Is VR dead?

  • Thread starter Deleted member 197115
  • Start date
AMD is going like gangbusters, but this is in server space and producing things like the upcoming 64 core chips that will be great for video rendering, but not taking top honors away for gaming.

I've yet to see anything showing that AMD is closing in. They don't seem to be getting the single core speed high enough to take the title of best gaming CPU. Financially they are going strong and they are expected to have 5nm process out in 2021 which is great. However Intel's 7nm will actually be denser than AMD's 5nm because of differences in how they define their processes.

What am I missing?
 
Last edited:
Why does it matter for users if it is AMS, Nvidia or Intel?
At the moment it isn't the hardware that is holding it back but the software poor use of the hardware.
Our sims are to a large extend still relaying on single core speed!
We all chase the last few fps and higher settings and think we will be happier if we achieve them.
This guy here looks to have fun and have likely more personally satisfaction with is setup than most of us have?
 
It doesn't matter what brand it is in theory. But the general landscape of lack of performance gains generation over generation is purely because the apparent saviours of PC gaming, namely Intel and Nvidia, have us exactly where they want us. It's crazy to look forward to new hardware from such companies that are completely happy to give us incremental upgrades for literally thousands of dollars. It's disgusting and I've had enough of it.

AMD has been in the wilderness for a long time and is the main reason why those companies are allowed to milk us for relatively negligible gains given the cost. It's time they get recognition for their Zen architecture and I am very very hopeful the same resurgence is coming to their GPU sector. From all reports, we will have a great battle to witness come the end of the year and that only benefits us.

Maybe it's about time game developers actually learn how to use modern hardware, instead of programming for single core clock speed. We have consumer CPU's available with 16 cores and you can bet the new consoles will be aiming to make the most of all of them. Perhaps when they are finally forced to come out of the dark ages, we will get better performance from clearly superior architecture.

I am not a fanboy of any company. But I detest the current practices from the giants in PC gaming hardware and welcome the day they meet their match, and then some.
 
If AMD takes the lead in performance they too will slow down progress. It would be crazy to compete with it own products.

But yes the sim software is way behind. Would love to see a Studio coming out with a sim that supports +8 cores. And I 'only' got a i7 7700k! But would upgrade in a heartbeat if there was some real advantages.
But who could/ will do that?
 
The problem is not that Sim developers don't want to have better multi-core support. The problem is that it is very hard to implement. The current development tools simply don't make it easy enough to get wide spread multi-core game support.

Let me put this another way. Pointer arithmetic and dynamic memory allocation are relatively simple by comparison to multi-core support. MS came out with the .NET framework with built in garbage collection to dynamically allocate needed memory and then free up memory not in use. That allowed many more developers to be productive since they couldn't be trusted with dynamic memory allocation. Many games are still written in C++ rather than C# for speed reasons and still handle dynamic memory allocation by hand. What we are looking for is a break through in multi-core support making it easier to break up processing into logical units that can occur in parallel, but at a very low level for performance. This is a tough ask, which is why we don't have it yet.

The only reason I'm explaining this is that it is currently unreasonable to expect sim studios to take this on. They need help in the form of a break through that allows them to implement it without tearing up their code base.


BTW I'm not a fanboy of any company either and I completely understand buying from a company that you respect if you feel another company has been taking you to the cleaners. I simply get the best if I can afford it. If AMD comes out with the next top end CPU than I'll reward that with a purchase. I felt like I bent over and grabbed my ankles when I got my 2080Ti, so I completely understand. Still I feel like my current setup will last me for the next year or so until the next big trifecta gets here ( CPU, GPU, VR headset )
 
The problem is not that Sim developers don't want to have better multi-core support. The problem is that it is very hard to implement. The current development tools simply don't make it easy enough to get wide spread multi-core game support.

Both MS and Codemaster could easy do it if they wanted to!
A small step would have been to implement Dx12 or Vulcan that would off load the CPU and people could finally get full performance out of their RTX 2080 ti GPU and give some more CPU cycles to Ai and online races.
Been using Dxvk with AMS 2 with good result! About +8-10% more FPS also works well in VR!
 
Both MS and Codemaster could easy do it if they wanted to!

Why do you believe this? This is very heady stuff. I know a number of the guys who train MS employees, write technical books and innovate many new features that go into each iteration of their development tools. I used to work with them. One of them actually patented a type of thread lock (a synchronization mechanism for multi-threading)that he both solid the patent rights and implementation to MS. This guy is amazing, and is currently a chief architect on their Azure team.

BTW I'm not saying it can't be done, but just that it will be very difficult. I don't care which company is behind it. What that means is that there are only a handful of people who might have the ability to crack this with current development tools. Keep in mind that right now a lot of the serious talent is going into AI because of the money there. So solving something like this for gaming is probably not going to be a top priority and it probably won't have dibs on the top talent to figure it out.

What I read recently was that there was some progress at the microcode level which is embedded in the CPU. The company which had patented that technology is also working on an implementation, but I haven't heard anything new in almost a year. I'm not sure what the software requirements of this would be, but there is no CPU that currently utilizes this.
 
Last edited:
BTW I'm not saying it can't be done, but just that it will be very difficult. I don't care which company is behind it. What that means is that there are only a handful of people who might have the ability to crack this with current development tools. Keep in mind that right now a lot of the serious talent is going into AI because of the money there. So solving something like this for gaming is probably not going to be a top priority and it probably won't have dibs on the top talent to figure it out.

That is nonsense!
Intel got free libraries out since early 2000's that are multi threaded. Lots of other games take advantage of +4 cores but not race sims!
My best guess is they all relay on very old libraries which might have seen some updates in calculations but are basically 20 years old.
When MS brings out Forza Motorsport 7 as a single threaded program is is not due to is is hard to make part of it multi threaded but because they are lazy and do not care about making a better product.
As they want it to run the same on high powered PCs as on old Xbox tech.
I do not want to further discus this with you as I do not know enough about this! But I do know enough to know that you do not.
And anyway this is a VR thread.
 
We aren't going to come to an agreement on this. Having a free multi-threaded library does not remotely make your software good with multiple cores. It's just not remotely that simple.

Like I said this is very easy for say a compression tool, or for video rendering and a bunch of other computationally intensive tasks which all supported this very shortly after these SDK libraries were available. A real time feedback loop with lots of branching is much much much more difficult.
 
Last edited:
  • Deleted member 197115

Forza Horizon 4 is already taking advantage of multi-core
On sims side, it's all limitations of used engines that still rely on single threaded rendering pipeline.
 
That's a perfect example. The rendering is all done past the decision making which is your realtime feedback loop. Once you get past that it's all on the rendering engine which is a pipeline that can be split up. This is exactly what NVidia has done in their non-generic processing blocks on their GPU's. The rendering engine is the one part of this that can be split up, but it has to run synchronously which places limitations on it. With rendering video and compression, individual chunks can be run asynchronously making them MUCH easier to deal with.

Going from 2 to 4 or 4 to 8 cores isn't simply adding to an array and recompiling. There is a lot going on.
 
Last edited:
Trying to get up to speed with VR stuff since I am finally almost done with my SFX100-based sim rig: Is there any way to take advantage of Vive Pro Eye's foveated rendering in any popular racing sims? Guessing not, but wanted to double check before making any purchases. Leaning towards the Valve Index unless I am missing something else about current/upcoming tech.
 
  • Deleted member 197115

Trying to get up to speed with VR stuff since I am finally almost done with my SFX100-based sim rig: Is there any way to take advantage of Vive Pro Eye's foveated rendering in any popular racing sims? Guessing not, but wanted to double check before making any purchases. Leaning towards the Valve Index unless I am missing something else about current/upcoming tech.
New Pimax, Oculus, HP Reverb v2, and Samsung are coming (Odyssey+ is officially discontinued).
Hold on a little longer, buddy.
 
Trying to get up to speed with VR stuff since I am finally almost done with my SFX100-based sim rig: Is there any way to take advantage of Vive Pro Eye's foveated rendering in any popular racing sims? Guessing not, but wanted to double check before making any purchases. Leaning towards the Valve Index unless I am missing something else about current/upcoming tech.

Right now for sim racing the Index or Reverb is probably your best bet.

If you do much room scale, the Index is a better choice by a mile. Also if you end up with a Pimax or StarVR in the future you keep the controllers and base stations from the Index. The Index also has a longer cable making it easier to route.

OTOH with an SFX-100 the Reverb may have less issues with EMI. The Index will work, but you may need some ferrite cores to keep it from blacking out.

The Reverb will let you see you brake points in the distance more clearly, but has less peripheral vision.

The Index is a much higher quality kit than anything else out. It's very comfortable, the audio is excellent and it has a very good FOV except compared with Pimax or StarVR.

Both the Pimax 8KX/8K+ are delayed, so there is no telling when they will actually be released. The StarVR is supposedly available, bit it appears they really aren't.

I know a number of people who own the Rift-S, Index and Reverb. The results are split. Some prefer the Reverb, others perfer the Index.

So it depends on your priorities. Personally I love my Index.
 

Latest News

What's needed for simracing in 2024?

  • More games, period

  • Better graphics/visuals

  • Advanced physics and handling

  • More cars and tracks

  • AI improvements

  • AI engineering

  • Cross-platform play

  • New game Modes

  • Other, post your idea


Results are only viewable after voting.
Back
Top