• 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.

;)

The gap between Radeon VII and 2080Ti is 35-40%. That's the gap to close.

The Radeon VII is 331mm^2 and doesn't use the better shaders in the 5700XT.

Remember the 2080ti die size is is 775mm^2.

It's possible but it is likely power consumption will get in the way. At 500mm^2 it might get close. Nvidia also wasted a lot of die space with RT cores and stuff, so they should be able to respond easily if they give up on that.
NV is not even on 7nm yet. They won't need to give up on RT to respond and won't do that anyway, they're well ahead I think. If anything AMD need to get RT on their cards and they may then go through the pain NV have already gone through with a smaller step in raw FPS increase when they first implement RT in hardware.
Overall I think AMD are a good few years behind Nvidia. Like with any tech, getting that extra few % can be so difficult too. 35%-40% is a huge gap bearing in mind the RT capabilities NV have stuffed on their cards, even if it's still early days for the tech.
Hopefully as the profits roll in from the CPU sales AMD can get more R&D done on their GPU's too. We know they've got the potential
 
Last edited:
He's almost right, the base architecture of Maxwell/Pascal/Turing is hitting bottlenecks but Ampere seems like it's a fresh start so should prove fruitful :)
Yep. They have loads of money and am sure been pouring a lot into R&D. Ampere will likely bring in a decent amount of performance together with 7nm. I think the 3070 will be a good card to get.

Though I might get a 3080Ti from members market to enjoy a few months of Cyberpunk etc. Depends on if PS5 is out around then and how I get on with it.
 
You were right, I was wrong. I now also don't see how AMD will find 60% more performance over the Navi 10 to equal RTX 2080 Ti.
I was confused, I thought the missing performance is between 30 and 40%.


https://www.techpowerup.com/review/amd-radeon-rx-5700-xt/28.html

AMD's next high end card likely won't even be RDNA, but instead RDNA2. They've confirmed that their high end will have hardware raytracing acceleration, and their first architecture to include hardware acceleration will be RDNA2.

An Nvidia style backwards launch would make sense for them. High end RDNA2 on 7nm+ with HBM2/3 around Feb-April, then mid range RDNA2 with GDDR6 to replace the 5700/5700XT around June-August. Then finally low end GPUs around Sep-Nov.
 
AMD's next high end card likely won't even be RDNA, but instead RDNA2. They've confirmed that their high end will have hardware raytracing acceleration, and their first architecture to include hardware acceleration will be RDNA2.

An Nvidia style backwards launch would make sense for them. High end RDNA2 on 7nm+ with HBM2/3 around Feb-April, then mid range RDNA2 with GDDR6 to replace the 5700/5700XT around June-August. Then finally low end GPUs around Sep-Nov.

Navi is a GCN-RDNA hybrid, the next one will be RDNA.

I'm curious to know how the limits that have plagued GCN (64CU-4096 bus) are changed by the new arch, or if it's even relevant.
 
Navi is a GCN-RDNA hybrid, the next one will be RDNA.

I'm curious to know how the limits that have plagued GCN (64CU-4096 bus) are changed by the new arch, or if it's even relevant.
Is it tho? I read somewhere that this isn't really the case.

All AMD cards for the foreseeable future will be an implementation of the GCN ISA. But allegedly RDNA is an all-new/mostly-new implementation of that ISA.

I'm not fully clued up on what precisely the GCN ISA is, but apparently it's similar in concept to the x86 instruction set. Although as with that, it has apparently evolved over the years in addition to the products implementing it having evolved also.

So sayeth the wise Alo--- Reddit.
 
Navi is a GCN-RDNA hybrid, the next one will be RDNA.

I'm curious to know how the limits that have plagued GCN (64CU-4096 bus) are changed by the new arch, or if it's even relevant.
Where you getting it's a GCN hybrid? Everything I have seen is it's all new
 
It's semantics, whether or not there's enough changes from GCN to be called a new arch is up to you, AMD say's it is in their as usual embellished slides, but realistically N20 is where it's supposed to be next year.

Believe it or not, it doesn't really matter.
 
It's semantics, whether or not there's enough changes from GCN to be called a new arch is up to you, AMD say's it is in their as usual embellished slides, but realistically N20 is where it's supposed to be next year.

Believe it or not, it doesn't really matter.
Yeah I thought the same thing until reading that internally AMD view GCN as an ISA, not a micro architecture. And that they are not the same thing.

So RDNA is a GCN (ISA) implementation. Navi 20 will be a GCN implementation also.
 
I very much doubt RDNA is all new in navi given it has the same PPW, heat and scaling problems of GCN, just not nearly as bad as Polaris or Vega (remember they said it was radically different GCN with this one too). Smells like a GCN1.6 to me.

I can't see how Navi would scale up to a bigger chip without HBM2 again.
 
Last edited:
As a side note I think the console contracts are now holding back there GPU designs, as anything "new" they come up can't be radically different fro compatibility and developer reasons ie, game and console dev's don't want to work with radically different architecture after building up skills and knowledge for year.

So I think the can or could do a massive shift in architecture which is needed like they did with Bulldozer to ryzen in the CPU space
 
Tbh the RT performance for the price is quite convincing now. That being said, I'm still looking for whatever's coming next year instead.

4K
index.php

461164bc-0e6b-481e-beaf-2d424daff9d5.png

index.php
 
Back
Top Bottom