• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Next Gen GPU's will be even more expensive

Status
Not open for further replies.
RDNA 4 was the chance for AMD to clinch some serious portion of the market share back. They needed a big Navi 41 on 3nm, in which case they would have a much faster card than the old 5nm RTX 5090.


No one knows how RTX 5000 series will perform

20% faster than RTX 4090.
 
No one knows how RTX 5000 series will perform either, all we have are clickbait and rumors. Are only positive feelings allowed in Nvidia threads?
Because Nvidia usually gives a pretty good bump gen over gen. Be it graphics, compute, or both.
But interesting you say that I've been seeing plenty of thread crapping on AMD topics.
Yeah.. because there are man-children all over. Seriously I hope some of these guys don't talk like this in real life face to face to actual people.
 
The 5090 probably will be, 512bit memory interface, 32GB of gddr7, massive die etc. 2k at least. This cycles version of 8800 ultra?

Mid range chips at high prices won't shift much in this environment though, same reason 7900xt being discounted every week and lots of stock of 700+ chips generally.
 
They will be an additional 60% more expensive. ;)
 
The 5090 probably will be, 512bit memory interface, 32GB of gddr7, massive die etc. 2k at least. This cycles version of 8800 ultra?
More like G(T)200 a.k.a. GTX 280.
 
Well, it don't matter what the prices are on the new cards, cause I had my AI buyer bot thingy buy all of them already, everywhere, all the time, 50, 60, 70, 80 & 90 series, & the TI, FE, SuperDuperMegaOverclockedUberPlatnium/Gold/Titanium editions too.....

And yea, I got a bulk-order discount, but even so, the 5090's were STILL almost $2.75k each :(

So if you want one, send me a PM around Q2/25, and I'll be glad to hook ya up 4 sure, hehehe :D



/s
 
Because Nvidia usually gives a pretty good bump gen over gen. Be it graphics, compute, or both.
I don't have any expectation for Jensen to be generous enough with an xx80 card faster than the 4090, maybe with a Super version. The $1500-1600 price seems realistic enough, if it would beat a 4090, and if the 5080 only beats the 4090 with whatever next gen AI DLSS RTX I think the prices will be higher than the 4080.

But I can understand how passionate people can be for Nvidia, its all I used to buy. The anticipation and launches were exciting, but not after the GTX 1000 series, performance/price value was gone with the RTX cards. Prices increasing with every gen is one reason why I've lost any excitement I had for new gpu's.
Yeah.. because there are man-children all over. Seriously I hope some of these guys don't talk like this in real life face to face to actual people.
I completely agree.
 
But I can understand how passionate people can be for Nvidia, its all I used to buy.
Likewise.

But I have no passion for hardware. That died a long time ago.

I do like to push hardware to see what it can do, or if I can do what someone else did, but that's about it.
 
Could history repeat itself?

From line-up standpoint, they (NV and AMD) are in similar situation as back then (NV vs. ATI).
NV made biggest GPU possible, and ATI competed with "sweet spot" strategy (HD 4000 series [RV770]).

I however doubt a repeat is possible now. Top SKUs from HD 4800 series used GDDR5 (new memory type) vs. GDDR3 NV used. Now AMD is limited to GDDR6 (I assume GDDR7 is in some kind of "exclusive" deal of sorts for NV [at least for now]), so AMD can't offset bandwidth advantage new memory type gives.
Flip side of that is G6 memory should be cheap (relatively) and abundant, which can give AMD a bit of edge in margin/price. Q : Can AMD now use G6X memory since NV moved on for G7, or is that still "exclusive" for NV ?

Lastly... there is R&D money problem.
I wouldn't be surprised if Nvidia current R&D GPU budget was at whole AMD worth scale (or more).
 
Last edited:
  • Like
Reactions: N/A
RX 7900 XTX costs 50% of RTX 4090, and still wins the performance benchmarks. I don't know how people can be so stupid to buy RTX 4090.

1735342383746.png
1735342398644.png


 
RX 7900 XTX costs 50% of RTX 4090, and still wins the performance benchmarks. I don't know how people can be so stupid to buy RTX 4090.

View attachment 377433View attachment 377434

Sampling the video it looks though the frame times are more consistent with the RX7900 XTX. To me that says better quality gameplay (not necessarily visuals) / less jerkiness / less stutter. Maybe I'm wrong as I live in the low end 60FPS gaming segment.

The video opening screenshot is reversed to the benchmark video where the comparison is swapped. I had to edit my comment once I noticed the difference.
Looks like 4090 wins in performance.
Looks like RX 7900XTX wins smoother FPS.
 
Last edited:
20% faster than RTX 4090.

20% faster than the 4090 is a conservative estimate for the 5080, it would make GB203 and AD102 full die versions relatively equivalent

The 5090 will be much faster, and yes, I warned before, AMD fans better brace for a bloodbath complete with a ritual sacrifice. It will have an entry price to match. Whatever, if not at launch I can surely buy one a few months down the road... it's the price to pay when the competition fumbles cycle after cycle.

RX 7900 XTX costs 50% of RTX 4090, and still wins the performance benchmarks. I don't know how people can be so stupid to buy RTX 4090.

View attachment 377433View attachment 377434


I want to know in what fantasy world does the 7900XTX win against the 4090
 
Supposed leak from an Aus retailer about 5080 pricing, Asus Prime 5080 $2500 AUS cost price, $2750 MSRP, - sales tax and to reflect $ US roughly $1500-$1600 end user pricing, video is obviously speculation/to be taken with a pinch of salt but doesn't seem unlikely

 
From line-up standpoint, they (NV and AMD) are in similar situation as back then (NV vs. ATI).
NV made biggest GPU possible, and ATI competed with "sweet spot" strategy (HD 4000 series [RV770]).

I however doubt a repeat is possible now. Top SKUs from HD 4800 series used GDDR5 (new memory type) vs. GDDR3 NV used. Now AMD is limited to GDDR6 (I assume GDDR7 is in some kind of "exclusive" deal of sorts for NV [at least for now]), so AMD can't offset bandwidth advantage new memory type gives.
Flip side of that is G6 memory should be cheap (relatively) and abundant, which can give AMD a bit of edge in margin/price. Q : Can AMD now use G6X memory since NV moved on for G7, or is that still "exclusive" for NV ?

Lastly... there is R&D money problem.
I wouldn't be surprised if Nvidia current R&D GPU budget was at whole AMD worth scale (or more).
The only way AMD could get a bandwidth advantage is using HBM memory, unfortunately all of it gets allocated for datacenter hardware. And I really hate the exclusive deals NV has with memory, it seems like anti-competition to me.
As for R&D, I wouldn't doubt it though hopefully unifying consumer and compute into UDNA saves AMD some R&D cost.
Sampling the video it looks though the frame times are more consistent with the RX7900 XTX. To me that says better quality gameplay (not necessarily visuals) / less jerkiness / less stutter. Maybe I'm wrong as I live in the low end 60FPS gaming segment.

The video opening screenshot is reversed to the benchmark video where the comparison is swapped. I had to edit my comment once I noticed the difference.
Looks like 4090 wins in performance.
Looks like RX 7900XTX wins smoother FPS.
Is there any particular reason the 7900XTX has smoother FPS? Lower driver overhead? It looks about the same to me, although I notice input lag a lot more than I do some subtle stuttering, which is why I prefer native resolution over any sort of upscaling aside from the visual quality.
 
20% faster than the 4090 is a conservative estimate for the 5080

RTX 5080 with only 10k shaders will be slower than RTX 4090. It has an increase of only 10%.

Sampling the video it looks though the frame times are more consistent with the RX7900 XTX. To me that says better quality gameplay (not necessarily visuals) / less jerkiness / less stutter. Maybe I'm wrong as I live in the low end 60FPS gaming segment.

The video opening screenshot is reversed to the benchmark video where the comparison is swapped. I had to edit my comment once I noticed the difference.
Looks like 4090 wins in performance.
Looks like RX 7900XTX wins smoother FPS.

Is there any particular reason the 7900XTX has smoother FPS? Lower driver overhead? It looks about the same to me, although I notice input lag a lot more than I do some subtle stuttering, which is why I prefer native resolution over any sort of upscaling aside from the visual quality.

Low system and graphics memory utilization.
Many users complain about the Nvidia driver not working properly.






 
AMD fans better brace for a bloodbath complete with a ritual sacrifice.
I will assume that you will pit a 5090 with a price tag of maybe 3k against a low end rdna4 with a price tag of maybe 300 bucks. Then yeah, it will be a bloodbath, ritual sacrifice, the devil himself will come to collect amd soul, etc.

I keep wishing at this point that amd drops out of the dGPU market, feed whatever is left of the gamers to Ngreedia and concentrate on the corporate world, since no matter what they do, its never good at all.
 
I want to know in what fantasy world does the 7900XTX win against the 4090
Price to performance ratio.

It doesn't realistically beat a 4090, it's more between a 4080 and a 4090, but... it costs nearly half what the 4090 costs.

I'd be lying if I said I haven't been very pleased with mine, and that's coming from a 3090ti not long before.
 
Price to performance ratio.

It doesn't realistically beat a 4090, it's more between a 4080 and a 4090, but... it costs nearly half what the 4090 costs.

I'd be lying if I said I haven't been very pleased with mine, and that's coming from a 3090ti not long before.

I mean, it is lighter on the wallet, but it has to be I guess

I will assume that you will pit a 5090 with a price tag of maybe 3k against a low end rdna4 with a price tag of maybe 300 bucks. Then yeah, it will be a bloodbath, ritual sacrifice, the devil himself will come to collect amd soul, etc.

I keep wishing at this point that amd drops out of the dGPU market, feed whatever is left of the gamers to Ngreedia and concentrate on the corporate world, since no matter what they do, its never good at all.

Then assume that... doesn't mean I will do it... and wishful thinking a GPU with 4080 like performance is gonna cost 300 bucks, you're setting yourself up for a disappointment ;)

RTX 5080 with only 10k shaders will be slower than RTX 4090. It has an increase of only 10%.

Where did you get the "only 10% IPC over Ada" figure?
 
I will assume that you will pit a 5090 with a price tag of maybe 3k against a low end rdna4 with a price tag of maybe 300 bucks. Then yeah, it will be a bloodbath, ritual sacrifice, the devil himself will come to collect amd soul, etc.

I keep wishing at this point that amd drops out of the dGPU market, feed whatever is left of the gamers to Ngreedia and concentrate on the corporate world, since no matter what they do, its never good at all.
I assume reviewers will do the same and not even talk about price to performance, its what most people consider when buying a GPU, at least those buying in the mid range.
And I wouldn't even be disappointed if AMD does leave the dGPU market at this point, it doesn't make any financial sense to keep trying, then the media and the loud Nvidia fans keep influencing gamers to buy the Ngreedia card.
AMD fans better brace for a bloodbath complete with a ritual sacrifice. It will have an entry price to match. Whatever, if not at launch I can surely buy one a few months down the road... it's the price to pay when the competition fumbles cycle after cycle.
Nvidia fans had better brace their wallets for a $2500+ 5090, more realistically at least $3000 because of the 32GB of VRAM, and I'm sure Nvidia will put the AI tax on it with the rumors of new AI features, which will probably only be available on the RTX 50 series. And I'll say it again because the Nvidia fans always ignore it, it isn't the fault of the competition when Nvidia buyers will keep buying Nvidia no matter what the leather jacket man does.
There is no bloodbath though, Nvidia fans have already gotten their wish after ridiculing the competition, of having no competition at all.
 
Last edited:
nothing forces you to run games in native res, the same way we dont have (and watch) all the video content in the same res than moni/tv.

i can run games in FHD/QHD just fine, but still enjoy a larger screen (50), with no screen-door effect, that i had while running games in native res on a 32" (1440p), at same distance (2-3ft).
ignoring you dont even take games into account, e.g. if i have games that are +5y old, its fine to run 4K with high IQ settings, even mu 2080S can do that.

so no, a 4k screen doesnt mean you need a faster gpu.
the game and used res (image quality settings) determine that.
Plus 4K screen does 4:1 pixel binning if you run at 1080p, so no issues there.

RX 7900 XTX costs 50% of RTX 4090, and still wins the performance benchmarks. I don't know how people can be so stupid to buy RTX 4090.

View attachment 377433View attachment 377434

In that comparison 0.1% lows are almost double on the 4090, 62 vs 37 FPS. Average that's only 20 lower on the XTX, but dips to 1/3 of average aren't fun, I'd prefer a locked 120 FPS with no fluctuations, than a 240 FPS average with dips to 60 FPS. It's immersion breaking.
 
Last edited:
5080 for $1500 is complete nonsense, I don't believe it, just like I don't believe the 5080 will be faster or even very close to the 4090. The 4080 was significantly faster than the 3080(+50%), and as it turns out, significantly more expensive. The 5080 will definitely not outperform the 4080 to the same extent as the 4080 did against the 3080. +20% maximum, i think.
 
RDNA 4 was the chance for AMD to clinch some serious portion of the market share back. They needed a big Navi 41 on 3nm, in which case they would have a much faster card than the old 5nm RTX 5090.
I disagree. Very few people need a $1500+ GPU. Focusing on the market where people actually buy things is the right choice, imo.
 
5080 for $1500 is complete nonsense, I don't believe it, just like I don't believe the 5080 will be faster or even very close to the 4090. The 4080 was significantly faster than the 3080(+50%), and as it turns out, significantly more expensive. The 5080 will definitely not outperform the 4080 to the same extent as the 4080 did against the 3080. +20% maximum, i think.
Even with such numbers in mind, the 5080 for 1500USD, imagine the cost of a ROG version! will push the 5090 to be 2699USD for MSRP? these are car prices there not GPUs anymore :( many factors are coming together that led the price to be so, but still, I can't imagine a pure gamer willing to have the best GPU to pay this amount! this is like you're asking your friend for the best car, and he just tells you: Get a Ferrari! it's future proof!!
 
I disagree. Very few people need a $1500+ GPU. Focusing on the market where people actually buy things is the right choice, imo.

what does focusing mean? why couldn't they have the 7900xtx (their most sold card according to steam) and focus on the "market where people actually buy things", they just had to price the cards better, that's were they failed.

i don't get this focus argument at all. If priced better they could even outsell Nvidia.

There are no bad products, only bad prices​

 
Status
Not open for further replies.
Back
Top