3DMARK Port Royal, the First Dedicated Ray Tracing Benchmark, has Launched

Oh I read it wrong, I see it now. Well no point in getting it since I don't have RTX and don't need to pay for a slideshow

Ah go on. It will be a pretty slideshow. TBH part of me wanted to buy it, despite not having an RTX card here either.

Just how badly do older GPUs run on this?

tenor.gif
 
So I updated everything and ran bare bones (you know benchmark style) and well nothing! Not a sausage.

VieWWbM.jpg


Every other bench works fine just not Port Royal, has anyone else with a non RTX card managed it?
 
Ah go on. It will be a pretty slideshow. TBH part of me wanted to buy it, despite not having an RTX card here either.

Just how badly do older GPUs run on this?

tenor.gif


GPUs without hardware DXR support won't work. There is a fallback emulation layer but must be implemented by the developer and has already been deprecated..
 
GPUs without hardware DXR support won't work. There is a fallback emulation layer but must be implemented by the developer and has already been deprecated..

Well that's disappointing, UL specified that "Port Royal will run on any graphics card with drivers that support DirectX Raytracing", which implies that any DX12 card will be supported if it offers DXR compute support in drivers. That said Nvidia may not be offering that support in their consumer drivers.

It kinda defeats the purpose of the benchmark for comparison purposes if the direct compute option cannot be used. You'd think that Nvidia would want the compute option to be available to showcase how bad Pascal is with DXR. (So that Jensen can showcase a new graph at Computex ti say Pascal isn't great anymore)

(From 3DMARK Technical Guide)
https://s3.amazonaws.com/download-aws.futuremark.com/3dmark-technical-guide.pdf

"At the launch of Port Royal, the only graphics cards with drivers that support Microsoft DirectX Raytracing are the NVIDIA GeForce RTX series, Quadro RTX series, and the NVIDIA TITAN RTX and TITAN V. More cards are expected to get DirectX Raytracing support in 2019"

Ugh.
 
Well that's disappointing, UL specified that "Port Royal will run on any graphics card with drivers that support DirectX Raytracing", which implies that any DX12 card will be supported if it offers DXR compute support in drivers. That said Nvidia may not be offering that support in their consumer drivers.

It kinda defeats the purpose of the benchmark for comparison purposes if the direct compute option cannot be used. You'd think that Nvidia would want the compute option to be available to showcase how bad Pascal is with DXR. (So that Jensen can showcase a new graph at Computex ti say Pascal isn't great anymore)

(From 3DMARK Technical Guide)
https://s3.amazonaws.com/download-aws.futuremark.com/3dmark-technical-guide.pdf

"At the launch of Port Royal, the only graphics cards with drivers that support Microsoft DirectX Raytracing are the NVIDIA GeForce RTX series, Quadro RTX series, and the NVIDIA TITAN RTX and TITAN V. More cards are expected to get DirectX Raytracing support in 2019"

Ugh.


Not really, there's no point comparing it. It's like saying lets compare tesselation on a card that didn't support tesselation. The only thing you can compare is visual quality vs performance on old cards, but they will be different implementations.


This benchmark is to bench ray tracing cards aginst each other, there's not many yet but that will change.
 
Not really, there's no point comparing it. It's like saying lets compare tesselation on a card that didn't support tesselation. The only thing you can compare is visual quality vs performance on old cards, but they will be different implementations.


This benchmark is to bench ray tracing cards against each other, there's not many yet but that will change.

Aye, I get it. I was just hoping that this would be a little more useful as a benchmark at launch.

As much as we know that any compute mode would be awful, it's an interesting data point and conveys the message that Ray Tracing is awesome but needs bespoke hardware to run well.

If I remember correctly Nvidia's RTX Star Wars Demo could be used on Pascal, just at stupid low framerates. I'd like to see a similar test from a 3rd party to put the whole "Nvidia RT cores do nothing" idiocy to bed.

Right now this benchmark is only really good for telling us that the 2060<2070<2080<2080 Ti<Titan RTX, which isn't exactly enlightening.
 
I don't see the point in benchmarking something like this with a card thats not design to use it if we knowingly expect low fps.

Had it been a hack/config which we could prove that RT is playable without RTX then fair enough. Some times, benchmarking something "just because we can" isn't always beneficial to the website and could be seen as click bait/timewasting of viewers time.

just my 2cents
 
I don't see the point in benchmarking something like this with a card thats not design to use it if we knowingly expect low fps.

Had it been a hack/config which we could prove that RT is playable without RTX then fair enough. Some times, benchmarking something "just because we can" isn't always beneficial to the website and could be seen as click bait/timewasting of viewers time.

just my 2cents

I get what you mean, but these days false information spreads so fast, so even smaller data points like this are important.

Even just a few weeks ago lots of places were going wild talking about Vega doing RTX because somebody enabled the setting. The setting didn't work, obviously (it just continued to use screen space reflections), but that didn't stop the rumour from spreading (and worse still some websites picking it up).

Similar information has started spreading recently as well, claiming that the Titan V offers the same performance as the RTX 2080 Ti in Battlefield V. Ray tracing works with that GPU, but the performance is not the same.

Even results that are as expected are useful. It sickens me how quickly controversy bait rumours spread sometimes.
 
Yeah the problem is this relies on DXR drivers and no GPU vendor is going to spend resources to make drivers for a card just to shown how bad it is at using certain technology, UL would have to do their own implementations which I guess isn't worth their time and isn't their goal.

To be honest, I think these rumours will die down naturally with time as the technology becomes more widely used and better understood.
 
Yeah the problem is this relies on DXR drivers and no GPU vendor is going to spend resources to make drivers for a card just to shown how bad it is at using certain technology, UL would have to do their own implementations which I guess isn't worth their time and isn't their goal.

To be honest, I think these rumours will die down naturally with time as the technology becomes more widely used and better understood.

Yeah, things will get better in time. I just don't like how a lot of the tech community works, so much conflict bait these days, especially from YouTubers (not TTL)

Score 17 621

NVIDIA Titan RTX(2x) @2055/4100

Intel Core i7-6950X Processor @4.47

Graphics Score 17 621

Drivers 417.35

https://www.3dmark.com/3dm/32198486?

I was wondering when Kaapstad would show up with some fresh results. Nice.

giphy.gif
 
Back
Top