Announcement

Collapse
No announcement yet.

Intel Pentium vs. AMD Ryzen 3 Performance For Linux Gaming

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • bridgman
    replied
    Originally posted by dungeon View Post
    Well, it goes down I am talking about this specific tendency. Michael benches in this article show exactly that this happen, especially when you look at nvidia results.

    amdgpu obviosly scale down with more cores in these examples, while nvidia driver scale up.
    Oh good grief. The amdgpu+mesa stack does *not* "scale down with more cores". If you run on the same CPU (eg Ryzen 1200 or whatever) with 2 cores enabled vs 4 cores enabled you will get the same or higher performance with 4 cores.

    If you want to prove that amdgpu+mesa goes down with more cores you need a graph that shows Ryzen with 2 cores enabled vs same Ryzen with 4 cores enabled, not two different CPUs with two different core designs. I don't believe that graph exists. Please do not create one with a paint tool.

    The amdgpu+mesa stack as tested (not sure if "as-tested" enabled threading) can not take much advantage of the Ryzen's additional cores, and so the Pentium's higher clocks end up having more impact on frame rate than the additional cores.

    With the NVidia driver the gain from additional cores outweighs the loss from lower clock speeds.
    Last edited by bridgman; 24 January 2018, 10:18 PM.

    Leave a comment:


  • dungeon
    replied
    Originally posted by F i L View Post
    Dude.. The AMDGPU drivers do not "go down" with more cores.
    Well, it goes down I am talking about this specific tendency. Michael benches in this article show exactly that this happen, especially when you look at nvidia results.

    amdgpu obviosly scale down with more cores in these examples, while nvidia driver scale up.

    They just go up on the i3's cores that are clocked higher and have more IPC (Instructions Per Clock).
    I know what is IPC, but that does not matter here . As nvidia driver scale up fine with both these Intel and AMD CPUs, while amdgpu scales down going from these 2 cores to 4 - i mean that is so obvious

    Multiple people have said this already.
    I know that but i don't care, as this does not matter in this case

    Everyone agrees with you that Nvidia's drivers take better advantage of higher-core CPUs. No one is disputing that.
    I don't care about that also and i am not trying to disputing anything, i don't care that nvidia driver is better there... i just comment that amdgpu is worse as it does not scale there which is worse for an user who will switch to these more core CPUs potentionaly, and got expected Horse Shit like this bench showed, etc... probably some people here does not clearly understand my point, but well what I can do about that

    Again i don't care that nVidia is better and i don't care about CPUs... i have in my mind average Joe who ask: Should these 4 core CPUs improve my gaming performance, i currently have 2 core? And answer is - yes, if you are using nvidia driver, with amdgpu no - there it will go mostly down
    Last edited by dungeon; 24 January 2018, 10:15 PM.

    Leave a comment:


  • F i L
    replied
    Originally posted by dungeon View Post
    This to me is so obvious amdgpu goes down with more cores, while nvidia scales up
    Dude.. The AMDGPU drivers do not "go down" with more cores. They just go up on the i3's cores that are clocked higher and have more IPC (Instructions Per Clock). Multiple people have said this already. Everyone agrees with you that Nvidia's drivers take better advantage of higher-core CPUs. No one is disputing that.

    Leave a comment:


  • nuetzel
    replied
    Originally posted by darkbasic View Post
    Very interesting article, it seems that mesa's multi-threading is almost nonexistent compared to nvidia.
    If you don't enabled it?

    @Michael:

    I'm running here _all_ the time with

    mesa_glthread=true
    radeonsi_assume_no_z_fights=true
    radeonsi_commutative_blend_add=true
    R600_DEBUG=sisched

    in /etc/environment

    R600_DEBUG=sisched show big boost with UH, UV (US)

    And don't forget to set this for _GL_ (only):

    cpupower frequency-set -g performance

    With _Vulkan_ it show ~10% decrease on my devel/testing system.

    Greetings,
    Dieter

    Leave a comment:


  • nuetzel
    replied
    Originally posted by dungeon View Post
    AMD, is there at least FullHD version of this Adrenalin Pro i want it as wallpaper There were some PRO walpapers like that somewhere on your site, does anyone have a link?

    This could become a trademark thing with NIVEA, Hamburg, Germany. --- Apart, that it looks like...

    Leave a comment:


  • dungeon
    replied
    Originally posted by sa666666 View Post

    Yes, it's enough for you since (a) it confirms your viewpoint, and (b) you had your mind made up before the benchmarks were even posted. So the reasoning is irrelevant.
    Well, i put a notice and pictures when everybody could spot these scale up scale down drivers tendencies.

    I think that my point is more than regular one and not irrelevant at all. This to me is so obvious amdgpu goes down with more cores, while nvidia scales up - here are more pictures, this is very easely noticable in majority GL benches Micheal posted in this article



    My comment is about amdgpu driver POV only really, i don't care which hardware combo is better and so on, since quality of drivers mostly makes things better or worse... Maybe i should just say hi bridgman bump mareko to see these
    Last edited by dungeon; 24 January 2018, 05:27 PM.

    Leave a comment:


  • grok
    replied
    Originally posted by debianxfce View Post
    Dual core cpus are over 10 year old technology. When a 100 usd Leeco Le S3 smartphone has 10 CPU cores, a PC should have at least four. Tomb Raider 2013 is a well implemented game and that benchmark is all that matters. Use a non debug 1000Hz timer custom kernel and disable unused systemd services. For a budget PC the AMD Athlon X4 950 3,5 GHz is the best CPU and a AM4 motherboard does have a upgrade path to Ryzen. GTX750 from aliexpress.com is the best GPU for a budget gaming PC.
    https://www.aliexpress.com/item/100-...08665165.html?
    "Outdated" or not I would sure like to have an i3 7350K.
    Or for something lower power still there's that dual core, 3 CU, single channel Zen 2 APU that comes out in 2019, if I'm not wrong on specs..

    Leave a comment:


  • bridgman
    replied
    Originally posted by cRaZy-bisCuiT View Post
    @AMD: What is wrong with you? For a few weeks / months the prices for 560 / 570 /580 were at least almost back to its original level.

    Now I'd need to pay 330 € for a 570 and 350 € for a 580? Instead of whining crypto miners buy your cards you should be happy to earn that much money! And at the same time you should make sure to deliver!
    We don't earn that money - AFAIK the retailers/distributors are marking up the cards, not us.

    Leave a comment:


  • darkbasic
    replied
    Very interesting article, it seems that mesa's multi-threading is almost nonexistent compared to nvidia.

    Leave a comment:


  • SvenK
    replied
    It doesn't look bad for AMD if you set the Pentium and the Ryzen 3 to 3.4 GHz, i calculated it with the results from this Benchmarks, so i am sure with the Ryzen 2000 series, we will see equal Performance on Pentium G4600 and Ryzen 3 2200.
    Game RX 560 | Pentium G4600 @ 3.4 GHz RX 560 | Ryzen 3 1200 @ ~ 3.4 GHz
    BioShock Infinite 129.63 121.95
    Counter Strike: Global Offensive 146.63 151.27
    Deus Ex: Mankind Divided 37.37 40.1
    Dawn of War III 36.62 38.14
    Mad Max (Camp Hollow Point) 58.41 62.59
    Mad Max (Stronghold) 52.79 55.52
    Mad Max (Hope, Glory and Dog is Dead) 56.35 59.08
    Mad Max (Landmover) 63.61 67.27
    OpenArena v0.8.8 231.48 230.3
    Portal 435.47 455.06
    Serious Sam 3: BFE 101.56 146.45
    Tesseract v2014-05-12 286.62 274.55
    Team Fortress 2 115.72 110.07
    Tomb Raider 109.84 114.37
    Xonotic v0.8.2 202.23 194.06
    DiRT Rally 87.54 86.91

    Leave a comment:

Working...
X