux-benchmarks-and-opengl-vs-vulkan-comparisons.9345.

It runs through a couple of scenes, all of them were played at maximum settings:

Camp - Hollow Point

Before:

Minimum FPS: 31
Average FPS: 58
Maximum FPS: 96

After:

Minimum FPS: 52
Average FPS: 63
Maximum FPS: 84

Minimum FPS showed an increase of about 62%, while the average showed an increase of just 8%. Not sure what happened to the maximum FPS there.

Stronghold - Tyrant's Lash

Before:

Minimum FPS: 20
Average FPS: 29
Maximum FPS: 37

After:

Minimum FPS: 38
Average FPS: 53
Maximum FPS: 62

Minimum FPS showed an increase of about 90%, while the average showed an increase of 80% and the maximum FPS increased a healthy 67% too! Seems this scene is quite CPU-heavy.

Cutscene - Hope, Glory, and Dog is Dead

Before:

Minimum FPS: 4
Average FPS: 56
Maximum FPS: 180

After:

Minimum FPS: 26
Average FPS: 59
Maximum FPS: 214

Minimum FPS showed an increase of about 550% (!), while the average showed an increase of just 5% and the maximum FPS increased 18%. Ryzen helps a lot with minimum FPS in this scene.

Cutscene - Landmover

Before:

Minimum FPS: 4
Average FPS: 56
Maximum FPS: 196

After:

Minimum FPS: 15
Average FPS: 57
Maximum FPS: 230

Minimum FPS showed an increase of about 275% (!), while the average remained more or less the same and the maximum FPS increased 17%. Ryzen again helps a lot with minimum FPS in this scene.

Unigine Superposition

As mentioned previously, I couldn't get it to run live on stream, but I got it fixed later. So I only have an older comparison point - with older kernel and Mesa, which makes this comparison less good. Nonetheless, here it goes:

Before:


After:


VR

One of the main reasons I wanted to get a new CPU is because I knew from watching other Linux people's videos (that have the same GPU), that my frame timings were quite a bit higher, bordering on the "red line" (which basically means frames weren't submitted fast enough).

So of course I was quite interested in seeing if Ryzen helped anything in this regard. Unfortunately, I didn't grab any screenshots before changing to Ryzen, but here it goes, playing Talos Principle VR:



As you can see, with the same settings, frame times are well below the red line. Before, the red line was being touched slightly by the purple line. I don't have the previous screenshot, so you're gonna just have to take my word for it :)

Conclusion

Looking at the upgrade, I'm quite happy, to say the least. I'm hitting Vsync in the tested games no problem, which was sometimes, but not always possible on the FX.
On the VR front, and on the main reason I did the upgrade, I can see improved frame times, though I am going to have to make a bit more tests.

That's all from me, thanks for reading, and have a lot of Linux gaming fun! :)