-
Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
-
We have made minor adjustments to how the search bar works on ResetEra. You can read about the changes here.
NVIDIA RTX 20 Series Reviews
- Thread starter Mecha Meister
- Start date
DICE did their earlier RTX/DXR work for BFV on Titan Vs before final RTX hardware. DF said so during the early RTX coverage at Gamescom 2018. Kinda makes sense that most of that carried over to the final game. Same for developers like Remedy, they worked with Titan Vs when doing the Northlight engine tech demo - that work probably carried over to what they're doing for Control.
I think it didn't use the Tensor cores, although it sounds like those might be put to use with DLSS in the (potentially near) future.
It's not "graphics cards". It's a Titan V which has officially supported RTX even longer than the 2xxx cards. Performance is also not really OK with these minimum FPS numbers. This is all on a $3000 card.
The RTX code in BFV was all coded for Volta. That's why they demoed it, it wasn't "1 week's work", it was 1 week port to Turing.
Turing's RTX side is an optimized Volta mod without HBM2. Probably running on Volta's tensor cores.
Last edited:
There is no "RTX code", DXR is supported on both Volta and Turing, the latter has RT cores which accelerate one specific part of RT computations which isn't necessarily the one which the code's performance is limited by. That being said, I wouldn't really trust this Reddit post until there will be a proper comparison by a reputable source as "XXX marketing at its best." (c) regawdless these days include such FUD shite from competition all the time.The RTX code in BFV was all coded for Volta. That's why they demoed it, it wasn't "1 week's work", it was 1 week port to Turing.
Turing's RTX side is an optimized Volta mod without HBM2. Probably running on Volta's tensor cores.
outside of battlefront 2, frostbite 3 titles have GREAT taa so its probably not going to be a good look for DLSS. we dont have information that suggests DLSS offers any benefit over straight 1800p upscaled to 4k
I haven't heard much about recent 2000 series card failures, but man... PC forums that got offended by pricing are still in full-on dogpile mode about it. Some of the enthusiast sites I generally enjoy are in full hive mind mode about these cards. Lots of Space Invaders GIFs and similar from people who never even purchased them.
Nothing new but a pretty interesting video I thought about the difference in how Pascal and Turing handle INT and FP workloads and how that would translate performance wise in some current (and future) games
I have a 580, and am interested in upgrading to a 2060. Is that a good move? Am I gonna miss the 8gb the 580 has?
OP
OP
I don't trust videos/channels like this.
Unless you're seeing a video side-by-side showcasing live gameplay with a frame rate counter like MSI Afterburner and things like clock speeds etc on the on-screen display don't trust these kind of videos, unless they're from a reputable source.
Anyone can put numbers together on a slide and calculate potential percentage differences in performance.
Check multiple sources for performance in the event you come across videos like this.
The video didnt show percentage differences. I have no reason to doubt them.
I would go to 2070 in your place. Going with less VRAM than you have is never a good idea, and 6GBs of 2060 is a real issue which you will likely run into quite soon.
Finally played some DXR on ultra on bfv mp. The performance is pretty much half. It does look good. The explosions are very much a step up and there is a lot of depth to them. The upgrade in explosions was so great that it really made me want well done HDR on the pc. On the swamp map, the water reflections are immediately noticeable and improve the overall IQ to a new level.
Performance hit is nasty. Dipping into the 50's on 3440x1440p on a 2080ti and 9900k. With DXR it's usually dips down into the 90's at worst.
Overall, there needs to be a lot more refinement in regards to the performance penalty.
Performance hit is nasty. Dipping into the 50's on 3440x1440p on a 2080ti and 9900k. With DXR it's usually dips down into the 90's at worst.
Overall, there needs to be a lot more refinement in regards to the performance penalty.
What Quality Setting did you have on DXR? High/Middle/Low? I heard that Setting the Textures from Ultra to High improves the Frame rate when running dxr.
I am curious what my 2080 msi Trio will manage in 1440p with dxr on. I am planning to set textures to high and dxr to medium. Looking at the digital foundry 2060 1080 60fps battlefield v Video i am thinking that this should give me a good result
Last edited:
You can actually check your expected framerates on BFV without actually having to play the game to test. The images that come on when you are in the menus are all rendered in real time and present a range of best and worse case scenarios for framerates with your settings. For example, the close-up image of a soldier holding a gun shows you the lowest expected framerate (for me with DXR on ultra at 3440x1440 on a 2070 this comes out at about 40fps or slightly lower), the plane image presents a good version of your average framerate (50fps on ultra) and the others present a mix of best and worse case. Put the framerate counter on and watch it run through the images.
Oh, change your post processing down to high and texture down to high. The texture setting on ultra seems to have an odd problem. You should be getting better framerates
Well over 60. I can get 60 fps locked with a 2070 at 1440p with DXR on medium. There's a big jump between medium and high as high raytraces your gun and medium doesn't. Still looks amazing though as all the other reflective surfaces are raytraced.
Thanks for the tip! I'll use that to tune it. I think hard forum did a review in IQ so i'll compare the IQ hit by going to "high"
Yeah I thought I'd be in the 80's which would work well with gysnc and x34p
You should get around 10 ish to 15fps more than I do so 60 should be okay in most scenarios. If not then drop internal rendering resolution down to 90 and you will be over 60 with ultra all the time. You will only notice a slight drop in clarity, if at all.
Also, my FPS is with a Ryzen 1600 so not exactly a great chip either.
Also, my FPS is with a Ryzen 1600 so not exactly a great chip either.
Good to know. I should have my new PC by the end of the week. I would love to get locked 60 with textures on high and dxr on high.
I'm updating now. I'll let you know.
Might take a while to download as it seems the servers are getting hammered.
So an upgrade from a 1060 to a 2060 is not that good? I'm concerned about 6GB VRAM only but at worst I feel I'll just have to play on medium textures or so, once new Gen games comes out.
For 1440p should be enough for at least 2 or 3 years I feel. Honestly a 2060 feels a better value than a 1070Ti which is also expensive on my country right now (not counting second hand which is something I would rather avoid)
For 1440p should be enough for at least 2 or 3 years I feel. Honestly a 2060 feels a better value than a 1070Ti which is also expensive on my country right now (not counting second hand which is something I would rather avoid)
Just updated and looked through the menus...no DLSS option.
I've definitely gotten steady 60fps basically locked in most modern titles just by taking quality settings down from Ultra to High. With Ray tracing though it's looking to be 1440p if Battlefield 5 is any indication.
oh god, the RTX 2060 is 530 USD in Sweden. because of course it is, don't know why I expected otherwise.
Dice never stated it was coming in this patch. They gave no ETA afaik
Someone pointed out that it's still being worked on and updated, albeit privately. Last updated yesterday even (find nvidia_dxr build)... https://steamdb.info/app/750920/depots/
My experiences with DXR in BFV pretty much match yours. I think it looks good, albeit the use in-game of raytracing is a little limited, but it is impressive where you see it in action. However the performance hit, even after patching, is extreme. I'm tried it out on a i9-7900X with a 2070, but only at 1920x1080... even at that resolution it's a stuttery mess, goes down into the 40's and is noticeably choppy.
Playing at 1440p ultra/DXR low with a 2070 and even though the framerate is fine most of the time, it can be very stuttery.
Using "perfoverlay.drawgraph 1" I can very often see some pretty big CPU spikes (6700k @4.5) when things get busy (GPU seems fine)
Using "perfoverlay.drawgraph 1" I can very often see some pretty big CPU spikes (6700k @4.5) when things get busy (GPU seems fine)
Anyone have recent failure issues here?
I follow a couple of tech forums and I'm starting to wonder about trolls and such. There are people claiming to be on their 3rd or 4th card at this point. I know there were some issues without a doubt, but when you get to 3-4 cards and are still looking for a "working" card.. I have to wonder if you're just stirring up shit or you have other issues.
I follow a couple of tech forums and I'm starting to wonder about trolls and such. There are people claiming to be on their 3rd or 4th card at this point. I know there were some issues without a doubt, but when you get to 3-4 cards and are still looking for a "working" card.. I have to wonder if you're just stirring up shit or you have other issues.
My 2080 Ti FE still going strong, and its also one with Micron memory.
Since the 2060 is roughly a 1070 ti with RTX and dlss support I think it's a good upgrade. Really no reason to buy a 1070 ti anymore.
All faulty cards were swapped for non-faulty ones if they failed. There wasn't a lot of them in the wild anyway. These people with a 3rd or 4th card are doing something wrong which isn't related to the cards they have per se.
This can't be possible because everyone on the internet is an IT expert, with borderline genius PC skills. /s
But yea.. I mean I had a 2080 Ti die space invaders style, but I agree once you get into 3-4 cards then you need to start reassessing things (or should have already). Some of the usually better tech forums became an echo chamber of nasty once the prices were announced, so there's lots of dogpile going on.
I wouldn't go as far to say every case of this is user error. I once went through 4 returns of the same SSD drive before the thing eventually stopped being sold. Sometimes you get these refurbished returns that should have stayed in the RMA bin.
