We were told, "It just works"

XzibitWe were told, "It just works"

No, the developers were told "Get to work!"

Sim Racers want above all things high, consistent FPS over graphic fidelity ALWAYS. Also being part a very large sim racing group, many have pretty low spec machines and can’t turn on the “shiny stuff” to begin with.

XzibitWe were told, "It just works"

I don't remember the exact wording around it but the context of "It just works" is different than what the meme says.

That quote was not directed at us, consumers. It was aimed at developers and the context is how much technical work needs to be done to get the results for effects demoed. The main effects that were demoed - soft shadows, AO and reflections - all exist in rasterization and have a range of methods to implement these (often on top of each other). In addition to that, there are always situations where the resulting effects do not look right and art/leveldesign/design teams make sure theses situations do not appear in the finished product. Now, the same effects in RT variation are more detailed and "physically correct" with effectively only using one method. That is what the quote is about.

This is not just Nvidia saying it. I strongly recommend watching game/engine developer presentations around RT - they do highlight these exact same things, mentioning that this has big potential to for real savings on less time being spent on this area. The current problem is performance both in terms of what hardware can do as well as how games/engines are written to set things up for RT.

I get it that Nvidia is an easy mark and RTX marketing is over-the-top but the fact of the matter is that most of it has very little to do with Nvidia specifically and applies to RT as a whole. Uncharacteristically for Nvidia, RT part of RTX is not proprietary technology, especially in light of DXR. If and when competitors (AMD primarily) come out with RT solution, this will benefit immensely from the work developers and Nvidia are doing right now.

so RTX is very important to developers LOL

XzibitWe were told, "It just works"

One game gone, 6 (Watch Dogs 3, CoD Modern Warfare, Cyberpunk 2077, Wallenstein Young Blood, Grimmstar, Bright Memory) joined the party.

A shame, really, since seeing proper reflections on the cars around you would be quite stunning.

XzibitWe were told, "It just works"

I keep seeing this claim. What I don't see is a source. So, just to make it clear: you and who else were told that?

there is no reason to steal development resources and time for a very low frame rate implementation

Lol..nice little bit of trolling there.

You listening, Jensen?...you want Ray Tracing in this game then start splashing the $$$ to the studio to pay for it.

bugA shame, really, since seeing proper reflections on the cars around you would be quite stunning.


I keep seeing this claim. What I don't see is a source. So, just to make it clear: you and who else were told that?

It's a direct quote from Nvidia's CEO, Jensen Huang during the reveal of RTX cards.
10 hour meme version:

erockerIt's a direct quote from Nvidia's CEO, Jensen Huang during the reveal of RTX cards.

Ah, I believe I was told that before. I really need to dig up that video, I never watch these things (mostly because they're full of bs ;) )

Mamya3084No, the developers were told "Get to work!"

You misunderstand, this is all temporary because right now its very hard to combine raster and gloriously simple ray tracing (that's the excuse today). Never mind the fact the software side still needs to exist to use that, but hey.

And guess what, its temporary for as long as market share is low because (see article) happens everywhere. I'm getting that PhysX vibe again.

Whoever says 'It just works' in anything related to code needs to get his head examined.

I'll never forget the absolute, long, and devastatingly awkward silence when Huang mentioned his 'Ten Gigarays' against the highly educated audience at SIGGRAPH. It spoke volumes and was exactly why I'm still convinced this RTX affair is a big pointless exercise driven by shareholders, not by 'we are ready for it'. We're not ready, the industry just needs a new cash cow.

bugAh, I believe I was told that before. I really need to dig up that video, I never watch these things (mostly because they're full of bs ;) )

Its really worth doing, maybe you'll understand why I'm so stubborn about this. I recognize a liar when I see one.

Vayra86I'll never forget the absolute, long, and devastatingly awkward silence when Huang mentioned his 'Ten Gigarays' against the highly educated audience at SIGGRAPH.

From what we have seen so far gigarays are not the primary problem right now. This part of performance hit can be and has been mitigated - less rays, noisier image.

londisteFrom what we have seen so far gigarays are not the primary problem right now. This part of performance hit can be and has been mitigated - less rays, noisier image.

The whole point was mitigating developer time to achieve results. So far, raster is doing that rather well, and much better than RT. The high degree of dynamic lighting part of RT is a fairy tale given the performance cost.

Economics don't lie. And so far, there is very little incentive to spend dev time on something that might just stick, but probably not for the foreseeable future. Any developer with a tight budget (all of them) will choose to cut cost and/or focus on core game features rather than RTX/DXR support for the happy few. This technology is being launched with a very rocky start and one-sided hardware support (NV only) in a marketplace where high-budget AAA titles are fast losing ground to copy/paste cosmetic grind games and F2Ps. What remains is indie devs that won't even consider spending time to look at it in the first place.

Vayra86You misunderstand, this is all temporary because right now its very hard to combine raster and gloriously simple ray tracing. Never mind the fact the software side still needs to exist to use that, but hey.

And guess what, its temporary for as long as market share is low because (see article) happens everywhere. I'm getting that PhysX vibe again.

Whoever says 'It just works' in anything related to code needs to get his head examined.

You listen here. We didn't pay to have our green logos in the game to here about "Development issues". Put the RTX in the game like we paid for...and make sure you use every NVIDIA specific effect to make the game run like crap on anything other than our GPUs. We are the standard.

Vayra86The whole point was mitigating developer time to achieve results. So far, raster is doing that rather well, and much better than RT.

That's not RTRT's fault. Developers are always more productive with the toolkit/framework/SDK they know than with a new one, no matter which is better.

bugThat's not RTRT's fault. Developers are always more productive with the toolkit/framework/SDK they know then with a new one, no matter which is better.

That's why most things work with iterations these days, slowly adding features instead of going all big bang on the world and failing miserably, like most of our Web 1.0 IT did. You know, gaining momentum, industry support, and then making it so widespread that you can't circumvent it. Nvidia knows how it works, they did that with CUDA.

CryTek got that memo, for example. They adjusted their engine and it can produce the magic without hardware, while the hardware can still add to performance.

Vayra86That's why most things work with iterations, slowly adding features instead of going all big bang on the world and failing miserably, like most of our Web 1.0 IT did.

Absolutely. But you can't migrate to RT iteratively, that's why we're on the fence. And will stay there for a while :(

bugAbsolutely. But you can't migrate to RT iteratively, that's why we're on the fence. And will stay there for a while :(

Yes you can, see Crytek. It all depends on the approach and the underlying goal. Nvidia's goal was to sell overpriced hardware, first and foremost.

Crytek, Epic, DICE and others are implementing RT iteratively and all of them are following roughly similar paths while going about this.

Edit:

Vayra86CryTek got that memo, for example. They adjusted their engine and it can produce the magic without hardware, while the hardware can still add to performance.

I would be careful about singling out CryTek with this. They did generate nice PR for them but looking at what they achieved more objectively does not reveal anything unique or strictly better. In CryTek's RT demo performance thread I used this example and I will do it again - Neon Noir is an RTRT reflections demo and CryTek got 30 FPS at 1080p on Vega56. This is not a bad result by any measure. However, for comparison, Battlefield V with DXR reflections (technically very similar to what Neon Noir does) runs at around the same 30 FPS at 1080p on GTX1080.

londisteCrytek, Epic, DICE and others are implementing RT iteratively and all of them are following roughly similar paths while going about this.

Maybe... just maybe they should have already been having something when Huang yelled about gigarays.

This is all about timing and perception. What wouldve landed better, that realtime Noir demo running on Pascal at 20-30 FPS, or giving us 3 FPS demos to push people to Turing post launch?

And lets not forget about the major price bump that goes along with that either.

Any thread on RT will devolve into the same thing - NVidia is evil and Turing GPUs are overpriced. This gets real old real fast.

We should see Neon Noir on GTX cards - and hopefully given DXR implementation on RTX cards - spring-summer of 2020 when CryTek will actually release the feature on CryEngine.

It is a chicken and egg problem. No hardware support - no games. No games - no hardware support. Someone has to be first and Nvidia jumped on that last year. Marketing hoopla is just part of this game.

londisteAny thread on RT will devolve into the same thing - NVidia is evil and Turing GPUs are overpriced. This gets real old real fast.

Honestly I have not seen either in this threat - what I have seen is a CEO being called on his own quotes...
If it the claim is that it is implemented (and optimized) easily, then it should be.

It works so well.
Almost a YEAR passed, No. of RTX DEMO > No. of actual RTX games.

@btarunr

Why the sensationalist title?

Why not use "drop" instead of "dump"? As a journalist you must remain neutral.

londisteAny thread on RT will devolve into the same thing - NVidia is evil and Turing GPUs are overpriced. This gets real old real fast.

We should see Neon Noir on GTX cards - and hopefully given DXR implementation on RTX cards - spring-summer of 2020 when CryTek will actually release the feature on CryEngine.

It is a chicken and egg problem. No hardware support - no games. No games - no hardware support. Someone has to be first and Nvidia jumped on that last year. Marketing hoopla is just part of this game.

The same boring response to every turing thread. "well well well chicken and egg! GAMES R COMING!!!!!!!"

It's only been a year. Do you REALLY think good RT is coming so soon? It took how long from DX11"s release to devs actually using it in more then 2 games? Over half a decade? And that had tangible benefits for devs and could run on a large variety of GPUs, given how much more advanced it was over DX9.

RT, meanwhile, is only available from one of the two vendors, only on their high end GPUs, only the top two can do it at all (the 2060 and 2070 RT capabilities are hilariously worthless). Many people think that RT makes the games look worse, not better, and the FPS hit you take is tremendous. I mean, 1080p30? That is pathetic. IF (and that a big IF) RT somehow becomes an industry standard, the RTX 2000 cards will be COMPLETELY worthless, much like DX8.1 or DX 10 cards. RT certainly wont catch on as a major graphical feature as long as you need a $500+ GPU to use it at all, and $800+ to get one that works correctly, for all that means. Especially if the new generation of consoles coming next year use RDNA, not RDNA2, and thus have no hardware raytracing capability, it will take another half a decade + before it begins to matter at all.