Battlefield V NVIDIA Ray Tracing: i9-9900K CPU Testing @ [H]

Thanks, now it seems to be clear that more threads not help with dx12/dxr. But what happened to cpu/gpu load on your 9900K? Is it going to be lower than on 7700K: some threads just not used? GPU load with 'DXR on' is always about 100%? MIN fps is at 100% loaded gpu and not loaded CPU? Is dxr loading cpu more than just dx12? It seems min fps is not consistent here, better use 1% low fps?

 

Haven't played this game and almost everyone complains its stutters alot despite having 8 core CPUs. Seems to happen in multiplayer mostly from what I read not single player.

 

Strange is that your results show even some dx11 min/avg fps scaling with more threads, but no dx12/dxr scaling. While Dx12 is supposed to be the api to better utilize cores/threads in general. Maybe bf5 engine dx12 code is so poor and is the root cause of all the problems? Maybe more patches are on the way.

 

The rtx is a lie... Yeah, i think that much we could gather so far.

 

So DX12 needs some overhauling?
So still gpu limited. Maybe sli/cfx would be cpu limited?
In short don't doubt the [H] team!

 

Lol, "chuck it to the trash, RTX is a lie". I love it.

I would be interested to see how the numbers look on stock clocks, seeing as how that is what the specs on the box would be looking for.

This is exactly the kinda stuff that Lisa Sure was talking about during CES reviews- Having the hardware there without having a proper software package is a disaster.

I wonder if the ray tracing tech... If it's an issue with the software, the hardware just not being powerful enough, or both?


Thanks for the write-up!

 

Thanks Kyle and Brent for the insane amount of testing you've done with this title.

It was nice to see increased documentation of DX12. I've witnessed it with my rigs in this game, and others, and after this I just don't know how to technically suggest causes.

Bottom line for anyone's claims of success with DX12 is that if it works for them it doesn't mean it will for all and I believe many dev's already display such things when you switch to DX12 in games.

 

Interesting results; would be neat to see some Ryzen RTX benches.

 

Now that the CPU cores/thread debate can be buried, we can focus solely on the new RTX. A $1250+ GPU provides 30 FPS min and about 60 FPS avg with DXR ultra settings. That is one hell of a steep price of admission to get what is widely adopted as bare minimum acceptable performance...and at only 1080p. How pathetic.

Maybe the future RTX 3xxx series will give a 60-80% gain, so 1440p can be barely playable?

 

Another great review thank you! I do have to say that after testing a lot of different graphics options I ran into a lot of stuttering, but after clearing the BFV's shader cache in the documents folder it seemed to have fixed it and also stopped the constant pop-in I was getting. Also it's worth comparing DX11 FFR on to DX12 FFR off as FFR on doesn't increase DX12 fps but does add a very slight input delay. If there's ever a Ryzen version done it would be great to see the scaling from 4 threads to 16.

 

Hahahaha 30Fps in 1080 with RTX and Ultra settings/ lol Looks like the dream of 4K is SHOT IN THE ASS how does that 1200$ price tag feel now guys? The cars are a JOKE! No wonder Nvidia stock is down 45% And I own a 1080Ti But will be ditching it fort the new AMD card Who gives a shit about the Ray Tracing right now.

 

so DXR is slower so we dont need nvidia LULZ !!!!!!!!!!!!!!

 

I won't be happy until you run this test on a 10900K....!

This is why I work in a field where I don't face a customer. People go out of their way with utmost effort to complain and whine than be a part of the solution.

 

This is not a big surprise. Other than "embarrassingly parallel" algorithms, Amdahl's Law (from 1967) always applies.

 

Ouch!
Just ouch at the end.

 

So, in a nutchell there are some fanboys who have 9900K rigs who wanted to see how much worse, cough better cough, their CPU performs compared to an older CPU such as the prehistoric 7700K and discovered through the H's testing that they upgraded too soon and spent money needlessly... I enjoy the comparion, but some guys need to man up and increase or start funding via Patreon if they want more of these comparisons in the future!

 

chunk it!! I love this.

 

Negative.

Some of us with higher core count processors get higher fps and were wondering if that was the reason why. For instance, I get 75fps at 3440x1440 ultra/low where Brent got ~66fps at 2560x1440 ultra/low. So I had 16% higher fps at 35% more pixels. I also don’t notice any stuttering.

The results aren’t so different they are bogus, but it is a decent amount, it’s awesome [H] did this testing and we got to see how minimums improved with the 9900k. It’s great they investigated at all. A big part of why I am a Patreon.

The overall jist I noticed on the forum is if you have a high Hz monitor, high Hz >> ray tracing. DLSS is supposed to come to BFV and cancel out the fps drop from ray tracing *according to nVidia* so take that as you want.

 

I'm playing this game on a 4.3GHz 4770K (I lost the silicon lottery on this chip badly) and I don't have any issues with stuttering in multiplayer, fwiw. 4C/8T, playing with a Titan Xp (2016 model) at 2076MHz. I play at 1440p Ultra and it's usually around 90-100 fps on most maps.

DirectX12 has never performed well in Frostbite. It's kind of unclear why, DICE clearly has talented software engineers so you would think that the DX12 implementation would be solid, especially given it's required for some of the flagship visual effects.

 

This is why I patreon.

Are there any Vulkan games that are doing ray tracing yet? Would be interesting to see the results in a different API.

I'm still happy with my upgrade to a 2080ti coming from a Radeon 7950, however I'm interested to see H's review on the Radeon 7. I max out my 2080ti's VRAM running in VR and 4K, so it'll be interesting to see if the Radeon 7's increased memory gives it any advantages and helps narrow the gap at higher resolutions and VR.

 

Maybe if they offloaded the RT to the CPU it would be faster :b

 

Excellent review and I like many others can't believe the amount of time you have spent bench marking this title.

 

Meh, sure the launch was a bit rushed and it's not quite fully ready for widespread implementation, especially in multiplayer, but Ray Tracing had to start somewhere, and frankly I don't think many people ever expected it to arrive as soon as it did. Ray Tracing isn't going to go away and will likely be fully functional on most systems possibly a lot sooner than it would have if Nvidia hadn't rushed things. The good thing is it looks like Nvidia is offering a really solid product in the 2060 at a very fair price. AMD is doing the same with their new releases. Thankfully we still have some solid choices and the competition between the two is healthy which is good for the consumer.

 

Not DX 12, but dices ability to use it properly, or better yet, code it properly. They have had issues with DX12 since day one. Yet there are many devs/games out there that have no issues with DX12. It really seems as though DX12 is an after thought with both BF1 and BFV. Then you throw in Ray Tracing and it gets worse.

 

I see people saying this a lot. Please bear in mind this is *not* full scene ray tracing. The engine is using rasterization for almost everything and ray tracing only some elements and lighting effects.

We are still many years away from full-scene ray tracing replacing rasterizing as the primary technology driving visuals in games.

 

2060 is not a fair price

 

So the 9900k was pretty much within the margin of error. I wonder if people OC their cards which is why they got a higher #?

 

For your best BFV multiplayer experience, chunk NVIDIA Ray Tracing in the trash bin, and run DX11. The RTX is a lie.​

Oof, That's gonna hurt some feelz from those who grabbed an i9-9900k and an rtx card. lolz

 

You recorded your framerate/frametime and CPU(per thread)/GPU load graphs on each run? I've seen in some previous RTX/GTX reviews you published them. Could you please share this data? It's valuable to get the sense of how bad are these stutters/pauses. BTW when measuring the GPU load - is it possible to detect somehow if RT/Tensor cores are loaded?

 

With the Windows settings tweaked for gaming you won't get any stutter at all. With the right tweaks to Windows it's possible to gain almost twice your fps.
I went from 130-140 fps to 230-240 fps with a few Windows/regedit tweaks and a config. Same graphics settings but gained 100 fps and stutter was completely gone.

 

I would really love seeing a test with Windows set up for BFV to give it more resources and see what kind of fps difference you'll see between the 7700k and 9900k. Obviously at that point you'll be gpu limited, but with lowered graphics it could be interesting.

 

Could you point us in the direction of these tweaks, as i am interested in looking into them, or to see if there any other tweaks i need to do that I haven't already done. I am sure there are others who would be interested in them as well.

Thanks!

 

Please share with the rest of us.

 

If only the [Hard] would have tested BFV using the Core i9-9990XE (available by auction only, when it becomes available if at all).

 

Yeah I am fully aware of that. This is exactly what expected years ago when I had a discussion with some gaming friends friends about ray tracing, that it basically would start out being partially implemented and go from there, kind of similar to how Wolfenstein 3D & Doom started out not being true/full 3D. The fact that we have something already is awesome, sort of a silver lining from the death of Moore's Law. They can't get much more raw FPS performance anymore, so now they are upping the quality. This is a great thing for gaming.

 

Yep, DICE doesn't do DX12 very well. I suspect NV went with them because they were the first dev that would at least try DXR within a game. I like how BFV looks, but it's pretty crappy for what it is at this moment. I suspect we'll see much better implementation in the future from better DX12 game developers.


IMO, people who judge DXR from the DICE implementation are very short-sighted.

P.S. How bout that great bang for the buck VII card?