Return to Level1Techs.com

Ray tracing on AMD Vega 56 NEON NOIR Crytek DEMO

gpu

#1

The demo is not yet public. It runs on a modified cryengine 5 featuring Raytracing made by Crytek.

For the demo a Vega 56 was used. It runs at 4K around 30FPS.


#2

Sweet, just wanted to create the same thread. It’s nice to see raytracing on AMD too!


#3

Just needs compute, I imagine?
The faster ray tracing is open source, the better. Then we can tackle VR again


#4

I am gonna die of laughter if this turns even 75% as good as RTX. What is even more interesting is this should just work in Linux. Cry Engine games tend to work under Wine/DXVK.

I can’t wait to get my hands on this tech demo to see if it’s really so good.


#5

So basically this topic will be the new gameworks topic for few generations


#6

Now that AMD is on board Raytracing is a very very very very very very very very very very VERY very very good thing.

Green brand bad.


#7

Ray tracing has always been a good thing. It’s also not been restricted by hardware vendor.

nVidia offers fast ray tracing but only if you join their ecosystem.

Also AMD has been on board for years, but only just announced real-time ray tracing with Radeon Rays 2.0 back in November. Still open source.


#8

com


#9

:rofl:


#10

It is nice to hear that one doesn’t have to buy Nvidia right now or in the future then …

and this is coming from a totally Green Team guy since I got into pcs.


#11

Im still confused why GPU’s have to be segmented so much. It’s all parallel math and nvidia has Cuda. AMD has OpenCL.

For the love of god why isn’t there just math and lets all the number roam run in army lines.

Lock ins are so anti consumer. What happens when Intel GPUs kick ass. Another compute standard Ispace ?

CPU wise we have x64, AMD’s standard. Why the hell GPU’s are not the same is beyond me. Non compatible software is stupid.

It is probably why apples Ipad beats laptops.


#12

RayTracing is just calculation. Done well it should run on everything.
Nvidias version of RayTracing works on Nvidia cards alone cause the algorithm is using their tensor cores, but the ray tracing is just a calculation process of lighting.
I’m not even slightly surprised it works on AMD…


#13

Because it would be slow. GPUs change instruction sets all the time to get the most out of the silicon. Imagine if GPUs were as locked into a single instruction set as processors are. Goodbye progress


#14

Im sorry Nvidia and AMD are making separate progress over time on GPU’s. While Intel and AMD focused on making an instruction set the best.

When intel enter the GPU market do you expect GPU’s to be better off when intel sponsors this game and AMD another and Nvidia makes a custom tee-bag-works API and funds that ? Kreestuh would need that GPU :slight_smile:


#15

I just wish I wish I could try this out myself.


#16

tens (hundreds?) of instruction sets.

Instructions are the middle ground between the programm and the atomic machine instructions. And as there are many ways to achive the same thing, there are many instruction sets.

This, plus the driver/low level API handles some stuff.

They better make sure to support DX12 completly and get OpenCL/GL and Vulkan right.


#17

To be honest, most of these lock ins are done by nvidia. AMD uses open standarda already. At this point, it is up to the consumer to decide which vendor to pick…


#18

AMD is forced to go open source because of the poor market share. This is their only hope of game developers accepting their tech.

The most impressive thing here is that AMD seems to bet on the right tech! Mantel > Vulkan, Freesync and Async Compute. I mean in the end Nvidia had to bow low to all of them.

Raytracing was one big exception where it was reversed.


#19

That is to be seen. PhysX did not go far either.


#20

We were talking about instruction sets. There is no reason for them to use different APIs.

What are you even on about? All three vendors support OpenCL and we can already see intel contributing to SYCL, OpenCL and other open standards. Nobody is introducing a third API.