Announcement

Collapse
No announcement yet.

Dota 2 Receives Optimization For AMD Ryzen CPUs

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

  • cip91
    replied
    Michael could you do a test of the new Dota2 with the 1800x and the 7700k ? We could then "manually" confront it with this
    Last edited by cip91; 22 March 2017, 10:06 AM.

    Leave a comment:


  • Sidicas
    replied
    Originally posted by schmidtbag View Post
    It's a CPU benchmark - you want to use lower resolutions, to eliminate any other bottlenecks. This is exactly why people were whining about Ryzen in the first place - at 2K and 4K resolutions, the gaming performance in fine, but at 1080p its noticeably worse.

    As long as both the before and after results are 800x600, how is this situation any different?
    Actually low resolution benchmarks are more of a RAM clock speed benchmark than anything. When you do artificial benchmarks like that you create bottlenecks that would not occur in real world scenarios. If he over clocks his RAM and runs that same benchmark, he will see significant FPS gains but in the real world, RAM would never be a bottleneck and only comes into play at FPS above 80.

    Leave a comment:


  • schmidtbag
    replied
    Originally posted by Sidicas View Post
    Going to have to call this out as deceptive benchmarks are a pet peeve of mine.

    Look, no gamer plays at 800x600 resolution with vsync off. Obviously any such benchmark would be entirely CPU bottlenecked and your results severely exaggerate the performance gains. If any, the overwhelming majority of users will get while playing the game. Redo at 1920x1080 with vsync on.
    It's a CPU benchmark - you want to use lower resolutions, to eliminate any other bottlenecks. This is exactly why people were whining about Ryzen in the first place - at 2K and 4K resolutions, the gaming performance in fine, but at 1080p its noticeably worse.

    As long as both the before and after results are 800x600, how is this situation any different?

    Leave a comment:


  • Sidicas
    replied
    Originally posted by LinuxID10T View Post
    Here is the before: https://docs.google.com/spreadsheets...it?usp=sharing

    Here is the after: http://openbenchmarking.org/result/1...TA-RYZENDOTA62

    It made a pretty fucking big improvement. Went from 67 FPS before to 108 FPS after...
    Going to have to call this out as deceptive benchmarks are a pet peeve of mine.

    Look, no gamer plays at 800x600 resolution with vsync off. Obviously any such benchmark would be entirely CPU bottlenecked and your results severely exaggerate the performance gains. If any, the overwhelming majority of users will get while playing the game. Redo at 1920x1080 with vsync on.

    Leave a comment:


  • schmidtbag
    replied
    Originally posted by LinuxID10T View Post
    Here is the before: https://docs.google.com/spreadsheets...it?usp=sharing

    Here is the after: http://openbenchmarking.org/result/1...TA-RYZENDOTA62

    It made a pretty fucking big improvement. Went from 67 FPS before to 108 FPS after...
    Wow! Thanks a lot for posting the results. This is very critical information, because it shows how much a little bit of optimization can make all the difference. It didn't seem to take Valve very long to release these patches either, so clearly it wasn't too difficult for them to accomplish this. All that being said, when you consider the performance of Ryzen without any optimizations, it fares modestly well.

    This gives me hope that Ryzen will really shape up into something very nice.

    Leave a comment:


  • LinuxID10T
    replied
    Here is the before: https://docs.google.com/spreadsheets...it?usp=sharing

    Here is the after: http://openbenchmarking.org/result/1...TA-RYZENDOTA62

    It made a pretty fucking big improvement. Went from 67 FPS before to 108 FPS after...

    Leave a comment:


  • LinuxID10T
    replied
    Originally posted by schmidtbag View Post
    @LinuxID10T
    Though I don't play DOTA2 and likely never will, I am still interested in these results. DOTA2 will be the first commercial example in years of what happens when you tweak software for a specific CPU architecture. To me, that's very interesting. Mind keeping us posted if Michael won't?
    Yeah, I'll start my tests with it and have some results this evening.

    Leave a comment:


  • sterky
    replied
    I diddn't save the run pre dota update, but i have a run @1440p with GTX1070, Ryzen 1800X @stock (mem:3200mhz CL16) vs FX-8320 @4.6Ghz
    Ryzen after dota update and FX pre.
    result_link

    Leave a comment:


  • schmidtbag
    replied
    @LinuxID10T
    Though I don't play DOTA2 and likely never will, I am still interested in these results. DOTA2 will be the first commercial example in years of what happens when you tweak software for a specific CPU architecture. To me, that's very interesting. Mind keeping us posted if Michael won't?
    Last edited by schmidtbag; 21 March 2017, 11:45 AM.

    Leave a comment:


  • Michael_S
    replied
    Originally posted by LinuxID10T View Post

    Assuming I haven't changed the hardware, which I haven't, it should be the same. I lock an overclock of 3.9 GHz for it so it is pretty consistent, no mobo/uefi/power management weirdness.
    I think AsuMagic thought you were offering to compare your before-patch benchmarks with after-patch benchmarks that Michael does. That wouldn't be useful, for the reasons he states.

    But before-and-after on your hardware makes complete sense.

    Leave a comment:

Working...
X