[SOLVED] Is there any Chance AMD will be able to produce the fastest card?

timf79

Distinguished
May 10, 2011
82
3
18,635
I am thinking to upgrade my TitanXP with the fastest gaming card available.
Wondering if there is any chance AMD will be able to beat the 3090 performance.
In the past AMD was able to get close of the mainstream cards 70s, 60s, 50s, but never the toptier 80s or the halo cards (titan's).
I have seen that BigNavi is expected to surpass the 2080TI by 30%, which would put it in TITAN RTX territory, but not close to 3090..

Wanted to foster the community for your opinion and thoughts on who will make the fastest gaming card in 2020/2021?
 
Solution
Also, the HD 7970 beat the GTX 580 upon release and had the top of the consumer field for about four months (the 590 was a dual GPU, still a thing back then, so doesn't really count) before the 680 came out. The 680 comfortable bested the 7970, but interestingly enough, the 2 GB of VRAM caused it to age far worse than the 7970 did; while the 680 kept beating the 7970 in the 2012 games, the VRAM limitations caused the 7970 to beat the 680 in the majority of the 2015/2016 games (Anand, I believe, did a look). That is, if your 7970 was still working; that thing required the power of a small city.

As sizzling notes, AMD/Nvidia traded blows for the next generation too, until AMD backed off from the flagship battle.

So I wouldn't expect...
Its always possible, but lately AMD hasnt been targeting the super high end market.
Why sell 5 flagships cards when you can sell 5000 budget ones instead?

Also, your historical research isnt quite right, for example the R9 290x outperformed every Nvidia GPU at launch, including the original TITAN.
Nvidia quickly released the 780Ti to respond, but still.
 
About 4-5 years ago AMD said their focus was low-mid range cards as that is where volume sales are and there is less risk. The problem with developing a market leading product which demands a high price is if you get it slightly wrong or it’s not the clear winner then it won’t sell and you won’t recoup the development costs. Unless AMD take a new strategy to gpu’s off the back of their cpu success I still think it’s very unlikely they will target 3090 level and maybe not even 3080 level.
 

DSzymborski

Titan
Moderator
Also, the HD 7970 beat the GTX 580 upon release and had the top of the consumer field for about four months (the 590 was a dual GPU, still a thing back then, so doesn't really count) before the 680 came out. The 680 comfortable bested the 7970, but interestingly enough, the 2 GB of VRAM caused it to age far worse than the 7970 did; while the 680 kept beating the 7970 in the 2012 games, the VRAM limitations caused the 7970 to beat the 680 in the majority of the 2015/2016 games (Anand, I believe, did a look). That is, if your 7970 was still working; that thing required the power of a small city.

As sizzling notes, AMD/Nvidia traded blows for the next generation too, until AMD backed off from the flagship battle.

So I wouldn't expect the next top AMD GPU to be on the same tier as the top of the Nvidia product stack. AMD's dipped their toes in that water, but given the experience with the ultimately disappointing Radeon VII, I can't see them jumping full-on into that fight again. And even if they did, the AMD solution that matches the Nvidia solution would likely require a lot more power than the comparable Nvidia one. AMD's ultimately a smaller company than Nvidia or Intel, so they try and pick where they get into fights.
 
  • Like
Reactions: digitalgriffin
Solution

timf79

Distinguished
May 10, 2011
82
3
18,635
Its always possible, but lately AMD hasnt been targeting the super high end market.
Why sell 5 flagships cards when you can sell 5000 budget ones instead?

Also, your historical research isnt quite right, for example the R9 290x outperformed every Nvidia GPU at launch, including the original TITAN.
Nvidia quickly released the 780Ti to respond, but still.
Sorry, I should have defined in most recent history (past 5 years)
 

timf79

Distinguished
May 10, 2011
82
3
18,635
Thanks all for the feedback, I see there is a general consensus based on the AMD strategy.
Economically I am not sure cards like the 3090 are generating profit.
 

mjbn1977

Distinguished
According to all recent rumors they will not have the fasted cards. Big Navi will most likely compete with the RTX 3080 or maybe even beat it slightly. We need to see. But the RTX 3080 and RTX 3090 line-up suggest that Nvidia doens't think they will be beat on the RTX 3090 but they're leaving nice update room on the RTX 3080. If the Big Navi turns out faster than RTX 3080, they can always come out in spring with a RTX 3080Ti.....
 
According to all recent rumors they will not have the fasted cards. Big Navi will most likely compete with the RTX 3080 or maybe even beat it slightly. We need to see. But the RTX 3080 and RTX 3090 line-up suggest that Nvidia doens't think they will be beat on the RTX 3090 but they're leaving nice update room on the RTX 3080. If the Big Navi turns out faster than RTX 3080, they can always come out in spring with a RTX 3080Ti.....

That's fine by me. 3090 is a power hog and if it's $1500+

I'm not even close in being interested. But I'm happy it created competition. Find me a nice $1000 Big Navi and I'm in.

That said it will be closer than people thing. Samsung's node is vastly inferior from what I understand.
 

mjbn1977

Distinguished
That's fine by me. 3090 is a power hog and if it's $1500+

I'm not even close in being interested. But I'm happy it created competition. Find me a nice $1000 Big Navi and I'm in.

That said it will be closer than people thing. Samsung's node is vastly inferior from what I understand.

It seems 3080 and 3090 are produced in TSMC 7nm, only the "smaller" cards are produced in 8nm from Samsung. Rumors from Friday....