VR 60% GPU Usage

Discussion in 'RaceRoom Racing Experience' started by tlsmikey, Mar 30, 2017.

  1. tlsmikey

    tlsmikey

    Messages:
    358
    Ratings:
    +127
    Anyone else running VR in Raceroom and noticed that the GPU and CPU performance never goes past 60%?

    No matter what i've tried, it never scales beyond that and it looks like i'm leaving a lot on the table in terms of performance.
     
  2. Jezziem

    Jezziem

    Messages:
    10
    Ratings:
    +9
    Personally, I don't have a VR setup, so I can't confirm any issues. What does your GPU usage look like without VR? Normally, that's an indicator for poor CPU performance in relation to your GPU. What are your PC components?
     
  3. Andrew Harper

    Andrew Harper
    Premium

    Messages:
    2,165
    Ratings:
    +1,533
    I'm sure someone will correct me if I'm wrong, isn't this an issue with DirectX 9 games?

    I read somewhere about DX9 titles not using all the capability of CPU/GPU's but I could be getting the details wrong.
     
  4. Thomas Wilen

    Thomas Wilen

    Messages:
    42
    Ratings:
    +6
    Yeah, pretty sure this is the issue. The RRRE engine is CPU bottlenecked. I have a 3370K@4.2 and a 1080 and I verified it boosts clocks if i force OC on, but GPU utilization is around 60-70% and no matter what wont go over that while CPU single core is 100% (but total CPU maybe 25%).
     
    • Like Like x 1
  5. Piotrek Blaszczak

    Piotrek Blaszczak
    Premium

    Messages:
    1,120
    Ratings:
    +746
    Me to :D.I have 1060 6GB , and GPU utilization across 60/70 %.For example I try AC and I have almost 90 GPU ussage , and only 35/45 CPU ussage so yes I thing R3R engine bottlenecked CPU.In Dirt Rally have this same ussage.
     
  6. jego

    jego
    Premium

    Messages:
    353
    Ratings:
    +384
  7. Kurupt CDN

    Kurupt CDN
    DirtT Tuned Motorsports Premium

    Messages:
    2,041
    Ratings:
    +1,427
    I noticed the same issue last night when setting up vr....i cant't get my gpu usage above 40%. I came across the same redit thread last night in search of a solution. Hopefully the devs can optimize vr some more as I cant justify buying faster ram for $100+ just to run Raceroom when all my other titles run fairly well in vr on an old system.
     
  8. RasmusP

    RasmusP
    AC Stuff and G27 help Staff

    Messages:
    3,678
    Ratings:
    +1,303
    To be honest it looks like the performance boost wasn't because of the ram speed, but because of dual channel.
    He bought 2x 8gb and had 8gb before. He doesn't mention if that was 1x 8gb or 2x 4gb.

    I stumpled upon the ram myth last year so I did some testing myself and the boost is not that heavy from my experience.
    Some data from cinebench R15:

    I7 2600k:
    4.2, 1333: 533
    4.2, 1600: 553
    4.4, 1333: 572
    4.4, 1600: 582
    4.4, 1600 + HT: 754

    I5 2500k:
    4.5, 2133: 557

    I highly doubt that a few points in cinebench would result in such a massive fps increase for gaming. My friend with the 2500k tested it with Witcher 2. Going from 1333 to 2133 and it was around 15 fps. But it was 150 to 135 fps and not 45 to 60 for example. Scaling!
    Tested at 800x600 for 100% taking the gpu out of the testing.

    Conclusion: yes, ram makes a difference but it isn't worth it to buy faster ram when you don't need more GB or build a new rig anyway!
     
    • Like Like x 1
  9. dud

    dud
    Premium

    Messages:
    434
    Ratings:
    +66
    You might want to verify that slow RAM is your problem. Downclock it. If the RAM is your problem you should see a significant further slowdown. If the speed stays the same then upgraded RAM is unlikely to make a positive difference either.

    You can generally downclock RAM safely and without research by simply setting the "multiplier" in the BIOS to something less than what you currently run.
     
    • Like Like x 2
  10. RasmusP

    RasmusP
    AC Stuff and G27 help Staff

    Messages:
    3,678
    Ratings:
    +1,303
    I really have to say, that I like your "opposite testing philosophy".
    Most suggestions are "overclock it and see the difference" whereas downclocking actually works absolutely well in most cases BUT is a lot easier! Seems like only a few people have this thought though..
     
    • Agree Agree x 1
  11. dud

    dud
    Premium

    Messages:
    434
    Ratings:
    +66
    Thanks for saying that. Hardware analysis both for power and for speed is painful, it's good to see it do something good. Assumptions should be checked and it isn't that people are too lazy to do it. It is that the only way that has been pointed out to you is so expensive that you cannot do it (in a job where a manager tries to budget your time). If you still want to do it then creativity kicks in.

    As a side note, it also helps that I always insisted on having physical test machines. All that datacenter and cloud stuff can go to hell. If I can't touch it I won't spend my time analyzing it in detail.
     
    • Like Like x 1