• 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.

AMD Navi 23 ‘NVIDIA Killer’ GPU Rumored to Support Hardware Ray Tracing, Coming Next Year

Status
Not open for further replies.
You can't be shafted when you buy a product knowing that it doesn't support a certain feature. Now if you bought a 2060 or 2070 expecting to be able to play games with decent fps and RT, I'd say Nvidia shafted you

This is the point i was trying to make (You made it far better in less words.. lol) but words aren't my strongpoint today. :)
 
AMD Radeon RDNA 2 GPUs will lead the way with every console port optimised for the RDNA 2 architecture, going to be a rough few years for Nvidia.

But, that's what people said with the last consoles and it never materialised. And it will make less difference now going forward, at least on Xbox games. Most games going forward will be developed and optimised for Dx12 Ultimate or Vulkan. Because the Xbox and PC will be using the exact same DirectX for the first time ever. I reckon it will still be the same as it is now for raster performance, there will be some outliers that play better on AMD or Nvidia GPUs, but, for most games the faster GPU will be better in games. Can't comment on RT performance.

Now it could still be a few tough years for Nvidia if AMD get their next couple of GPU launches done without problems, especially with the mess Nvidia have made with the 3080/3090 launch. The ball is in AMD's court now, can they start off on a good note with their RDNA 2 launch??
 
Always moving goalposts.

I am using a 2080 ti SLi setup and I don't like the performance drop with RTX on...I won't be enabling RTX on Ampere or RDNA2 cuz it's going to be the same story... I won't be possibly using RTX till the performance drop is less than 2%... which might take 3 or more generations of upgrades (I doubt this, the problem seems intractable)..RTX is a waste of computational power which could have been used elsewhere. I am upgrading my knowledge about RTX in the meanwhile..but it's looking nothing more than a grunt approach to IQ
 
How many decent games with RTX over the past two years? :p fact is now the consoles with RDNA 2 are soon to be on the scene, raytracing will become more mainstream and on the desktop AMD Radeon RDNA 2 GPUs will lead the way with every console port optimised for the RDNA 2 architecture, going to be a rough few years for Nvidia.

In the same way games were optimised for GCN and 8 core CPUs for the last generation?
 
How many decent games with RTX over the past two years? :p fact is now the consoles with RDNA 2 are soon to be on the scene, raytracing will become more mainstream and on the desktop AMD Radeon RDNA 2 GPUs will lead the way with every console port optimised for the RDNA 2 architecture, going to be a rough few years for Nvidia.

Why would every console port on the PC be optimised for AMD, when the PCs got Nvidia in it :p
 
Always moving goalposts.

Typical childish response when faced with cold hard facts.

Your a troll and a not very good one at that.

Quit trying to stir up controversy in this AMD thread, its more than obvious it is what you are doing, and equally as transparent you are a blatant Nvidia fanboy.
 
https://twitter.com/CyberCatPunk/status/1312745996888154115

Confirmed:
Navi21 --- Sienna Cichlid
Compute --- GFX10 (gfx1030)
80CUs/5120 shaders/320 TMUs /96 ROPs
~20TF (FP32)

CFO Devinder Kumar has confirmed Hardware-Level RT
'will use ~ 20CUs from the 80CU pool for HWRT'

Not Confirmed:

- Exact Memory Configurations
- Peak Core Clock

80 CU's and 96 ROP's apparently confirmed.

20 CU's will be given over to Ray Tracing when required.
 
https://twitter.com/CyberCatPunk/status/1312745996888154115



80 CU's and 96 ROP's apparently confirmed.

20 CU's will be given over to Ray Tracing when required.

What's great about this is you can have them 80CU when you not wanting or using RT :D

On Paper this is much better than how Nvidia is doing it. But time will tell just how well AMD does RT, I expect they shouldn't be any issue reason being AMD has been working on RT behind the scene for years now while working along side Microsoft and Sony for the next gen consoles.
 
20 CU's will be given over to Ray Tracing when required.

Interesting.. I'd heard that RT would just be pushed down the same pipeline so as not to waste hardware? It would be very dependent on good scheduling to make it work with minimal impact that way, but it seems better than dedicating units - unless I'm misunderstanding?
 
Interesting.. I'd heard that RT would just be pushed down the same pipeline so as not to waste hardware? It would be very dependent on good scheduling to make it work with minimal impact that way, but it seems better than dedicating units - unless I'm misunderstanding?

No idea.

I think that tweet may well be a load of Rubbish

Its just another rumour, but what makes you think this one is rubbish?

It makes a lot of sense to me given that the RDNA2 architecture has dual CU Work Groups, you see on this slide they are labelled as Ray Tracing or Texture Mapping.

Mi8iJu7.jpg.png


XBox Series X

wOvywMK.jpg.png
 
I thought of rechecking the Xbox specs
15.3 B transistors on 360.4 mm2
Around 42.5 M transistors per sq mm
At 536 mm2 for Big Navi.. works out to approx 23 B transistors which is nowhere close to Coretek's 27 B estimate (but still slightly bigger than the 3080).. are there further efficiency gains expected over Xbox SOC? cuz that chip can theoretically pack in 50 B transistors

I looked at this article on Xbox:
https://www.eurogamer.net/amp/digitalfoundry-2020-xbox-series-x-silicon-hot-chips-analysis

Total area 360.4 mm2
56 CUs take 47% of area (assuming this includes RB and TMU going by the dieshot)
per CU area 3 mm2
11% area for memory controllers which is 39.6 mm2
I don't know what the multimedia accel zone contains but looks like 4.8% which is 17.3 mm2.. am taking this into the GPU die

Total available area 536 mm2
Total fixed area 56.9 mm2
Total available area for CUs 479.1 mm2
Works out to 160 CU.. this might also mean that the PC CU is twice the size of Xbox CU
Hard to believe but if true the largest sku will have 10240 shaders to match Nvidia
Or some of that die space is taken by infinity cache.. too many variables :(
 
I looked at this article on Xbox:
https://www.eurogamer.net/amp/digitalfoundry-2020-xbox-series-x-silicon-hot-chips-analysis

Total area 360.4 mm2
56 CUs take 47% of area (assuming this includes RB and TMU going by the dieshot)
per CU area 3 mm2
11% area for memory controllers which is 39.6 mm2
I don't know what the multimedia accel zone contains but looks like 4.8% which is 17.3 mm2.. am taking this into the GPU die

Total available area 536 mm2
Total fixed area 56.9 mm2
Total available area for CUs 479.1 mm2
Works out to 160 CU.. this might also mean that the PC CU is twice the size of Xbox CU
Hard to believe but if true the largest sku will have 10240 shaders to match Nvidia
Or some of that die space is taken by infinity cache.. too many variables :(

I'm reluctant to make a thing of this because people in here have a habit of holding it against you when you get your speculations wrong, but, count the CU's in this slide, ignore the blue tab on the side actually count the CU's on the diagram. 20 in each array. 8 arrays.

Mi8iJu7.jpg.png
 
Status
Not open for further replies.
Back
Top Bottom