• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

3DMark DirectX Raytracing Feature Test & Port Royal Bench Thread.

I thought 10 series cards could do RT too? They'll be at the bottom of the table but still can take part :D

Other cards can do ray tracing, yet, that is if game engine offloads it to the cpu, so essentially everything bellow RTX cards will be CPU tests, and not GPU.
What I would love 3D Mark devs to do is (since its their day job to create benchmarks) to code an engine with various API, like DX12/Vulkan and DX11, so if the card like 1080Ti runs best on DX11, benchmark runs in DX11 mode, if the card like Vega 64 runs that bench best in DX12/Vulkan, it will be run in that mode automatically. nVidia and AMD do driver optimisations for those benchmarks anyway, so they could leave some flags for each GPU of theirs on what API to use. Same with all these RTX, async shaders etc, features. If a GPU supports RTX, let it run, if it doesn't offload it to CPU or other alternative, if GPU supports async shaders, let it use it, if it doesn't it doesn't. That way benchmark would actually test the cards completely to their fullest potential, instead now we get half assed shortcutted benchmark which ignores many GPU features to cater the industry leader however backwards they might be. This seems to be a quick PR stunt for nvidia, that hey look we have engine built for 3 cards in the market who can make use of it, and hey look, at the launch event we have certain product placement.
I understand game devs, who ignore certain low volume features in their game engines, because it doesn't make financial sense to support everything, but for benchmark devs, who push latest hardware to the limit, it doesn't matter if one manufacturer does not support certain features while the other does, they still gonna get the sales
 
Other cards can do ray tracing, yet, that is if game engine offloads it to the cpu, so essentially everything bellow RTX cards will be CPU tests, and not GPU.
What I would love 3D Mark devs to do is (since its their day job to create benchmarks) to code an engine with various API, like DX12/Vulkan and DX11, so if the card like 1080Ti runs best on DX11, benchmark runs in DX11 mode, if the card like Vega 64 runs that bench best in DX12/Vulkan, it will be run in that mode automatically. nVidia and AMD do driver optimisations for those benchmarks anyway, so they could leave some flags for each GPU of theirs on what API to use. Same with all these RTX, async shaders etc, features. If a GPU supports RTX, let it run, if it doesn't offload it to CPU or other alternative, if GPU supports async shaders, let it use it, if it doesn't it doesn't. That way benchmark would actually test the cards completely to their fullest potential, instead now we get half assed shortcutted benchmark which ignores many GPU features to cater the industry leader however backwards they might be. This seems to be a quick PR stunt for nvidia, that hey look we have engine built for 3 cards in the market who can make use of it, and hey look, at the launch event we have certain product placement.
I understand game devs, who ignore certain low volume features in their game engines, because it doesn't make financial sense to support everything, but for benchmark devs, who push latest hardware to the limit, it doesn't matter if one manufacturer does not support certain features while the other does, they still gonna get the sales
It's difficult really as benchmarks are really used for comparing things(GPU based systems in this case) and the settings used need to be like for like otherwise it's not really a benchmark. You'd otherwise end up with benchmarks based on a particular generation or even model of GPU.
It'd be better if benchmarks just reflected the current gaming trends which then should (in theory) show how well a particular GPU and CPU combination would work with latest games
 
It's difficult really as benchmarks are really used for comparing things(GPU based systems in this case) and the settings used need to be like for like otherwise it's not really a benchmark. You then end up with benchmarks based on a particular generation or even model of GPU.
It'd be better if benchmarks just reflected the current gaming trends which then should (in theory) show how well a particular GPU and CPU combination would work with latest games.

Yeah unless you can assure identical output quality for any given render path it makes the benchmark less than useful.

Theoretically DX has a GPU based fallback path for Ray Tracing functions but not sure off the top of my head if that is comparable quality wise to the cards that have equivalent hardware just much slower or whether it also reduces feature level support to make it work at all.
 
It's difficult really as benchmarks are really used for comparing things(GPU based systems in this case) and the settings used need to be like for like otherwise it's not really a benchmark. You then end up with benchmarks based on a particular generation or even model of GPU.

Yeah, but as you go, features are being added, new technologies developed to improve performance. Though not in case of nvidia where every of their advancement halves the fps :D
Currently 3dmark benchmarks are based on generation of cards. You had DX10 only benchmark, then dx11 only, then dx12 only, then VR only, and now, it seems they are just targeting Ray Tracing.
 
Yeah, but as you go, features are being added, new technologies developed to improve performance. Though not in case of nvidia where every of their advancement halves the fps :D
Currently 3dmark benchmarks are based on generation of cards. You had DX10 only benchmark, then dx11 only, then dx12 only, then VR only, and now, it seems they are just targeting Ray Tracing.

I don't have a problem with a Ray Tracing only benchmark.

You don't go off road racing in a Formula 1 GP car do you.
 
Last edited:
Typical of some people on this forum to down play the importance of this new benchmark.

Thing to think about is, it is a Microsoft DXR based benchmark, not an NVidia RTX based benchmark, so in a couple of years when AMD get in on the ray tracing action, they will be able to run it. Hey they might even do well in it like they do in the multitude of other 3d mark benches.
 
Typical of some people on this forum to down play the importance of this new benchmark.

Thing to think about is, it is a Microsoft DXR based benchmark, not an NVidia RTX based benchmark, so in a couple of years when AMD get in on the ray tracing action, they will be able to run it. Hey they might even do well in it like they do in the multitude of other 3d mark benches.

Totally agree.

It will also give everyone an idea of how well Ray Tracking runs on current hardware and what is realistic to expect in games.
 
Typical of some people on this forum to down play the importance of this new benchmark.

Thing to think about is, it is a Microsoft DXR based benchmark, not an NVidia RTX based benchmark, so in a couple of years when AMD get in on the ray tracing action, they will be able to run it. Hey they might even do well in it like they do in the multitude of other 3d mark benches.

I am actually quite looking forward to it. If all current cards can run it as it's dxr based it should be half decent. Will be interesting to see just how good the Turing cards are with there dedicated hardware compared to the older gens. I wonder if AMD will bring out a driver for this as i am sure it was said in another thread that AMD don't have support for this yet.

At least overtaking would be easier with a bit of mud and a few trees to dodge.:D

Yea some races this year needed something like this to spice it up.
 
Not typical. But you have to question the motives, when you are releasing benchmark for 3 card models from same manufacturer, and release event is featuring one of the models of that manufacturer.
Are you interested to see how your non RTX system performs in ray tracing? I can tell you straight away: it performs exceptionally horrible. That's the reason Nvidia created new arch including dedicated cores to do ray tracing and that's why you all pay through your noses for RTX cards due to their massive die sizes because it was needed to house RT dedicated cores. There is a reason that until now ray tracing in real time was not used.
I never had any issues with other benchmarks because they were targeting a lot of cards.
This on the other hand seems like a lovely PR stunt for massively expensive RTX cards.
 
Not typical. But you have to question the motives, when you are releasing benchmark for 3 card models from same manufacturer, and release event is featuring one of the models of that manufacturer.
Are you interested to see how your non RTX system performs in ray tracing? I can tell you straight away: it performs exceptionally horrible. That's the reason Nvidia created new arch including dedicated cores to do ray tracing and that's why you all pay through your noses for RTX cards due to their massive die sizes because it was needed to house RT dedicated cores. There is a reason that until now ray tracing in real time was not used.
I never had any issues with other benchmarks because they were targeting a lot of cards.
This on the other hand seems like a lovely PR stunt for massively expensive RTX cards.


You do realise that this isn't made by NVidia?
It isn't using a NVidia only functions.
It is using the standard Microsoft DXR API

Yes being launched with the Galax HOF card demoing it is of course a PR stunt, but then isn't Time spy sponsored by Galax as well.
 
You do realise that this isn't made by NVidia?
It isn't using a NVidia only functions.
It is using the standard Microsoft DXR API

Yes being launched with the Galax HOF card demoing it is of course a PR stunt, but then isn't Time spy sponsored by Galax as well.

One of the reason 3d mark is no longer a benchmark. IIRC they did start compiling for Intel and had some very friendly Nvidia code.
 
https://www.guru3d.com/news-story/3...mark-new-trailerrelease-date-and-pricing.html


8 Jan 2019 for $2.99? Not bad!

Stunning ray tacing demo. Strange, these androids in suits with bow ties reminded me of Working Joe androids from Alien: Isolation. :)

https://benchmarks.ul.com/3dmark/buy-port-royal-upgrade

* To date, the only graphics cards with drivers that support Microsoft DirectX Raytracing are the NVIDIA RTX 2070, RTX 2080, RTX 2080 Ti, Titan RTX, Quadro RTX series, and Titan V. More cards are expected to get DirectX Raytracing support in 2019.

More cards in 2019 expected to be RTX 2050, RTX 2050 Ti, RTX 2060 and maybe 7nm RTX 3000 series too but Navi 10 RX 3000 series wont have Ray Tracing support so look like Portal Royal will not have DXR fallback support.
 
Last edited:
https://www.guru3d.com/news-story/3...mark-new-trailerrelease-date-and-pricing.html


8 Jan 2019 for $2.99? Not bad!

Stunning ray tacing demo. Strange, these androids in suits with bow ties reminded me of Working Joe androids from Alien: Isolation. :)

https://benchmarks.ul.com/3dmark/buy-port-royal-upgrade



More cards in 2019 expected to be RTX 2050, RTX 2050 Ti, RTX 2060 and maybe 7nm RTX 3000 series too but Navi 10 RX 3000 series wont have Ray Tracing support so look like Portal Royal will not have DXR fallback support.
Benchmark looks really cool, some nice reflections, creepy robots too!
 
Typical that some people on this forum pretend to think that 3D mark is an irrelevant representation graphical power.;)
 
Typical that some people on this forum pretend to think that 3D mark is an irrelevant representation graphical power.;)

Those same people used to think different when the HD7970 was on sale.

It will be interesting to see how the NVidia RTX cards run on this benchmark, I don't think they will do as well as some people are expecting.
 
Back
Top Bottom