Battlefield V NVIDIA Ray Tracing - CPU Testing @ [H]

Discussion in 'Video Cards' started by Kyle_Bennett, Jan 2, 2019.

  1. Kyle_Bennett

    Kyle_Bennett Editor-in-Chief HardOCP Staff Member

    Messages:
    52,075
    Joined:
    May 18, 1997
    Battlefield V NVIDIA Ray Tracing - CPU Testing

    We have been doing some deep dives into playing Battlefield V 64-person multiplayer lately and testing what exactly the cost of using NVIDIA Ray Tracing is in terms of framerate performance using new NVIDIA RTX 2070 and RTX 2080 cards. We did get questioned on using a 5GHz overclocked 7700K instead of the suggested CPU that EA recommends.

    If you like our content, please support HardOCP on Patreon.
     
  2. DF-1

    DF-1 2[H]4U

    Messages:
    2,518
    Joined:
    Jun 17, 2011
    good on you.

    sometimes the results aren't the most interesting. They're still useful to know.
     
    Aioeyu likes this.
  3. Dayaks

    Dayaks [H]ardness Supreme

    Messages:
    6,609
    Joined:
    Feb 22, 2012
    Welp, Kyle was right about there not being a CPU bottleneck. Good to know.
     
  4. Nightfire

    Nightfire [H]ard|Gawd

    Messages:
    1,346
    Joined:
    Sep 7, 2017
    Good on you for checking anyhow, Kyle. Still, I would want 6/12 config for this game to help with frametimes in multiplayer.

    The 2080ti seems to do WAY better than the 2080 than the 2080 does in comparison to the 2070 running DXR ultra at 1440p (53.9 vs 42.8 vs 39)

    Kyle, don't flame me for asking the following, but would having 32 GB of DDR4 help with the GTX 2080? I know it used 'only' around 11 GB of the 16 GB that you had, but I wasn't sure how much the other programs you were running used. Steve from Techspot saw a slight reduction going from 32 GB to 16 GB and that was with the 2080ti, which had a bigger memory 'pool' of 11 GB + 16 GB. He may have been running more programs, though.

    In any case, it is too bad that the GTX 2080 does not have more available vram.
     
    ltron likes this.
  5. OmenemO

    OmenemO Limp Gawd

    Messages:
    134
    Joined:
    Feb 25, 2016
    I was hoping you were wrong, only so I have an excuse to upgrade.
     
    AceGoober, Dayaks and alxlwson like this.
  6. Brent_Justice

    Brent_Justice [H] Video Card Managing Editor Staff Member

    Messages:
    17,825
    Joined:
    Apr 17, 2000
    While we did not test frametimes specifically, my experience with the 9700K is that I experienced the exact same stuttering I encountered with the 7700K with DXR enabled at the higher levels at 1080p and 1440p.

    The stuttering, as we have found, is actually related to VRAM capacity, and the game surpassing the 8GB of VRAM on the 2070 and 2080. We will talk about this in our upcoming article.

    The CPU had no impact for me, on the smoothness of multiplayer while gaming, with or without DXR.

    In terms of system RAM, the 11GB-12GB peak was total system RAM being utilized, out of the 16GB available. Just idling, right now, with nothing running, Windows at the desktop, the system is using 1.7GB of system RAM.

    I do not feel that increasing RAM capacity will make any difference in BFV gaming.
     
    luke51087, N4CR, AlphaQup and 9 others like this.
  7. Al Capwn

    Al Capwn n00b

    Messages:
    28
    Joined:
    Nov 27, 2010
    In the conclusion you mention that using a 4 core CPU isn't really an issue in BF4, but aren't you running HT on the i7? I would really love to see some comparisons of it running as a simulated 4-core as my i5-2500k @ 4.5ghz and GTX 1080 (and now 2070) can't keep FPS above 30 when the action gets intense on Rotterdam or really any 64 player map with lots of people around; DXR is disabled (as well as those terrible options such as chromatic aberration disabled). And yes I know my CPU is not even minimum but @ 4.5 I would hope it's sort of close to a stock i5-6600k.

    The only thing that actually helped slightly was running DX12 w/ FFR off. Running DX11 FFR on definitely gives slightly higher fps until the action starts and then it drops far lower for me, and the slight delay from FFR is noticeable.

    As a side note on the stuttering in general, that's the one issue I no longer have after clearing the BFV shader cache in the documents folder, but you already mentioned you are exceeding your VRAM capacity so that obviously wouldn't help. Interestingly I haven't seen VRAM usage go above 5.5GB so far, even when I was trying out ultra DXR to see how brutally slow I could go.

    Also thank you for this review as there really aren't any CPU comparisons available for BFV.
     
    Last edited: Jan 3, 2019
    Sulphademus likes this.
  8. Kyle_Bennett

    Kyle_Bennett Editor-in-Chief HardOCP Staff Member

    Messages:
    52,075
    Joined:
    May 18, 1997
    Once we get done here, I am likely to send out the 9900K to Brent to look at comparisons there.
     
  9. noko

    noko [H]ardness Supreme

    Messages:
    4,134
    Joined:
    Apr 14, 2010
    One kickass 7700k system.

    EA recommendation for the higher end is 6 core but the processor listed is 12 threads. Have no idea if that would make any kind of difference. I doubt it but Good to see some 9900k testing planned.

    Top notch follow up and confirmation with real testing.

    Since my testing is so different and not applicable to DXR, Turing etc. I best just start another tread dealing with BF5, Vega and Ryzen.

    Really appreciate [H]ardOCP attention to detail, strict controlled testing for accurate results.
     
    lostin3d, Maddness and Pieter3dnow like this.
  10. lostin3d

    lostin3d [H]ard|Gawd

    Messages:
    1,838
    Joined:
    Oct 13, 2016
    As someone who's on the verge of a new build in the next 1-2 years I appreciate these CPU tests. It's nice to see how a healthy OC will still keep an older gen performing well. I recently was able to get 4.3GHZ on my 4930k on air and just having that has shown it's got life for at least one more x80TI gen in 4k/60fps.

    I remember around 6-7 years ago many believed 5GHZ to be the halo for 4c/8t and this seems to show how true that is. It'd be massive undertaking, but it would be cool, to see someone bench the most popular I7's 2nd-6th(left out 7th since you already have it) 4c/8t CPU's all clocked at, or closest to, 5GHZ to see how they fare in BFV.
     
  11. GoodBoy

    GoodBoy [H]ard|Gawd

    Messages:
    1,161
    Joined:
    Nov 29, 2004
  12. trick0502

    trick0502 [H]ardness Supreme

    Messages:
    5,163
    Joined:
    Apr 17, 2006
    Even though EA suggests 6-core CPUs for playing Battlefield V, it is obvious to us that a highly overclocked 4-core / 8-thread CPU can still get the job done.
     
    AceGoober and lostin3d like this.
  13. Starrbuck

    Starrbuck 2[H]4U

    Messages:
    2,408
    Joined:
    Jun 12, 2005
    Some Ryzen results would be nice to see.
     
  14. agentbb007

    agentbb007 n00b

    Messages:
    3
    Joined:
    Jul 9, 2015
    Even though a 4-core CPU can still get the job done I would be curious what the minimum fps is. I have a feeling a 4-core CPU is going to stutter. Might have a decent average FPS but minimum fps is just as important as average.
     
  15. noko

    noko [H]ardness Supreme

    Messages:
    4,134
    Joined:
    Apr 14, 2010
    Frame times would show this or not. I am sure Brent or Kyle would jump all over obvious stutter different from the 9700K.

    I will do some more Ryzen testing and post in different thread. May have to pull the Vega FEs and put in the Vega 64 LC for this to be on the latest supporting drivers.
     
  16. Rauelius

    Rauelius 2[H]4U

    Messages:
    2,231
    Joined:
    Jan 2, 2008
    Rule of thumb for gaming, 2 Threads = 1 Non-SMT Core So and i7 7700k @ 5Ghz SHOULD perform about the same as an i5 9600k @ 5Ghz.
     
  17. NKD

    NKD [H]ardness Supreme

    Messages:
    7,431
    Joined:
    Aug 26, 2007
    Is it normal to be pissed off to see shitty as sites making more in Patreon then Hardocp?

    I can understand that they have a larger youtube presence. But to me everyone that loves this site should be able to donate a one Grande Starbucks a month to Hardocp so kyle never has to sign another fuckin NDA again lol. Truly boggles my mind to see little over 400 supporters.

    I honestly never gave money to anyone but when I sat down and thought about the news Kyle has brought us and level of details in the reviews I could cut down on coffee a bit easily.

    Wish we had 1000 users averaging 5 bucks.

    Rant over. LOL!
     
  18. psyclist

    psyclist Gawd

    Messages:
    852
    Joined:
    Jan 25, 2005
    You guys are awesome! thanks for the follow up. I appreciate that you guys take input from your readers and paint a more complete picture of performance. Id love to see how my machine stacks up to these results, glad I went HEDT 5 years back!
     
  19. blandead

    blandead Limp Gawd

    Messages:
    202
    Joined:
    Nov 6, 2010
    Everyone is really hung up on 6 cores, those recommended specs are general guidelines

    I somehow doubt you'll find any benchmark showing the 7700k @ 5 Ghz being slower

    Great follow up data though!
     
    Last edited: Jan 4, 2019
    Kyle_Bennett likes this.
  20. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,382
    Joined:
    Oct 29, 2000
    This is very interesting. It suggests that unlike current rasterization calls, even with Nvidia's new RTX technology, there is more that is not accelerated on the GPU than with rasterization, and thus is left to the CPU. Small differences indeed, but this is only in the ~60fps range, which is typically a breeze for just about any CPU, and now we are seeing differences in framerate (albeit small) between two very highly clocked CPU's.

    Interesting indeed.

    My prediction would be that if Nvidia is right about this, and real time ray tracing is the future of gaming, then we will slowly see more and more of this move to the GPU, just like we did with rasterization back in the day, when bit by bit more things became GPU accelerated. First there was hardware transform and lighting (T&L) and then feature after feature moved from the CPU to the GPU. If ray tracing becomes popular, this will likely happen here too.
     
    noko and Maddness like this.
  21. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,382
    Joined:
    Oct 29, 2000

    I don't understand why the retards love YouTube so much. Everyone knows a written review is by far superior to a video review in just about every case. Yet because of the sheeple, YouTube and video is sadly where the money is :(
     
    DrezKill, noko and Maddness like this.
  22. Nightfire

    Nightfire [H]ard|Gawd

    Messages:
    1,346
    Joined:
    Sep 7, 2017
    A 6/12 is much closer to an 8/8 so 3 threads = 1 core. In most games a 6/6 beats the 4/8 by a good margin.
     
    noko likes this.
  23. noko

    noko [H]ardness Supreme

    Messages:
    4,134
    Joined:
    Apr 14, 2010
    I think the media or video has some advantages as well and good to compliment a well written research, professionally edited work, Kyle does this and not to extremes either complimenting article with video - like how to install a TR CPU in the socket - that was A+ for clarity and to the point and saved many folks from many problems and frustrations. Somethings I think videos are utterly poor at, like a motherboard review, if one actually puts out in the video what a good motherboard review consists of it would be so long winded, boring and slow to watch it would loose most of the audience. Supplementing as needed the article for things video does well in will just make both forms of content better overall.
     
  24. noko

    noko [H]ardness Supreme

    Messages:
    4,134
    Joined:
    Apr 14, 2010
    Noted, fixed. Well at least add one that is. Thanks for the guilt :D
     
    Aioeyu likes this.
  25. Nolan7689

    Nolan7689 [H]ard|Gawd

    Messages:
    1,189
    Joined:
    Jun 5, 2015
    YouTube is low effort on the part of the viewer. Like watching a letsplay or streamer it’s a low effort way of experiencing the game. It can also roll in the background as you do something else and just listen to the commentary.

    Personally I’d rather read though, it’s quicker for me and easier to drop and pick up later than a video. Especially on my phone as I’m away from PCs most the day.
     
    Kyle_Bennett likes this.
  26. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,382
    Joined:
    Oct 29, 2000
    I guess I don't consider reading to be effort.

    I'm more concerned with my time, and videos tend to waste my time a lot.

    Everything from the intro sequence, to someone else's taste in music and the obligatory "hey guys..." introduction just takes so much time.

    Reading is also faster than spoken word. The few times that I need to watch YouTube videos I tend to watch them at 1.5-2x speed depending on how the person speaks.

    Also kills me to have to pause and look at charts, and the fact that skipping forwards and backwards to reference things is near impossible unless you remember the timestamp where the information you wanted happened to be, where as skimming back and forth in something written is super simple.

    Anyway, I don't want to drag this off topic any more than we already have...
     
    DrezKill and Nolan7689 like this.
  27. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,382
    Joined:
    Oct 29, 2000
    Agreed. There certainly are things videos excel at. How to guides are among them. And there is a place for short clips embedded in reviews to support this. Im just decrying the full video review of stuff, for people who are apparently too lazy to even read, sometgijg your brain just does automatically when you look at something.

    I fear we are becoming illiterate.
     
    psyclist and noko like this.
  28. Slash3

    Slash3 Limp Gawd

    Messages:
    209
    Joined:
    Jan 25, 2008
    The Russian site GameGPU has an indexed benchmark for Battlefield V, comparing both CPUs and GPUs amongst other things. It's a bit dense, but scroll down near the bottom and you'll see there's actually a core test result feature as well. Selecting the i7-7700K and the i7-9700K and you'll also see that the 9700K is tested with HT on and off. With HT off, it's CPU utilization is very similar to the 7700K.
     
  29. Kyle_Bennett

    Kyle_Bennett Editor-in-Chief HardOCP Staff Member

    Messages:
    52,075
    Joined:
    May 18, 1997
    We are starting 9900K testing today, so that should put an end to all the discussion about cores and threads and we will all see exactly what is going on.
     
    DrezKill, Dayaks, Chimpee and 3 others like this.
  30. Nightfire

    Nightfire [H]ard|Gawd

    Messages:
    1,346
    Joined:
    Sep 7, 2017
    Are we to believe that this website did all 30 of these GPU tests using all 30 cpus in 3 different resolutions? That is 2700 tests if they did them just once each.
     
  31. Ranger101

    Ranger101 [H]Lite

    Messages:
    75
    Joined:
    Sep 11, 2015
    Agree 100% that a written review is (almost) always preferable, but there may come a time when the old adage, "if you can't beat them, join them" may prove to be a prudent course of action.
     
  32. Sailindawg

    Sailindawg [H]Lite

    Messages:
    78
    Joined:
    Dec 20, 2006
    Thanks for review. The facts, nothing but the facts. I like the very direct conclusion regarding RTX.
     
    Kyle_Bennett likes this.
  33. Zarathustra[H]

    Zarathustra[H] Official Forum Curmudgeon

    Messages:
    27,382
    Joined:
    Oct 29, 2000
    Yeah, that is what I worry about.

    Welcome to the de-evolution. For all of human history technology generally improved over time. Starting in the early 2000's that changed, and things started getting worse.
     
  34. Kyle_Bennett

    Kyle_Bennett Editor-in-Chief HardOCP Staff Member

    Messages:
    52,075
    Joined:
    May 18, 1997
    That will not be happening. There is no way to present the in-depth coverage we do in a video format. It is simply impossible.
     
    DrezKill, noko and Maddness like this.
  35. Nolan7689

    Nolan7689 [H]ard|Gawd

    Messages:
    1,189
    Joined:
    Jun 5, 2015
    Nah, it’s just impossible to do it timely and concisely and in a watchable manner. You could still display the relevant graphics while someone just reads the review word for word as written.

    Please don’t do that, it would be terrible. Like a dry college professor giving a lecture.
     
  36. Kyle_Bennett

    Kyle_Bennett Editor-in-Chief HardOCP Staff Member

    Messages:
    52,075
    Joined:
    May 18, 1997
    Like I said, no way to do that. I could make shit all day long that no one wants to watch.
     
    PhaseNoise and Legendary Gamer like this.
  37. dad_H

    dad_H n00b

    Messages:
    5
    Joined:
    Jan 11, 2019
    How 9900k testing is going? When the results are ready? Could you please publish more data on cpu performance: CPU load graphs, GPU load graphs? How is cpu load changing when DXR is on/off. Also it's interesting to see the results of more high-end cpus with 10-32 cores if possible. On VRAM amount results it also looks like we need more data: 24 GB TITAN RTX should be tested to find how much GB is enough for DXR :)