• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

$2500 5090 I’m out

Leaks suggest the RTX-5090 will start @ $1,900 and up.

F this.

I'm not a console gamer, but for that price you can buy a brand new PS5 and a new OLED 4K TV.

I'm a big PC Gamer and my current RTX-4080 I got brand new for $999 runs all my games ultra max highest settings with ease.

This is a bad precedent to to have such outrageously expensive video cards. I guess fools are easily parted from their money.
For gaming, I agree, the 5090 is a terrible value.
NVIDIA isn't targeting gamers with the 5090, though.
 
I love how if you're is willing to spend money on your hobby that you enjoy then that automatically makes you some fool chasing e peen.
I agree as well. PC's are the only hobby that has kept my interest over the years. People will have their opinions on things no matter what so I try to block out the "noise" and do what makes me happy.
 
Any word if Nvidia redesigned the power connector?

I recall a couple years ago all these scary issues of power supplies burning out the cards or the connection, and that always seemed like a loose wobbly connector not totally secure.

They've had 2 years to correct it. Hopefully it's a new setup.
I've had no issue with the original 12VHPWR connector using the included 4 8-pin PCI-E adapter included with my card. I occasionally check to see if it's gotten loose and it's still solid as a rock.

That being said the connector, was redesigned a year ago. It is now called 12V-2x6. Also, the connector was designed by PCI-SIG, not NVIDIA. NVIDIA's proprietary connector was a 12-pin that was featured on the RTX 3000 series that was a stopgap.
 
The situation is what it is. Blame AMD for dropping the ball, not Nvidia for producing superior products and capitalizing on it.
I think I did do that. ;)

For gaming, I agree, the 5090 is a terrible value.
NVIDIA isn't targeting gamers with the 5090, though.
Then they should call it a Titan again and then probably hike the price some more frankly since X02 chips in any gen can command much more $$$ in non-GeForce products.
 
Last edited:
To be honest they could put a $4000 price tag on it and it'd sell out anyway because there's millions of people that HAVE to have the best card no matter what. I've always looked more at value, IMO I think the 4070 Ti super is the best bang for the buck right now.
 
Some are actually delusional enough to think the 4090 will drop to $800 or something once the 5090 comes out lol.
What is a 4090 worth right now?

And what will it be worth when the 5090 drops?

I have an extra 4090 in my backup rig that I should probably sell cause it's never used lol.

Seriously asking...
 
What is a 4090 worth right now?

And what will it be worth when the 5090 drops?

I have an extra 4090 in my backup rig that I should probably sell cause it's never used lol.

Seriously asking...

The latest sold listings on ebay from from 1800-2000. I can see it dropping a little bit when the 5090 launches but I would imagine they will still be highly sought after for AI n stuff that it won't drop to 800 for a long time.
 
What is a 4090 worth right now?

And what will it be worth when the 5090 drops?

I have an extra 4090 in my backup rig that I should probably sell cause it's never used lol.

Seriously asking...
It is the 5080 that will have the most influence on the 4090 price. If the 5080 is to the 4090 what the 3080 was to the 2080ti then the price of the 4090 will sink like a rock, but if it is more like most 4000 series cards in price/performance then the 4090 will maintain it's value quite well. So far there are lots of rumors, but we won't know the actual performance until the new cards drop so it is just guesswork until then.
 
As an eBay Associate, HardForum may earn from qualifying purchases.
The 3090 is a less expensive high RAM board so its frequently bought for AI, a factor in its resale value. As a gaming gpu the value is more questionable now.
 
The 3090 is a less expensive high RAM board so its frequently bought for AI, a factor in its resale value. As a gaming gpu the value is more questionable now.
True, but the 16gb in the 4080's has been a limiting factor for sure with gaming on 4k+. They should have given the 4080 or at least the 4080 super 20gb.
 
True, but the 16gb in the 4080's has been a limiting factor for sure with gaming on 4k+. They should have given the 4080 or at least the 4080 super 20gb.
Yeah quite a gulf which is only growing with 50-series between the 90 card and the 80 card. Just speaking in terms of core count and hell even memory bus.
 
new GPU will probably cost 30% - 40% higher coz of the tariff starting next year. Same thing for new consoles, expected around 40% higher.
 
new GPU will probably cost 30% - 40% higher coz of the tariff starting next year. Same thing for new consoles, expected around 40% higher.
I wanna see people blowing 3k on the 5090 lol who's gonna be the first person to post about it on [H} make sure Jensons infiniti pool is always hot so he can skinny dip with his harem lmao

As the 4090 climbs in value. Thanks Trump rofl...
 
True, but the 16gb in the 4080's has been a limiting factor for sure with gaming on 4k+. They should have given the 4080 or at least the 4080 super 20gb.
16GB is a limiting factor for 1440p if you want to use DLSS3 feature.

Only 1080p gets a clean pass for not exceeding 16GB VRAM usage on the most demanding RT titles.

So nvidia should have the courtesy to add minimum of 18-20GB on 70 and 80 series cards.

12GB is like an insult, why are they releasing a RT card that cannot play RT even on 1080p resolution because of lack of vram?
 
16GB is a limiting factor for 1440p if you want to use DLSS3 feature.

Only 1080p gets a clean pass for not exceeding 16GB VRAM usage on the most demanding RT titles.

So nvidia should have the courtesy to add minimum of 18-20GB on 70 and 80 series cards.

12GB is like an insult, why are they releasing a RT card that cannot play RT even on 1080p resolution because of lack of vram?

Well, well to be fair the 4080 was released in 2022 when 16gb was ''enough'' then but clearly wasn't futureproof enough. As I said, when they put out the 4080 super it should have included 20gb in that update.
 
The 3090 is a less expensive high RAM board so its frequently bought for AI, a factor in its resale value. As a gaming gpu the value is more questionable now.

3080 Ti's, which have roughly the same gaming performance, are much less expensive than the 3090. It's solely about the RAM, and probably for AI, yes. With DLSS, it does quite fine even in 4k. Most games where you would need DLSS are the ones where VRAM would be an issue anyway, so just for pure gaming I don't think a 3090 provides too much over a 3080 Ti.

I saw someone here say something about a 3090 being about the same as a 4080 performance wise, but that's a bit eyebrow raising. Haven't seen recent benchmarks, but I think especially a Super, on pure performance grunt, far outdoes a 3090.
 
I saw someone here say something about a 3090 being about the same as a 4080 performance wise, but that's a bit eyebrow raising. Haven't seen recent benchmarks, but I think especially a Super, on pure performance grunt, far outdoes a 3090.

For sure. A 4080 Super is still decently better than a 3090 Ti even.

1732385820414.png
 
Also forgot to point out that around that price range you can generally find open box or just some sort of deal on the 7900XTX, which has the same amount of VRAM and is just generally better for gaming (if we ignore DLSS vs FSR and all that jazz). It's just that afaik Nvidia has always been easier to set up for AI compute, so another point in the "it's just because of AI use" bucket for the current 3090 sale prices.
 
Also forgot to point out that around that price range you can generally find open box or just some sort of deal on the 7900XTX, which has the same amount of VRAM and is just generally better for gaming (if we ignore DLSS vs FSR and all that jazz). It's just that afaik Nvidia has always been easier to set up for AI compute, so another point in the "it's just because of AI use" bucket for the current 3090 sale prices.

DLSS is huge though. Kind of a hard feature to ignore. So many games utilize it and in the most recent iteration picture quality rivals native 4k with the Quality preset. I can hardly tell the difference at this point. In demanding modern games it makes all the difference for getting the most out of high refresh displays. I wanted AMD to succeed here and would love to buy their GPUs, but sadly they've dropped the ball and aren't competing anymore at the high end.
 
DLSS is huge though. Kind of a hard feature to ignore. So many games utilize it and in the most recent iteration picture quality rivals native 4k with the Quality preset. I can hardly tell the difference at this point. In demanding modern games it makes all the difference for getting the most out of high refresh displays. I wanted AMD to succeed here and would love to buy their GPUs, but sadly they've dropped the ball and aren't competing anymore at the high end.
Monitors are improving faster than chips, and will likely continue to do so for some time. Chips are closing in on the wall as far as feature size goes, while if you have good eyesight you can still see individual pixels on a screen with the naked eye. Pixels are enormous compared to transistors on a chip and have tons of room to get smaller. We'll get 8k screens, then 16k screens, and they'll take 4x and 16x as much processing power to render at native res compared to 4k, respectively. 4k is already about 2x 1440p. GPUs will not be able to keep up with screen tech, so we'll all be using scaling eventually.
 
DLSS is huge though. Kind of a hard feature to ignore. So many games utilize it and in the most recent iteration picture quality rivals native 4k with the Quality preset. I can hardly tell the difference at this point. In demanding modern games it makes all the difference for getting the most out of high refresh displays. I wanted AMD to succeed here and would love to buy their GPUs, but sadly they've dropped the ball and aren't competing anymore at the high end.
Yeah currently you absolutely need DLSS if you are going to play a game with true path tracing (which is the only way RT even makes sense).

Otherwise you're stuck at below 20FPS even on a 4090 lol.


Portal-RTX-benchmarks-1.png
Portal-RTX-benchmarks-3.png
 
DLSS is huge though. Kind of a hard feature to ignore. So many games utilize it and in the most recent iteration picture quality rivals native 4k with the Quality preset. I can hardly tell the difference at this point. In demanding modern games it makes all the difference for getting the most out of high refresh displays. I wanted AMD to succeed here and would love to buy their GPUs, but sadly they've dropped the ball and aren't competing anymore at the high end.

I mean sure, but I remember using FSR (on my 3080 Ti) in Starfield just fine. Sure there were some weird glitchy spots, but I don't know if it would really be a deal breaker for anyone. DLSS is better, but I'm sure FSR will keep improving over time. Maybe it won't quite reach feature parity, but at the price range it's kind of still a tossup between the 4080 Super (which technically is very rarely within that price range, actually) and the 7900 XTX (which is, and has more VRAM). A few friends, after I laid the facts out before them (and relatively unbiased, too; after all I have a 3080 Ti, 2080 before, that, and went 4090 this gen... I told them to do their own research, too), ended up all choosing the 7900XTX vs the 4080 Super. And I think 7900XTX vs the 3090 especially is just a no-brainer if they're at roughly the same price. The AMD chip just has more power for the money.
 
I posted this in the other thread cuz it had some information about the new high core count X 3D chips only having one CCD cache like the last generation at least that's what's rumored but it also has good information about some of the specs of the 5000 series and it looks like it's just a refresh like touring maybe just larger and terrible value this guy goes over some of the specs very interesting and kind of disappointing was just going to be more power hungry an extremely expensive and if the tarrifs kick in it's going to be 60% worse than what it was last series


View: https://youtu.be/HxyfpbUUrWc?si=hEwa9pJ_5aE79Nr8
 
Combine this with this article saying 4000 series stock is drying up.. Nothing is getting much of a price reduction / sale for the holidays and you better buy them all now now now!

CONSUME

https://www.techpowerup.com/329125/...rtx-40-series-gpus-could-be-in-shortage-in-q4

With the 4090s going up in price the 5090 seems like its going to be expensive or maybe even climb in price over time too. If someone is really wanting a 5090 then getting in the first wave may be the play. Seems like it worked out for the people that 'overpaid' 1600+ for a 4090 on launch.
 
Combine this with this article saying 4000 series stock is drying up.. Nothing is getting much of a price reduction / sale for the holidays and you better buy them all now now now!

CONSUME

https://www.techpowerup.com/329125/...rtx-40-series-gpus-could-be-in-shortage-in-q4

With the 4090s going up in price the 5090 seems like its going to be expensive or maybe even climb in price over time too. If someone is really wanting a 5090 then getting in the first wave may be the play. Seems like it worked out for the people that 'overpaid' 1600+ for a 4090 on launch.
Let's just hope and pray (or whatever you fancy) that there's not a crypto gpu hash mining resurgence or people won't see a 5k series card.
 
Combine this with this article saying 4000 series stock is drying up.. Nothing is getting much of a price reduction / sale for the holidays and you better buy them all now now now!

CONSUME

https://www.techpowerup.com/329125/...rtx-40-series-gpus-could-be-in-shortage-in-q4

With the 4090s going up in price the 5090 seems like its going to be expensive or maybe even climb in price over time too. If someone is really wanting a 5090 then getting in the first wave may be the play. Seems like it worked out for the people that 'overpaid' 1600+ for a 4090 on launch.
I really wish there was legit competition in the high end, and pressure in the rest of the range.

Rumors make is sound like NVIDIA is going to keep being stingy with VRAM this generation again.
 
16GB is a limiting factor for 1440p if you want to use DLSS3 feature.

Only 1080p gets a clean pass for not exceeding 16GB VRAM usage on the most demanding RT titles.

So nvidia should have the courtesy to add minimum of 18-20GB on 70 and 80 series cards.

12GB is like an insult, why are they releasing a RT card that cannot play RT even on 1080p resolution because of lack of vram?
No.

1732563996617.png


At 4K the 4080 is limited by its GPU, not VRAM.

1732564342458.png
 
Let's just hope and pray (or whatever you fancy) that there's not a crypto gpu hash mining resurgence or people won't see a 5k series card.
They will all be bought by scalpers with the 4090 stock limited this is going to be the 3080Ti and 3090 shortages all over again except much MUCH more expensive. I'm guessing we will see 5090 reselling on etailers and scalpers close to 3k dollars stick that in a pipe and smoke it yikes 😬 😳
 
No.

*image snipped*

At 4K the 4080 is limited by its GPU, not VRAM.
*image snipped*
These graphs only serve to prove the point that NVIDIA is stingy with VRAM. In that graph, Alan Wake 2 is already using more than 16 GB with path tracing and frame gen (which is the only way you're going to get 60+ FPS with PT at 4K, even with a 4090). And it's exceeding 12 GB at 4K, even with no PT. This is a current-gen game, and they will only get more demanding over time. If the 5070 performs anywhere close to a 4080, it's going to need more than 12 GB or it can't be considered present-proof, much less future-proof. Same goes for the 5070 Ti and 5080 and their 16 GB if they're at all close to the 4090.

Also, a lot of people will buy these cards to run high-end VR, which means a rendering resolution at 6K or even 8K, most VR games without good DLSS support (although this is improving), and counter-productive frame gen (it increases frametime, which is what matters for VR). You're definitely going to need more than 16 GB to push the latest Unreal Engine games using UEVR, the latest flight simulators, etc. My 6800 XT can only run Mechwarrior 5 in UEVR at ~5k rendering res, medium settings, and min 37 FPS (to get 72 FPS with spacewarp). I could probably get to the Quest 3's full ~6k res with high settings with a 4090 (but still only 37 FPS), but it would take a 5090 to get to 60 -> 120 FPS at high settings, and I still wouldn't be able to max out a higher-end headset like the latest Pimax or Varjo. I don't know what the VRAM usage is at those settings, but I bet it'll take almost everything you can throw at it. I personally wouldn't buy a mid-range card with less than 16 GB or a 70 Ti / 80 series card with less than 20.
 
No.

View attachment 693828

At 4K the 4080 is limited by its GPU, not VRAM.

The issue is 12GB. 2560x1440 is already hitting its limit in some games with ray tracing, as shown here.

But even with 16GB, if you move up to 4K ray tracing without frame generation is essentially hitting a wall. Turn on FG and what would likely have been playable frame rates may hit some VRAM shortage.

You can forget about path tracing, the hardware isn't there yet. But in the other scenarios the VRAM limitations can be problematic.
 
The price of these things after tax and shipping are going to be crazy. All cards like the guy a few posts above said.
 
Yep, this. Plus you'd be using dlss quality to improve framerates anyway, which means even less vram usage. People way overblow how much is really needed for their cards nowadays at a given performance tier.
Isn't that a bit of game by game if dlss save vram at all ?

vram.png


Or a general rules that you save vram with dlss on as started to be a trend ?

https://www.techspot.com/review/2856-how-much-vram-pc-gaming/

here lot of example with dlss increase vram, but they do it with frame gen on which maybe increase the issue
 
Isn't that a bit of game by game if dlss save vram at all ?

View attachment 693868

Or a general rules that you save vram with dlss on as started to be a trend ?

https://www.techspot.com/review/2856-how-much-vram-pc-gaming/

here lot of example with dlss increase vram, but they do it with frame gen on which maybe increase the issue
Frame generation uses vram. As for dlss, that's a very old one and it does indeed reduce vram consumption as it is lowering internal rendering resolution.
 
Last edited:
Allocated VRAM and needed VRAM is not the same thing. A game might only need 11GB but allocate a large portion of the rest as cache that could be loaded later from disk without any issues. The signs of running out of VRAM is typically lots of stuttering and 1% lows that are abnormally bad compared to the average FPS when compared to cards with similar average FPS.

You want the game to use as much VRAM as possible as it is generally efficient use of resources. That doesn't mean it is impossible to use similar settings on a card with less resources. It is very possible for a game that uses 17GB on a 4090 to run just fine if it only has access to 12GB or 16GB, it mostly depends on how the memory usage is implemented.
 
Back
Top