AMD Radeon RX 6800 vs. NVIDIA RTX 30 Linux Performance Heating Up

[qarium]

Hey as an Nvidia teamster, I will give AMD kudos on this one! If you are strictly looking at gaming, the AMD 6K series is very impressive (sans Ray Tracing). No point in comparing them to Nvidia on availability as getting any one of these cards these days is brutal.
It is nice to see some competition from team red finally! Hopefully they can get their act together for the ray tracing in the next gen (7k series?) as its rapidly looking like RT is going to be the "new hw T&L".

I think AMD world on 4-5 things:

die shring to 5nm TCSM for the GPU core
HBM3 modells with 32-64GB vram for high paying customers like apple
increase the 128mb infinity cache to 256MB and maybe go from 12nm to 12nm-LP+ or 7nm-

https://www.techpowerup.com/269309/globalfoundries-announces-12lp-enhanced-12nm-silicon-fabrication-node

Also Global Foundries invest to double the 12nm capacity (FD-SOI)

https://www.eetimes.com/globalfoundries-offers-ambitious-tech-plans-while-eying-an-ipo/

right now 5nm and 7nm is complete overloaded without any capacity free so the 256bit infinity cache chip will be FD-SOI 12LP+ nm

I think AMD also workd on OoO Out of Order execution shaders.

https://en.wikipedia.org/wiki/Out-of-order_execution

Multi-GPU-Chiplet design for VR gaming... with 2 gpu cores for each eye one gpu and 2 infinity cache and HBM3 vram.

as you can see Nvidia is doomed...

[hiryu]

Hmm... I think Sapphire is the best brand for AMD, their Nitro cards are really Good.

I remember I used to buy Sapphire back in the day and seemed pretty good. Thanks!


I'm currently trying to determine where ray tracing performance is... Lower than the RTX 30 series of course. But what about compared to the RTX 20 series? I see the RTX 2080 Ti pulling ahead at least in some games... but perhaps the AMD simply isn't a full drop-in replacement for ray tracing and some AMD specific optimizations are needed..?

[qarium]

I'm currently trying to determine where ray tracing performance is... Lower than the RTX 30 series of course. But what about compared to the RTX 20 series? I see the RTX 2080 Ti pulling ahead at least in some games... but perhaps the AMD simply isn't a full drop-in replacement for ray tracing and some AMD specific optimizations are needed..?

no... you think to one-dimensional... but the topic is multibe dimensional...

yes the 6900XT is like 25% slower in Nvidia implementation of Raytracing
but if you use the AMD implemenation of Raytracing the AMD 6900XT is 10-15% faster than nvidia 3090.

so yes the AMD raytracing implementation is the AMD Specific Optimisation.

but now you think like all games will use the nvidia implementation... WRONG!!!

Why wrong?.. well playstation5 and xbox use the AMD implementation because of this many games will use the AMD impementation.

[hiryu]

no... you think to one-dimensional... but the topic is multibe dimensional...

yes the 6900XT is like 25% slower in Nvidia implementation of Raytracing
but if you use the AMD implemenation of Raytracing the AMD 6900XT is 10-15% faster than nvidia.

so yes the AMD raytracing implementation is the AMD Specific Optimisation.

but now you think like all games will use the nvidia implementation... WRONG!!!

Why wrong?.. well playstation5 and xbox use the AMD implementation because of this many games will use the AMD impementation.

Absolutely! I am aware of all of this, and I should have been more clear in my previous post.

If AMD's ray tracing implementation is faster than the RTX 20 series' in cases where both are optimized for their respective platforms... I can live with that, and I will likely pull the trigger.

Presently, I'm searching the web to find such comparisons right now. If anyone has good ones to share, I'd appreciate it!

[Crashone]

These results are misleading. It seems above all that the Nvidia graphics cards has drastically lost in performance compared to November 2020. For instance, in F1 2017, the 3080 fell from 183fps to 163fps in 2160p high preset, from 97fps to 92 fps in Shadow of the Tomb Raider (High), from 142 to 136fps in Total War (Low), etc.

https://www.phoronix.com/scan.php?page=article&item=amd-rx6800-linux&num=11

[Volta]

These results are misleading. It seems above all that the Nvidia graphics cards has drastically lost in performance compared to November 2020. For instance, in F1 2017, the 3080 fell from 183fps to 163fps in 2160p high preset, from 97fps to 92 fps in Shadow of the Tomb Raider (High), from 142 to 136fps in Total War (Low), etc.
https://www.phoronix.com/scan.php?pa...0-linux&num=11

There's nothing misleading there. Blame nvidia for your lost performance. Another birdie's account?

[ms178]

These results are misleading. It seems above all that the Nvidia graphics cards has drastically lost in performance compared to November 2020. For instance, in F1 2017, the 3080 fell from 183fps to 163fps in 2160p high preset, from 97fps to 92 fps in Shadow of the Tomb Raider (High), from 142 to 136fps in Total War (Low), etc.
https://www.phoronix.com/scan.php?pa...0-linux&num=11

What is the basis of accusing Michael of intentionally skewing his testing? Just because the numbers are different, doesn't mean Michael has done anything wrong or intentionally presenting these cards in a different light. These numbers reflect the current state of these cards with the current driver, Proton, BIOS and game versions. Regressions can happen over time, also minor differences could be attributed to some run-to-run variance, different BIOS or Kernel versions or any other of all the variables which might have an impact on performance which are different between the last and today's testing.

[qarium]

Absolutely! I am aware of all of this, and I should have been more clear in my previous post.
If AMD's ray tracing implementation is faster than the RTX 20 series' in cases where both are optimized for their respective platforms... I can live with that, and I will likely pull the trigger.
Presently, I'm searching the web to find such comparisons right now. If anyone has good ones to share, I'd appreciate it!

i read such comparisons in the past and i maybe post you some.

you write "faster than the RTX 20 series"

you just do not get the point. the 6900XT is 15% faster in raytracing than a 3090 as soon as the AMD implementation is used.

games who use the AMD implementation is for example: world of warcraft of Blizzard

see here are many raytracing benchmarks and in the AMD implemenation in world of warcraft AMD is much faster:

https://www.pcgameshardware.de/Radeon-RX-6900-XT-Grafikkarte-276950/Specials/RX-6900-XT-oder-RX-6800-XT-OC-Vergleich-mit-RTX-3090-1363216/

[aufkrawall]

see here are many raytracing benchmarks and in the AMD implemenation in world of warcraft AMD is much faster:

https://www.pcgameshardware.de/Radeo...-3090-1363216/

Nvidia hits some weird CPU bound performance issue in WoW with RT. Also Watch Dogs Legion DX12 runs ~20% faster in CPU bound scenes on AMD, even without RT.

[Crashone]

What is the basis of accusing Michael of intentionally skewing his testing? Just because the numbers are different, doesn't mean Michael has done anything wrong or intentionally presenting these cards in a different light.

Because it does not seem objective to me to say that AMD has increased in performance without specifying that in the same time that of NVIDIA decreased and conclude" To much surprise overall the Radeon RX 6800 manages to effectively tie with the GeForce RTX 3080 overall while the RX 6800 XT had around a 12% advantage in these benchmarks conducted." If we compare the best results of the two cards, then Nvidia is still ahead. This does not undermine the wonderful work of the community which has managed to get the best performance from AMD cards. I myself own a 6800xt and I am delighted to get such results.