• 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.

CES 2025 AMD & Nvidia Keynote Live Chat [06/01/25]

So now that all of our emotions have subsided and we can reason again, what are we talking about here in reality (based on Nvidias relative performance charts at least, excluding the MFG 4 x frames sorcery)? 20-30% uplift in good ol' fashioned raster performance?
 
yes around 30% improvement for the 5090
edit: i will go by the timespy benchmarks till we are on dx12, as a fair indicator of performance
So it looks like the cards are roughly a tier up from their 40 series counterparts. 5070 = 4070 ti etc. Be interested in a proper comparison between the uplift in performance and changes in pricing between this generation and the last, as well as 30-40 series change. Obviously too soon for that but I'd be interested to see how this gen's price/performance compares to the last couple of generations. In terms of marketing, Nvidia are certainly massaging those figures with all of the AI fake-framery.
 
yes around 30% improvement for the 5090
edit: i will go by the timespy benchmarks till we are on dx12, as a fair indicator of performance


I have the numbers Jensen refused to show you

RTX 4090: 83 Teraflops
RTX 5090: 104 teraflops

Performance improvement: 25%

You can take the RTX4090's Timespy score and multiply by 1.25 to get the RTX5090's Timespy score, and you can take that to the bank boo-yeah
 
Last edited:
I have the numbers Jensen refused to show you

RTX 4090: 83 Teraflops
RTX 5090: 104 teraflops

Performance improvement: 25%

You can take the RTX4090's Timespy score and multiply by 1.25 to get the RTX5090's Timespy score, and you can take that to the bank boo-yeah

He can't refuse anything. Reviews will be out before the card will which will inevitably show you what's what.

Anyone with half a brain understands he is using MFG to compare cards.

You know what else we can take to the bank? Grim will soon be running a second tier card. How delicious that will be after all your high and mighty posts over the years :D

Boo-yahh baby! :cry:
 
Last edited:
We're getting to the point though, with both the graphical demands from games, the ever chasing of higher frame rates and more advanced monitor tech, that it sort of has to exist. You can only brute force raster so much and if the goal is always more then they'll find a way to do that and then each generation hopefully improve on how that works.
no it doesn't ..... they purposely make the graphiucs crap sop you have to use dlss and other nonsense to cover up how terrible they really are.

there's a pretty viral video about it that you may have missed


graphics have gone almost no where in 10 years because they only care about dlss , frame gen and other gimmicks


who needs frame gen on a 5090? thgat should be for the budget card at the bottom that's willing to sacrifice IQ
 
Hardware Unboxed's latest video has them estimating a 20% cost-per-frame improvement over 40 Super series (skip to the 9:33 mark in the video for his explanation):

<iframe width="560" height="315" src="
" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen></iframe>
 
Last edited:
Also worth noting Digital Foundry got a hands on with the 5080 and said despite there being 2 additional frames generated. (2x fg vs 4x fg) the latency wasn't noticeably worse.
6.4ms of added latency for 71% extra frames being generated

50.9ms average (2xfg) vs 57.3ms average (4xfg)
That's telling nothing so all unless we know exact input FPS. Also, let's not forget it's being tested on non-public version of cp2077, very well optimised by Nvidia engineers as with all the other things - it's their flagship tech demo for years now. And then we have IJ games where fg cause horrible ghosting and attracts on hands when running and general latency makes it near unplayable for me even though input FPS is fine - implementation matters, it seems. We'll see it in practice in other games, when is actually released to the masses.
 
who needs frame gen on a 5090? thgat should be for the budget card at the bottom that's willing to sacrifice IQ
This is the funny bit about FG - it mostly works well on higher end cards, as it requires minimum input FPS being close to 60, otherwise it's way too laggy to use. However, we have now monitors of 400Hz+ and in such case it has its use. But to think it's to bring playable FPS from 30 to 120 is just silly - unless one loves feeling sick all the time.
 
So yes having had a day out in the car with my thoughts and some youtube playing in the background, the initial wow factor for the 2 slot cooler design and performance isn't as wow as it was at 02:30 AM and I am more excited about the DLSS 4 model updates coming to 40 series shortly instead.

This also means a 5080 is nowhere near 4090 performance either when factoring in the above. So other than the initial "omg 40 series used prices plummet" panic for no real reason, I don't think it's ultimately going to affect the 4090 used prices much given the VRAM USP and relative raw power still.

So unless reviews show something completely different and new revelations show the 5090 is the must have, then na I'll stick with the 4090 thanks.
For once I can say we are in full agreement with opinions here. I'm just more excited about software updates and proper AI models added to dlss etc. than me hardware which brings little new as shown so far.
 
  • Like
Reactions: mrk
Good video which looks at each card 1 by 1 compares to the equivalent 4000 series. Basically outside of the 4090 to 5090 and the AI performance increase across the stack, the cards aren't really a big jump at all. In the 5070 case, the 4070S actually looks like it may out perform it (4090 performance, my arse).

 
Last edited:
watching that video made me think the 5070 just needs a zip and it looks like 90s pencil case

the finish on it looks like some weird cheap thing, all bloated and rounded looking.

doesnt make me think of the future at all, its like the the past coming back


be interesting watching someone play dark souls or something with the new framegen, see if the extra frames make dodging, parrying etc easier or harder, since they aren't really real frames
 
Last edited:
With Hardware Unboxed, they've quoted AMD saying the performance leaks are completely inaccurate

Now people have been assuming it's around 7900XT performance (Which I've read here for ages) whereas on the HUB video it mentioned about a 7800XT type performance in regards to leaks
 
@TNA will be amused by this because it will prove his theory wrong but... I am staying with my 4090 :p

I've been on the road all day today as was attending a press event for some new devices and came home with a free watch, phone and buds so not sad about missing all the action here hah.

On the drive there and back had the chance to catch up with some youtube from LTT, Digital Foundry (who actually showed their own hands on as well as LTT), Gamers Nexus and Mr floaty head himself who @Neil79 is a big fan of, Daniel Owen, as well as a take by one of my fv tubers, Optimum Tech:




The takeaway is this, with some of my own thoughts included.

Everything Nvidia showed was obviously with the massive dose of PR spin, all of the metrics were essentially with DLSS 4 FG applied with MFG, and the two games that were only DLSS3, the 5090 raw uplift over a 4090 is far below 50% - It remains to be seen what the no Frame Gen differences are until reviewers get their hands on them. Also the 40 series charts are running DLSS 3 not 4, note that Nvidia state that all RTX cards are getting all of the Enhanced DLSS updates, MFG is the only thing exclusive to the 50 series, so if 240fps is what you need/want, then 50 is the only option, though nobody has a single bit of material showing what the framerate and IQ gains are yet on 40 series with the new Enhanced DLSS4 update, because there absolutely are gains as shown by Digital Foundry and stated by Nvidia. And it is in Nvidia's best interest to not show the 4090 with Enhanced DLSS as the performance gap would be much smaller still most likely in games not using MFG.

With that in mind I am more inclined to skip 50 series entirely and ride out the next year or two with the 4090 with confidence that there are new framerate gains and image quality gains coming very shortly, it's not like performance is lacking or anything lol.

I do want that 2 slot cooler though, but we don't know what the noise that setup will make given it has to control up to 575 watts of TDP even if it's going to be 200W below that in actual games when not being maxed out in benchmarks or DLAA path tracing, otherwise even the 4090 exceeds 400W quite easily, just play Control at 4K DLAA and witness the fans spinning at 2000rpm+ and approx 440 watts being drawn by it., laugh and then enable DLSS Quality and instantly see a 200W drop.

For me to get the 5090 I want to see at the minimum, 50% uplift in raw performance over the 4090, nothing we see right now shows anywhere close to that lol.

So yes having had a day out in the car with my thoughts and some youtube playing in the background, the initial wow factor for the 2 slot cooler design and performance isn't as wow as it was at 02:30 AM and I am more excited about the DLSS 4 model updates coming to 40 series shortly instead.

This also means a 5080 is nowhere near 4090 performance either when factoring in the above. So other than the initial "omg 40 series used prices plummet" panic for no real reason, I don't think it's ultimately going to affect the 4090 used prices much given the VRAM USP and relative raw power still.

So unless reviews show something completely different and new revelations show the 5090 is the must have, then na I'll stick with the 4090 thanks.
I'm starting to agree tbh. I'm gonna wait for the proper hands on reviews before making my choice. I don't mind all the extra frame gen stuff, its cool and it works. But its not enough to make me want to shell out 2k. The 4090 I have is still really good and if I can expect slightly better performance then great. I'm happy to be proved wrong but this generation does seem like a bit of a let down at the high end.
 
This is the funny bit about FG - it mostly works well on higher end cards, as it requires minimum input FPS being close to 60, otherwise it's way too laggy to use. However, we have now monitors of 400Hz+ and in such case it has its use. But to think it's to bring playable FPS from 30 to 120 is just silly - unless one loves feeling sick all the time.
I agree with this.
 
With multi frame gen being the only real upgrade from the looks of things. Whats really stopping someone with the skills from finding a way to enable it on 40 series cards? Like I get why Nvidia has locked it, but for real as a 4090 owner if you take that upgrade away it really doesn't feel like much of a jump for the asking price?
 
We did this dance last Gen with everyone doubting the 4090 performance in raster based on the slides.

Nvidia probably just want to work on the drivers more and plan on a massive positive push because it's raw performance is a "surprise" when the reviews hit.
 
We did this dance last Gen with everyone doubting the 4090 performance in raster based on the slides.

Nvidia probably just want to work on the drivers more and plan on a massive positive push because it's raw performance is a "surprise" when the reviews hit.
Last gen went from Samsung 8nm(10nm TSMC really 30 series) to TSMC 4nm (40 series) ... 50 series is still 4nm... Check the transistor counts and be shocked..

The dance is justified this time.. 60 series and a new node shrink again and we will see a real jump again if Nvidia doesn't gimp the silicon size.
 
Last gen went from Samsung 8nm(10nm TSMC really 30 series) to TSMC 4nm (40 series) ... 50 series is still 4nm... Check the transistor counts and be shocked..

The dance is justified this time.. 60 series and a new node shrink again and we will see a real jump again if Nvidia doesn't gimp the silicon size.

I'll look at the reviews the day before I guess. If a 5080 is mostly 4090 performance with a grand in my pocket, that may be the better option after all.
 
Back
Top Bottom