Article World of Tanks Implements Ray Traced Shadows for ANY DX11 GPU

[Stuka87]

Meant to post this yesterday, but I forgot. The video link below goes through how the implemented it, and its an interesting watch. For anybody wondering why its only shadows, its because each tank has in excess of 50k polygons, and you can have up to 50 tanks on screen at any given time. Currently no hardware exists that can handle that kind of load with RT reflections/GI. Wargaming has always tried to make their engine run on even low end hardware, and not require special hardware.

However, with how they implemented the RT Shadows, with help from Intel, it can run on any GPU. I play the game a lot, so I downloaded their test client, with the game settings at Ultra, and RT Shadows at HIGH, my system with a 4.5GHz 4690K and an RX480 4GB ran it really quite well at 1080P. It did drop below 60fps a few times. As a note, this does not use any specialized RT hardware, even if the GPU has it. This uses a combination of CPU and GPU to handle the calculations, and they use only 1 ray per pixel, with a custom dithering filter that they came up with which is shown in the video.

Details: https://worldoftanks.com/en/news/general-news/ray-tracing/

Developer Video:

 

[kondziowy]

RT on ultra vs off cuts performance exactly in half on my Vega64. Visual difference is of course very small unless you stop and stare at shadows. In normal gameplay you notice only performance drop. And I think that normal non-raytraced shadows look really bad for today's standards.

My card boosts 150MHz higher on RT ultra vs RT high which is weird. I never see so high clocks in any other game. It usually happens when my gpu is not fully utilised(low utilisation cause high clocks), but here it uses max power limit in both RT ultra and RT high. Weird.

Fun to run once and turn off later because we are still 19 years too early for fully ray traced games. Bring that Max Payne 1 now, Nvidia Lightspeed Studios

And BTW it uses 9% of Ryzen 2600. CPU is almost idling.

 

[Stuka87]

Yeah, I don't think I will play with it turned on, unless I end up with a faster GPU at some point. I didn't run with RT on Ultra, just on High. The CPU load is all on the front end, once the volume map is created for the tank, the CPU goes back to normal utilization. Which in this demo is very minimal as its not regular game play.

BUT, I think its cool from a technology aspect. Works on any DX11 GPU, and the shadows are being placed off some complex objects. I STILL stand by my thought that RT for regular use is quite a ways off. But the technology is cool when its done in an open way that can benefit any player with any brand GPU.

 

[Dribble]

Some charts: https://wccftech.com/world-of-tanks...d-polaris-vega-navi-pascal-and-turing-tested/

Hits performance pretty hard for fairly minor improvements, that said it's a game about tanks so you don't need 144fps to aim. If you've got a sufficiently powerful gpu might as well turn it on.

 

[mopardude87]

Its a nice attempt for sure, not needing nothing to insane to max out 1080p and hold a constant 60. Maybe more games will go this way and you won't need a 2080ti in something like BF5 to hold 60+ at 1080p. Sure the RT cores could help but they won't be a absolutely needed thing.

As mentioned in the video, current hardware isn't going to do much of anything right now. A balance between graphics and the needed power to run it is always nice to consider. Even if they added reflections as well and used RTX anything but the 2080ti will tank anyways. No pun intended.

Something tells me this will be a big thing before and after next gen consoles drop. Still kind of sad the game is more or less still single core base,i would welcome a feature or technology to ravage at least 4 of the threads constantly. It's 2019,most gamers have to be on 4 threads by now.

 

[Stuka87]

Its a nice attempt for sure, not needing nothing to insane to max out 1080p and hold a constant 60. Maybe more games will go this way and you won't need a 2080ti in something like BF5 to hold 60+ at 1080p. Sure the RT cores could help but they won't be a absolutely needed thing.

As mentioned in the video, current hardware isn't going to do much of anything right now. A balance between graphics and the needed power to run it is always nice to consider. Even if they added reflections as well and used RTX anything but the 2080ti will tank anyways. No pun intended.

Something tells me this will be a big thing before and after next gen consoles drop. Still kind of sad the game is more or less still single core base,i would welcome a feature or technology to ravage at least 4 of the threads constantly. It's 2019,most gamers have to be on 4 threads by now.

The game will use multi-cores now. That was added late last year.

 

[DeathReborn]

I wonder if this would break 10FPS @ 1080p on an Intel GPU.

 

[Stuka87]

I wonder if this would break 10FPS @ 1080p on an Intel GPU.

Integrated, unlikely. (It is playable on standard low settings). BUT, Intel has discrete GPU's coming out. I am sure this was on their mind when developing this with WG.

 

[Muhammed]

It uses RT shadows in a very limited way, only for healthy tanks, damaged tanks and all the other objects and environments don't cast RT shadows. Despite this it cuts fps by 55% on NVIDIA cards and by 75% on AMD cards.

RTX games achieve full RT Global Lighting and Shadows, as well as scene wide reflections and Shadows. So RTX cores really do help accelerate what's possible RT wise.

 

[Stuka87]

It uses RT shadows in a very limited way, only for healthy tanks, damaged tanks and all the other objects and environments don't cast RT shadows. Despite this it cuts fps by 55% on NVIDIA cards and by 75% on AMD cards.

RTX games achieve full RT Global Lighting and Shadows, as well as scene wide reflections and Shadows. So RTX cores really do help accelerate what's possible RT wise.

First, there is no concept of "damaged" tanks using different models in WoT. Any tank that isn't dead has RT shadows.

Second, your percentages are wildly wrong. 5700XT drops by 41.4%, and 2060S drops by 36.7%. No where even close to 75%?!

Third, if you watched the video, you would know that no GPU on earth could run WoT with full ray tracing, or even with just reflections. The average tank has 50K+ polygons, the developers state we are a LONG ways off before we have games that can do full RT.

 

[DooKey]

LOL. Specialized hardware is needed for reasonable RT. We'll be there sometime...just not today.

In regards to the lame comment about in an open way...DXR is an open way. Sorry that only one company accelerates DXR with hardware at this point.

 

[soresu]

Third, if you watched the video, you would know that no GPU on earth could run WoT with full ray tracing, or even with just reflections. The average tank has 50K+ polygons, the developers state we are a LONG ways off before we have games that can do full RT.

Makes me wonder if the recent work to use ML to accelerate ray/path tracing will find its way into the real time domain.

Look up "Path Guiding", "Neural Importance Sampling" and "Zero Variance Sampling Schemes" to get an idea of what I'm talking about.

 

[soresu]

In regards to the lame comment about in an open way...DXR is an open way. Sorry that only one company accelerates DXR with hardware at this point.

Vulkan is truly open and free to implement on whatever platform you can stick it onto, there seems to even be efforts to make the certification process less money bound too, though I may have read into that incorrectly.

DXR however is part of DirectX, and therefore proprietary to Microsoft platforms - it is only open in so much as the spec is available openly, which you would expect if they have any interest in proliferating support for it with hardware and software partners.

 

[mopardude87]

The game will use multi-cores now. That was added late last year.

I saw a bit more usage but nothing extreme. Looked like the last time i played it still used the majority of a single core with a bit of a second core being used of course. I think they processed some of the sound or something to another core? It's been a minute since i gave WOT any real attention right till this RT update.

 

[Muhammed]

First, there is no concept of "damaged" tanks using different models in WoT. Any tank that isn't dead has RT shadows.

Check the video again, ONLY healthy tanks cast RT shadows. Damaged ones don't.

Second, your percentages are wildly wrong. 5700XT drops by 41.4%, and 2060S drops by 36.7%. No where even close to 75%?!

What? You are bad at math too?

5700XT no RT: 203
5700XT RT: 119

70% drop in fps

World of Tanks enCore RT Demo Benchmarked - Polaris, Vega, Navi, Pascal, And Turing Tested

The team over at Wargaming is bringing the much-anticipated update to their World of Tanks game engine with the move to enCore RT. Wargaming has partnered with Intel and is using their Embree library (part of OneAPI) to bring ray traced shadows to World of Tanks. This is not full scene ray...

wccftech.com

Third, if you watched the video, you would know that no GPU on earth could run WoT with full ray tracing, or even with just reflections. The average tank has 50K+ polygons, the developers state we are a LONG ways off before we have games that can do full RT.

Yes we can, we've done whole scene reflections before in games like Battlefield V with much more complex models, but we need hardware RT for that. Not some lame software one.

 

[DXDiag]

First, there is no concept of "damaged" tanks using different models in WoT. Any tank that isn't dead has RT shadows.

In the video it clearly states that intact tanks alone cast ray traced shadows, any other object or tank don't use ray traced shadows. The official game website also mentions this:

As the first step we chose to include Ray Tracing shadows for the "main actors" of our game - tanks, the intact vehicles (not destroyed)”; their smallest details will give super-realistic shadows when the sun hits them.

Ray Tracing Explained

World of Tanks will soon support one of the most cutting-edge rendering technologies.

worldoftanks.eu

So in essence that implementation, really enforces the need for RTX cores/hardware ray tracing, because with them you do get comparatively massive increases in quality for the same fps hit. For example, in Tomb Raider, the whole scene cast ray traced shadows and it runs well on an RTX 2070 Super for example. In Metro the whole scene cast ray traced Global Illumination as well as Ambient Shadows, which -together- are basically the heaviest form of Ray Tracing, and still maintains good fps on one GPU.

 

[Hitman928]

5700XT no RT: 203
5700XT RT: 119

70% drop in fps

Oh the math call-outs when you're the one doing the math wrong, an internet classic. Let me ask you a sanity check question, what would a 50% drop in performance (half the frame rate) be and why is a 50% drop bigger than your supposed 70% drop?

 

[Muhammed]

Oh the math call-outs when you're the one doing the math wrong, an internet classic. Let me ask you a sanity check question, what would a 50% drop in performance (half the frame rate) be and why is a 50% drop bigger than your supposed 70% drop?

Lets rephrase then, the no RT fps is 70% faster than the RT fps. Satisfied now?
The drop in this case would be 58% to be exact. But we have to be pedantic about it, don't we?

At least have the audacity to correct his laughable 41% numbers.

 

[Hitman928]

Lets rephrase then, the no RT fps is 70% faster than the RT fps. Satisfied now?
The drop in this case would be 58% to be exact. But we have to be pedantic about it, don't we?

At least have the audacity to correct his laughable 41% numbers.

Just stop, you're making it worse. He was correct, you were wrong. Let's move on.

Edit: And it's not being pedantic to correct a statement that says it's a 75% drop and a 55% drop when in reality it's a 41.4% and 36.7% drop. That's a huge difference.

 

[Muhammed]

Just stop, you're making it worse. He was correct, you were wrong. Let's move on.

Oh you are bad at math too? Good to know!


5700XT no RT: 203
5700XT RT: 119


119 is 58% of 203 not 41%!
203 is 70% faster than 119!

God! didn't know math was so hard on so many people!

Edit: And it's not being pedantic to correct a statement that says it's a 75% drop and a 55% drop when in reality it's a 41.4% and 36.7% drop. That's a huge difference.

That's not how percentages work! The drop is calculated by the amount of fps you lose when going from the lowest amount to the highest amount. So it's 70% drop here. You lose 70% more fps going from no RT to RT.

 

[Hitman928]

Oh you are bad at math too? Good to know!


5700XT no RT: 203
5700XT RT: 119


119 is 58% of 203 not 41%!
203 is 70% faster than 119!

God! didn't know math was so hard on so many people!

You're really digging yourself deep aren't you?

Yes, 119 is 58.6% of 203, that means it is a 100 - 58.6 = 41.4% drop.

Alternatively, you can do 100 * (119 - 203) / 203 = -41.4% change.

Yes 203 is 70.6% faster than 119 -> 100 * (203 - 119) / 119 = 70.6%

But that's not what you said initially which is why you were corrected. It makes a huge difference. You easily could have accepted the mistake and moved on but you just keep digging deeper trying to contort some way into making it seem like you were right all along while throwing insults at others who actually did have it right all along. Good luck with that.

 

[BFG10K]

What? You are bad at math too?

5700XT no RT: 203
5700XT RT: 119

70% drop in fps

Comedy gold.

Oh you are bad at math too? Good to know!
5700XT no RT: 203
5700XT RT: 119
119 is 58% of 203 not 41%!
203 is 70% faster than 119!
God! didn't know math was so hard on so many people!

You don't understand how percentages work. Hitman is correct. You're repeatedly wrong even after your back-pedal.

 

[Muhammed]

But that's not what you said initially which is why you were corrected. It makes a huge difference. You easily could have accepted the mistake and moved on but you just keep digging deeper trying to contort some way into making it seem like you were right all along while throwing insults at others who actually did have it right all along. Good luck with that.

I stand by my statements, its you guys who don't understand how percentages work during benchmarking or when comparing fps .. It's STILL is 70% difference! Go read some books before you babble about trivia.

Here is how people do compare fps numbers:

 

[Hitman928]

I stand by my statements, its you guys who don't understand how percentages work during benchmarking or when comparing fps .. It's STILL is 70% difference! Go read some books before you babble about trivia.

Here is how people do compare fps numbers:

So your argument is that by your third try you got it partially right so you were never wrong?

 

[Stuka87]

In the video it clearly states that intact tanks alone cast ray traced shadows, any other object or tank don't use ray traced shadows. The official game website also mentions this:

Ray Tracing Explained

World of Tanks will soon support one of the most cutting-edge rendering technologies.

worldoftanks.eu

So in essence that implementation, really enforces the need for RTX cores/hardware ray tracing, because with them you do get comparatively massive increases in quality for the same fps hit. For example, in Tomb Raider, the whole scene cast ray traced shadows and it runs well on an RTX 2070 Super for example. In Metro the whole scene cast ray traced Global Illumination as well as Ambient Shadows, which -together- are basically the heaviest form of Ray Tracing, and still maintains good fps on one GPU.

I said destroyed tanks would not have RT Shadows. However, "damaged" tanks don't exist in game. Either they are fully intact, or destroyed. That was my point.

The difference between TR/Metro compared to WoT is the complexity of the objects. In WoT you can have up to 60 tanks on screen at one time, each having 50K+ polygons. Thats significantly more than either of the other games have on screen.