• 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.

Poll: Do you think AMD will be able to compete with Nvidia again during the next few years?

Do you think AMD will be able to compete with Nvidia again during the next few years?


  • Total voters
    213
  • Poll closed .
I think AMD compete very well at the lower ends of the market. As far as the mid market goes they were just too late to the party and despite now owning a vega and seeing some of the benefits (it was cheap) vega was hardly launched with critical success.

As for the high end well I suspect Nvidia will keep moving the goal posts. AMD hasn’t yet matched top end Pascal for performance and when it finally does Nvidia will be streaks ahead again. I do realise this is a very small market though and I suspect AMD are happy cleaning up at the lower end of the market.
 
As for the high end well I suspect Nvidia will keep moving the goal posts. AMD hasn’t yet matched top end Pascal for performance and when it finally does Nvidia will be streaks ahead again. I do realise this is a very small market though and I suspect AMD are happy cleaning up at the lower end of the market.

Are you suggesting that Nvidia are sandbagging their raster performance with the RTX line? In order for Nvidia to be "streaks ahead" that would mean they have a lot more raster performance available than we've seen in RTX.

Obviously RTX's big selling point is ray tracing and AI but the core raster performance hasn't really improved that much over GTX 10. Given how big a 12nm die is, Nvidia couldn't really ramp up the CUDA cores much more given how much space is taken up by RTX tech, so in theory a shrink to 7nm will give them plenty of room for gen 2 RTX and a boat load more CUDA. But then AMD will have Arcturus in 2020, which is a brand new arch based on 7nm+.

If Nvidia are not sandbagging their raster performance, it is a distinct possibility that AMD could seriously close the gap with Arcturus, much like they've done with Zen against Intel.

But yes, as part of this rebuilding process I don't think AMD are too bothered about chasing the halo performance crown just yet. Navi as leaked will happily take on RTX 2070 and down for a common-sense amount of money and that's bread and butter, high volume stuff. It would be nice if the alluded "Big Navi" could raster punch with the 2080 Ti, but thus far AMD haven't touched the GTX 1080 Ti, let alone another 30% on top.

I doubt we'll see any top-end competition until Arcturus at the very earliest.
 
When have NVIDIA arrived late?
Dozens of times over the decades, the last example being the GTX600 series which arrived late with worse performance/$ (and yet still outsold the competition due to fanboy-ism).


I'm just saying in an ideal world AMD being competitve would force NVIDIA to be competitive. AMD should price their stuff better because they arrive 6 months late to the party.
In fairness the 1070ti was a direct response to the Vega 56, that is literally AMD forcing Nvidia to be competitive. As far as pricing their stuff better, the V64 and GTX1080 being equal on price and performance (until mining demand mooned the Vega prices) did mean anyone interested in Freesync/Gsync (which was the big thing at the time) had a cheaper option with AMD.

This non-competitive thing seems to stem mainly from AMD having no answer to the 1080ti, which was indeed a problem, but a problem that affected less than 1% of GPU buyers. For the 99% AMD were quite competitive and the industry/consumers benefited from it.
 
When have NVIDIA arrived late?

Dozens of times over the decades, the last example being the GTX600 series which arrived late with worse performance/$

Personally I'd say RTX is the latest example. Nothing new for 2 years then when it finally shows up the new tech is woefully underpowered, software support might as well be vapourware, core raster performance is identical to the previous generation, just shifted a product tier, and then ridiculously priced to boot.
 
Personally I'd say RTX is the latest example. Nothing new for 2 years then when it finally shows up the new tech is woefully underpowered, software support might as well be vapourware, core raster performance is identical to the previous generation, just shifted a product tier, and then ridiculously priced to boot.
RTX 2080ti is crazy’s fast at 4K though. Maintains 60fps in just about anything. Wipes the floor with 1080ti. It’s too bad the price is crazy too.
 
Variable Rate Shading, A Possible Addition To The AMD Strategy?
https://wccftech.com/variable-rate-shading-a-possible-addition-to-the-amd-strategy/

"AMD's patent which was filed all the way back in February of this year seems to suggest that for either an upcoming update or upcoming graphics card, VRS or Variable Rate Shading, would be implemented. Variable Rate Shading is a technique already in use by Nvidia in its graphics cards, VRS allows the GPU to spend more of its focus to the more complex/action orientated part of the screen, by saving resources on the less complex parts of the screen. The technique has helped Nvidia stay ahead of the AMD when it comes to raw performance.

AMD's new graphics cards could come with VRS (Variable Rate Shading) to help them beat Nvidia's overwhelming odds in the GPU market.
While there has been no confirmation from AMD revolving around this issue, to be able to use VRS would most definitely shorten the gap between their GPUs and NVIDIA.

The RX Vega 64 is the GPU that I currently have in my system and it uses much more power than the NVIDIA counterpart because it isn't able to take advantage of VRS to only spend power on the more complex part of the screen. This is also one of the reasons that AMDs Graphics cards are usually more power-hungry than their NVIDIA counterparts."



Let's hope they will manage to get anything out of this.

The other thing is to enable/use CrossFire/ GPU chiplets.
And to implement some type of ray-tracing acceleration.
 
So how can AMD patent it? There's prior art so no dice.

You probably can't patent the concept/idea of VSR, but you can patent a technique for doing it (as long as no one else has patented that same technique).

Generally you don't patent the idea, you patent the implementation.
 
The RX Vega 64 is the GPU that I currently have in my system and it uses much more power than the NVIDIA counterpart because it isn't able to take advantage of VRS to only spend power on the more complex part of the screen.

Vega64 was competing against Pascal, nVidia only implemented their version of VRS with Turing and not before.
 
You probably can't patent the concept/idea of VSR, but you can patent a technique for doing it (as long as no one else has patented that same technique).

Generally you don't patent the idea, you patent the implementation.

Exactly this, you cannot patent a concept or idea, only how to achieve the end result, Nvidia will have patented that, AMD will have to find and then patent a different way of achieving the same thing, which they have and it will come with the second generation of RDNA, along with Hardware Accelerated Ray Tracing.

Nvidias Ray Tracing is 'In my opinion' extremely inefficient.
 
Care to elaborate why? Genuinely curious :)

Ray Traced lighting and reflections have been in games for years, some do it better than others, most implementations of it are not as extensive as RTX and sometimes have subtle but no less undesirable graphical side effects.

Still, when done well it can be just as good as RTX without the need for Hardware Acceleration, the fact that Nvidia are using apparently very powerful Hardware Acceleration and still managing to get pretty poor performance is what i'm talking about.

The most demanding of Ray Tracing is Global Illumination, here's me in a 5 year old engine switching between it, no Hardware Acceleration.

 
Last edited:
Lol. Spamming this everywhere I see.

what good is it them killing an already 14 month old GPU? By the time it is out it will be over 18 months and Nvidia will kill it with ease with their 3000 series. I am not excited at all really. Plus look at the link. Wccftech. Why do you guys even post links to that fake news site? Desperate much?
 
So according to these slides.

2020 Navi cards will only be able to do Ray Traced global illumination, it won't support ray traced reflections and shadows.

That's a bummer. But then on the same slide, indicates that in the future AMD wants to do full scene ray tracing using the almight "cloud"... goodluck with this, we waited years to see how Microsoft's cloud physics would work and it was a massive failure.
 
So according to these slides.

2020 Navi cards will only be able to do Ray Traced global illumination, it won't support ray traced reflections and shadows.

That's a bummer. But then on the same slide, indicates that in the future AMD wants to do full scene ray tracing using the almight "cloud"... goodluck with this, we waited years to see how Microsoft's cloud physics would work and it was a massive failure.
Yeah, not interested in anything cloud until proven to work perfectly. It seems to me Nvidia will come out on top again with their implementation. Shame.
 
So according to these slides.

2020 Navi cards will only be able to do Ray Traced global illumination, it won't support ray traced reflections and shadows.

That's a bummer. But then on the same slide, indicates that in the future AMD wants to do full scene ray tracing using the almight "cloud"... goodluck with this, we waited years to see how Microsoft's cloud physics would work and it was a massive failure.


What like this? Real Time Reflections have been around long before "RTX" was a thing, this is a 4 year old engine, i did this the other day for something but never used it, might as well use it now...

The one that is static is using Cube Maps.


Oh and BTW, RTX on a Vega 56 using a much later version of the same engine.

 
More here, this is the same engine 'under another name' as the one i was using in the first video, its subtle because of the roughness of the textures used but you can see the reflections in the floor, i'll go to that area later and recod it, Realtime Reflections all over the place, all over the game infact, no one notices this because its doesn't have "RTX" stamped all over it, tells you a lot that.....

https://youtu.be/OngP6uEfQoE?t=274
 
I reckon the gap will be closed on average when not taking into games that favour heavily one camp or the other (ray tracing off as well).

All AMD needs to do is not charge the same price, AMD needs a halo product that performs on par with Nvidias halo product and this being the Titan, not the x080ti ( they should have a card ready for this as well ).

AMD hasn't brought out a stacked line up in a long time based on the same design in the longest of times, think what? HD7000 series? thats pretty rough going.
 
Back
Top Bottom