NVIDIA GeForce RTX 4090 Founders Edition

[ARF]

@W1zzard Would you make this page active? NVIDIA GeForce RTX 4090 Specs | TechPowerUp GPU Database

 

[R0H1T]

Wow look at that insane VFM

 

[W1zzard]

the Ray Tracing results show how much further 4090 can stretch its leg

You mean relative to Ampere? That's because of the improved RT units and the additional power headroom

 

[vlad.coolish]

For next tests of 4090 please make an experiment with power limit -100W, -150W and -200W to current max in 460W.
And check power efficient Performance by Watt after.

 

[clopezi]

Would be interesting see the performance on PCIe 3.0...

 

[amit_talkin]

Intimidating !

 

[Vayra86]

I start to understand why Nvidia only releases two SKUs. The rest is just not worth it versus Ampere.
Basically, Ada is what Ampere shouldve been.

Perf/watt is the only good thing I see here. But they had catching up to do anyway.

 

[THU31]

Less than 15% improvements vs 3090ti on 1080p and if you don't care about that useless RT bs. Amazing! So much power draw and wasted transistors on RT and tensor cores... Btw the article would attract more ppl if it included CSGO and Overwatch 2.

Amazing to see this GPU with 78 BILLION transistors has less FPS vs a 3080 (less than 28b transistors) on some games that do not use RT and on 1080p all the while using up to 600 WATTS!!! This has to be a new record of stupidity.

You are blaming the card for bad 1080p performance? Do even know anything about how video games work?

 

[outpt]

These will sale like hotcakes.
Will you do a RT only review or does that go beyond the call of duty?

 

[defaultluser]

hey man, proof that doom eternal IS SUCH AN EASY GAME TO RENDER, ITS ALMOST GOT THEW SAME PERFORMANCE AS A 3300X AT 1080P :

BUT NOW WE CAN SEE JUST HOW MUCH CPU THE GPU TEST PLATFORMS S HAVE LEFTl

 

[mrpaco]

What CPU do you have in your computer?

does it matter? I'm no journalist trying to make a review and showcase the true power of a product, that question is just a fallacy

if you are limiting artificially the limits of the product how people can know what the true unlocked performance is?

do you make CPU reviews using a 10€ cpu cooler that would throttle the device at 50%? Then why would do it with this?

Don't get me wrong, I am happy to see this review vs others with the latest cpu because I can now see something what under other circumstances we wouldn't see. But at the same time, it was not the best choice to display the 4090. Adding the top tier CPU changes it so much, that maybe should have been a second CPU test like the ryzen 7000 ones with air cooling and AIO's at the same time, but if you are going to do only one, then I think this was the wrong choice, just my 2c

 

[EatingDirt]

Damn Ray Tracing performance has taken quite a leap up, impressive indeed!

And it's actually rather more efficient too despite what was expected.

Raytracing is only a ~5% improvement over the 3090 Ti in terms of frame cost(Original FPS vs Raytracing on FPS). It's just a much faster card overall, so it's just much faster with raytracing enabled, but it's not much more efficient at raytracing.

I'd say the very small raytracing improvement is the largest surprise for me.

 

[arandomguy]

@W1zzard

The clock states table on page 40 "Efficiency and Clock Speeds" seems be the numbers from the Intel Arc 770.

 

[Deleted member 177333]

Nice writeup, thanks. I can easily afford the 4090 or even several of them, but I simply will not pay $1600 for a video card. Not gonna happen. Especially for one that doesn't even have the correct cooling solution installed. A premium card like that should have a waterblock. Period.

So, cut that price in half and offer it with a FC waterblock and I'd buy one immediately. Otherwise I'm happy to wait for 2nd-hand market in a couple years.

 

[R0H1T]

does it matter? I'm no journalist trying to make a review and showcase the true power of a product, that question is just a fallacy

if you are limiting artificially the limits of the product how people can know what the true unlocked performance is?

do you make CPU reviews using a 10€ cpu cooler that would throttle the device at 50%? Then why would do it with this?

Don't get me wrong, I am happy to see this review vs others with the latest cpu because I can now see something what under other circumstances we wouldn't see. But at the same time, it was not the best choice to display the 4090. Adding the top tier CPU changes it so much, that maybe should have been a second CPU test like the ryzen 7000 ones with air cooling and AIO's at the same time, but if you are going to do only one, then I think this was the wrong choice, just my 2c

Artificially how? Are you going to subsidize every household to get a (free) upgrade to the 13900ks or 7950x3d & play it with 4090/Ti

Most users out there will have an inferior platform than the one used here, in fact more than 90% or so!

 

[grammar_phreak]

Dear Lord

View attachment 265026

Someone should make a GPU cooler that resembles a retro 80's Boom Box.

 

[AusWolf]

The table on the "clock speeds" page shows the Intel Arc cards' clock speeds.

Nice review by the way. It's a shame the card is pretty underwhelming considering its price. Only slightly faster than the 6950 XT for double the price. What a massive slap in the face for the 5 green fans who will buy it!

 

[W1zzard]

@W1zzard Would you make this page active? NVIDIA GeForce RTX 4090 Specs | TechPowerUp GPU Database

Done

Would be interesting see the performance on PCIe 3.0...

Soon

hey man, proof that doom eternal IS SUCH AN EASY GAME TO RENDER, ITS ALMOST GOT THEW SAME PERFORMANCE AS A 3300X AT 1080P :

Different test system, different game version, different graphics drivers, slightly different test scene

do you make CPU reviews using a 10€ cpu cooler that would throttle the device at 50%?

I tested that, throttling is actually minimal: https://www.techpowerup.com/review/amd-ryzen-9-7950x-cooling-requirements-thermal-throttling/
Let me retest on 13900K and we'll talk again soon

The clock states table on page 40 "Efficiency and Clock Speeds" seems be the numbers from the Intel Arc 770.

I knew that looked wrong, fixing

 

[AdmiralThrawn]

Not impressed with the price chart TPU is using here. You cannot use current amazon prices to compare against the launch price of the 4090. The RTX 3080 is not 660$. Please fix this.

I am not defending the 1,600 dollar price but it is incredibly dishonest to say the 3090 is 900 bucks and not state that you are using current ebay/amazon prices instead of launch pricing.

 

[Legacy-ZA]

For that price and power draw, this "thing" is an abomination. I expected more performance to make up for it, much more.

 

[RandallFlagg]

I think amd will not have eny chances

Actually for the first time in a long time, I'm interested in what AMD provides.

The real game is in the mid and upper midrange. Call it below the 6800. Stuff that costs <=$500.

It's entirely possible to win in the midrange, and lose at the halo product or perhaps not even try to make a monster like the 4090.

Beyond just cost and power, that thing will not even physically fit in a lot of ATX cases.

 

[grammar_phreak]

does it matter? I'm no journalist trying to make a review and showcase the true power of a product, that question is just a fallacy

if you are limiting artificially the limits of the product how people can know what the true unlocked performance is?

do you make CPU reviews using a 10€ cpu cooler that would throttle the device at 50%? Then why would do it with this?

Don't get me wrong, I am happy to see this review vs others with the latest cpu because I can now see something what under other circumstances we wouldn't see. But at the same time, it was not the best choice to display the 4090. Adding the top tier CPU changes it so much, that maybe should have been a second CPU test like the ryzen 7000 ones with air cooling and AIO's at the same time, but if you are going to do only one, then I think this was the wrong choice, just my 2c

If you know how to read benchmarks, then you can see where there is a CPU bottleneck and where there isn't..... or where there is a limitation with the Game engine. Testing with a 7950x isn't going to magically make a 3090 look any better or worse.
Since all the cards in the chart are tested with the same setup... the results are not going to be a whole lot different compared to Jayztwocents.... especially at 4k where it really matters.

 

[Bwaze]

Not impressed with the price chart TPU is using here. You cannot use current amazon prices to compare against the launch price of the 4090. The RTX 3080 is not 660$. Please fix this.

I am not defending the 1,600 dollar price but it is incredibly dishonest to say the 3090 is 900 bucks and not state that you are using current ebay/amazon prices instead of launch pricing.

It says just that - and that is actually relevant to the buyers that are deciding right now what to buy. What should it be compared to? Release MSRP? Increased MSRP mid-generation? Highest achieved prices during cryptoinsanity (then the RTX 4090 would actually look affordable, cheaper than RTX 3080, about the max price RTX 3070 achieved).

"We looked up each card's current USD price on Newegg and used that and the relative performance numbers to calculate a performance-per-dollar index."

 

[Dimitriman]

After reading the 4090 reviews, I feel like this now:

 

[clopezi]

I really hope that AMD make battle, because it's good to all, but without DLSS (FSR actual implementation it's like DLSS 5 years ago) and without RT good performance, they cannot make a fight in the same conditions...