- Joined
- Mar 20, 2019
- Messages
- 428 (0.19/day)
- Location
- Australia
| System Name | Ryzen |
|---|---|
| Processor | AMD Ryzen 7 5700X |
| Motherboard | Asus TUF Gaming B550-Plus (Wi-Fi) |
| Cooling | Cryorig H7 |
| Memory | Kingston Fury Beast DDR4 3200MHz 2x8GB + 2x16GB |
| Video Card(s) | Sapphire NITRO+ Radeon RX 6700 XT GAMING OC |
| Storage | WD_Black SN850 500GB NVMe SSD + Adata XPG SX8200 Pro 512GB NVMe SSD |
| Display(s) | Gigabyte G27QC |
| Case | NZXT H510 Flow |
| Audio Device(s) | SteelSeries Arctis Prime |
| Power Supply | Corsair RM650x Gold 650W |
| Mouse | Logitech G502 X |
| Keyboard | HyperX Alloy FPS Cherry MX Blue |
| Software | Windows 11 Pro |
Intoduction
Hi all, I thought I would show some Linux love on the TechPowerUp forums by performing some gaming benchmarks and posting the results here.
With Valve pushing development of Steam on Linux through the Proton compatibility layer and AMD Radeon drivers getting better and better, now is as good a time as ever to start looking at some Linux gaming performance benchmarks.
Much of these kinds of GPU performance benchmarks are detailed by Michael Larabel over in the excellent Phoronix.com, where sometimes a dizzying array of GPUs are pitted against each other on a regular basis. Check out his recent reviews of the GTX 1660 and Radeon VII here for instance: https://www.phoronix.com/scan.php?page=category&item=Graphics+Cards
Here I want to explore how a (perhaps) somewhat typical, older system fares between recent Ubuntu releases (Kubuntu to be precise, but this is based on the same core), namely 18.04 LTS, and the recently released 19.04.
For testing I will be using the Radeon RX 570 - a card that has been on the market since April 2017 - more than enough time for amdgpu Linux drivers to be well matured at this point in terms of stability. I'm going to pair this with an overclocked i5 3570k on the Z77 platform - far from the latest and greatest CPU and chipset, but I believe this to be a fair baseline given that it won't be a bottleneck for the mid-range RX 570.
Measuring performance improvements (or regressions) between the latest Ubuntu release and the latest LTS release will be the goal of these benchmarks. This will hopefully highlight how much work AMD has (or has not) put into refining Mesa and their amdgpu drivers through two releases of Ubuntu, noting that Mesa 18.2.8 is default now on Ubuntu 18.04, and Mesa 19.0.3 is in use on Ubuntu 19.04. We’ll see whether upgrading from a LTS release is worth it from a GPU performance perspective (ignoring the possibility of upgrading to the latest kernel or Mesa releases on 18.04 by using PPAs).
Unfortunately the selection of games is quite limited. It is quite difficult to find benchmarks for PC games, let alone games that will also run under Linux – although that second point is improving fortunately. Secondly, I am relying on what is available in my own personal Steam collection – also rather limited. Hopefully enough though to make a few observations.
As an additional point of reference, I'll also compare benchmarks here to the same benchmarks in Windows 10 1809 running the latest Adrenalin 2019 Edition 19.4.3 Radeon drivers.
System Specs
Video Card: Gigabyte Radeon RX 570 Gaming 4G
CPU: Intel Core i5 3570k @ 4.2GHz
Mobo: Asus P8Z77-M Pro
RAM: DDR3 16GB (4x4GB) 1600MHz CL3 Corsair Vengeance
Storage: Samsung 840 Pro 256GB SATA3 SSD
PSU: Corsair CX650 650W Gold
Benchmarks
All benchmarks are run at 1920x1080 screen resolution, and all games are run under Ubuntu using the native version of the game where possible (i.e. without the Proton compatibility layer), Thief being the exception here.
Benchmark 1: Tomb Raider (Crystal Engine) – 2013
There is a small performance gain here n Ubuntu 19.04 over 18.04 – about a 4 AVG FPS gain on the newer release. Windows is well ahead here.
Benchmark 2: CS:GO (Source) – 2012
There is a small performance gain here n Ubuntu 19.04 over 18.04 – about a 7 AVG FPS gain on the newer release. Still some catching up to Windows to do though.
Benchmark 3: Thief (Unreal Engine 3) – 2014
This is a bit of a quirky one. The only way that I was able to get this running in Ubuntu was by using the Steam launch command “PROTON_USE_WINED3D11=1 PROTON_NO_ESYNC=1 %command%”, which unfortunately affects performance under Linux significantly as you’ll see below. If someone has a workaround, please let me know as I would love to retest this benchmark.
Interestingly on Ubuntu 19.04, the benchmark didn’t hit rock bottom at 3.1 FPS minimum in the same fashion as 18.04 did, although I still wouldn’t call 23FPS an acceptable average for the newer release. Obviously this is a result of the workaround to get this game to even run in the first place as the game runs smooth as silk under Windows.
Benchmark 4: Metro: Last Light Redux (4A Engine) – 2014
Metro Last Light Redux actually shows a slight performance regression for the latest release of Ubuntu.
Benchmark 5: Dota 2 (Source 2) – 2013
Benchmark file taken from the Global eSports Cup Finals, posted on Phoronix here:
https://www.phoronix.com/forums/for...marks-with-valve-s-dota-2?p=873909#post873909
Using this launch option:
-vulkan +timedemo *demo-file-name* +fps_max 0 -novconsole -autoconfig_level 3 -high
A nice little gain is made in Dota 2 between releases – almost catching up to the Windows release with only 2.6 AVG FPS separating them in the eSports Cup Final timedemo.
Conclusion
I’ll let you be the judge of the results of this very small array of games in my Steam collection, although in general you could come to the conclusion that by upgrading to the latest version of Ubuntu, you’ll see at lease a moderate improvement in performance if you’re using one of these RX 570 cards.
Closing Thoughts
This is the first time that I’ve compiled something like this, but I’ve really enjoyed the process and being able to compare results between configurations. In the future I’d like to throw in some more benchmarks with additional hardware to gain some more in-depth insight as to how things are improving in Penguin Land. Until then though, I hope you’ve enjoyed my little analysis.
Hi all, I thought I would show some Linux love on the TechPowerUp forums by performing some gaming benchmarks and posting the results here.
With Valve pushing development of Steam on Linux through the Proton compatibility layer and AMD Radeon drivers getting better and better, now is as good a time as ever to start looking at some Linux gaming performance benchmarks.
Much of these kinds of GPU performance benchmarks are detailed by Michael Larabel over in the excellent Phoronix.com, where sometimes a dizzying array of GPUs are pitted against each other on a regular basis. Check out his recent reviews of the GTX 1660 and Radeon VII here for instance: https://www.phoronix.com/scan.php?page=category&item=Graphics+Cards
Here I want to explore how a (perhaps) somewhat typical, older system fares between recent Ubuntu releases (Kubuntu to be precise, but this is based on the same core), namely 18.04 LTS, and the recently released 19.04.
For testing I will be using the Radeon RX 570 - a card that has been on the market since April 2017 - more than enough time for amdgpu Linux drivers to be well matured at this point in terms of stability. I'm going to pair this with an overclocked i5 3570k on the Z77 platform - far from the latest and greatest CPU and chipset, but I believe this to be a fair baseline given that it won't be a bottleneck for the mid-range RX 570.
Measuring performance improvements (or regressions) between the latest Ubuntu release and the latest LTS release will be the goal of these benchmarks. This will hopefully highlight how much work AMD has (or has not) put into refining Mesa and their amdgpu drivers through two releases of Ubuntu, noting that Mesa 18.2.8 is default now on Ubuntu 18.04, and Mesa 19.0.3 is in use on Ubuntu 19.04. We’ll see whether upgrading from a LTS release is worth it from a GPU performance perspective (ignoring the possibility of upgrading to the latest kernel or Mesa releases on 18.04 by using PPAs).
Unfortunately the selection of games is quite limited. It is quite difficult to find benchmarks for PC games, let alone games that will also run under Linux – although that second point is improving fortunately. Secondly, I am relying on what is available in my own personal Steam collection – also rather limited. Hopefully enough though to make a few observations.
As an additional point of reference, I'll also compare benchmarks here to the same benchmarks in Windows 10 1809 running the latest Adrenalin 2019 Edition 19.4.3 Radeon drivers.
System Specs
Video Card: Gigabyte Radeon RX 570 Gaming 4G
CPU: Intel Core i5 3570k @ 4.2GHz
Mobo: Asus P8Z77-M Pro
RAM: DDR3 16GB (4x4GB) 1600MHz CL3 Corsair Vengeance
Storage: Samsung 840 Pro 256GB SATA3 SSD
PSU: Corsair CX650 650W Gold
Benchmarks
All benchmarks are run at 1920x1080 screen resolution, and all games are run under Ubuntu using the native version of the game where possible (i.e. without the Proton compatibility layer), Thief being the exception here.
Benchmark 1: Tomb Raider (Crystal Engine) – 2013
- Quality: Ultimate
- Texture Quality: Ultra
- Texture Filter: Anisotropic 16x
- Anti-Aliasing: FXAA
- Shadows: Normal
- Shadow Resolution: High
- Level of Detail: Ultra
- Post Processing: Yes
- Tessellation: Yes
- Hair Quality: TressFX
- Reflections: High
- Depth-of-Field: Ultra
- SSAO: Ultra
- Motion Blur: Yes
- Screen Reflections: Yes
There is a small performance gain here n Ubuntu 19.04 over 18.04 – about a 4 AVG FPS gain on the newer release. Windows is well ahead here.
Benchmark 2: CS:GO (Source) – 2012
- Global Shadow Quality: High
- Model/Texture Detail: High
- Effect Detail: High
- Shader Detail: Very High
- Multicore Rendering: Enabled
- Multisampling Anti-Aliasing Mode: 8x MSAA
- FXAA Anti-Aliasing: Enabled
- Texture Filtering Mode: Trilinear
- Wait For Vertical Sync: Disabled
- Motion Blur: Disabled
There is a small performance gain here n Ubuntu 19.04 over 18.04 – about a 7 AVG FPS gain on the newer release. Still some catching up to Windows to do though.
Benchmark 3: Thief (Unreal Engine 3) – 2014
This is a bit of a quirky one. The only way that I was able to get this running in Ubuntu was by using the Steam launch command “PROTON_USE_WINED3D11=1 PROTON_NO_ESYNC=1 %command%”, which unfortunately affects performance under Linux significantly as you’ll see below. If someone has a workaround, please let me know as I would love to retest this benchmark.
- Preset: Very High
- Texture Quality: Very High
- Depth-of-Field Quality: High
- Texture Filtering Quality: 8x Anisotropic
- SSAA: High
- Automatically Limit Texture Quality: Default
- Screenspace Reflection: Yes
- Parallax Occlusion: Yes
- FXAA: Yes
- Contact Hardening Shadows: Yes
- Tessellation: Yes
- Image-based Reflection: Yes
Interestingly on Ubuntu 19.04, the benchmark didn’t hit rock bottom at 3.1 FPS minimum in the same fashion as 18.04 did, although I still wouldn’t call 23FPS an acceptable average for the newer release. Obviously this is a result of the workaround to get this game to even run in the first place as the game runs smooth as silk under Windows.
Benchmark 4: Metro: Last Light Redux (4A Engine) – 2014
- Quality: Very High
- SSAA: Off
- Texture Filtering: AF 16x
- Motion Blur: Normal
- VSync: Off
Metro Last Light Redux actually shows a slight performance regression for the latest release of Ubuntu.
Benchmark 5: Dota 2 (Source 2) – 2013
Benchmark file taken from the Global eSports Cup Finals, posted on Phoronix here:
https://www.phoronix.com/forums/for...marks-with-valve-s-dota-2?p=873909#post873909
Using this launch option:
-vulkan +timedemo *demo-file-name* +fps_max 0 -novconsole -autoconfig_level 3 -high
A nice little gain is made in Dota 2 between releases – almost catching up to the Windows release with only 2.6 AVG FPS separating them in the eSports Cup Final timedemo.
Conclusion
I’ll let you be the judge of the results of this very small array of games in my Steam collection, although in general you could come to the conclusion that by upgrading to the latest version of Ubuntu, you’ll see at lease a moderate improvement in performance if you’re using one of these RX 570 cards.
Closing Thoughts
This is the first time that I’ve compiled something like this, but I’ve really enjoyed the process and being able to compare results between configurations. In the future I’d like to throw in some more benchmarks with additional hardware to gain some more in-depth insight as to how things are improving in Penguin Land. Until then though, I hope you’ve enjoyed my little analysis.
Last edited:
