NVIDIA GeForce RTX 4090 Founders Edition

[RandallFlagg]

The only bad information here is that you are implying that $350 is the price the card was broadly available for. $350 was the founder's edition price which almost no one saw. Aside from the scalping, AIB pricing started at $400+ as was usual when comparing founders edition to non-founders edition.

Mind you that's comparing the 6GB 2060 to the 6GB 1060. Zero increase in VRAM. The 12GB 2060 cost a whopping $600 at time of review on TechPowerUp.

You paid $309 for a 2060 KO, which launched March 3rd 2020, a whopping 14 MONTHS after the Jan 7th 2019 launch of the 2060. Suffice to say, it's extremely misleading to compare late generation prices to MSRP.

It is you who is trying to interject "market price" from 3-4 years ago. I am using MSRP. I'll keep on using MRSP. The reasoning for using that to compare has been explained ad nauseum not only here, but on countless other sites. If demand is too high, then the AIBs and retailers pocket the difference. If it's too low, then they take a hit. MSRP is the only thing we have, anything else is chaos and cherry picking.

My comment about paying $309 was just pointing out that I got a discount on the card by buying late in the cycle, since you seemed dead set on comparing discounted prices to new release prices, or high demand scalp prices to MSRP. Anyone can do that, just cherry pick launch vs mid cycle vs late cycle, crypto bust vs crypto mania. That's chaos.

As I said, I think MSRP is the only reasonable way to compare pricing. And +$50 on a $299 card is only +16.7%. For that you get +45% at 1080P and +67% at 4K. That's not hard to understand.

You admitted you own a 2060 and I believe your ownership is clouding your judgement in this case. Even when you take the best turning SKU value wise in the best possible light it's at best underwhelming.

That's a laugh. I posted charts, I've been posting charts, I don't usually say stuff without proof when it comes to numbers because proof is so easy to find. It's 45% faster than 1060 at 1080P and 67% faster at 4K. I paid about $30 over what the 1660 Ti was selling for at the time (about 12% more) to get +20% over that card. The reason was likely because everyone said that the 2060 wasn't worth it, which drove up 1660 Ti demand, while the 2060 wasn't selling. So I got to take advantage of both bad information in circulation and its results. I don't do these things because fanboy gibberish like you imply, I do it because the math works out, and I already showed that it did.

 

[Dr. Dro]

this review is worthless. why would you test a high end GPU on a budget CPU? you are knee capping the performance. a lot of the benches here, even those at 4K, are CPU limited. you guys couldn't spring for a 5800X3D? or an intel bench?

no one who is the market for this GPU is going to be using a 5800X system.

everyone else used ryzen 7000 or high end intel. but you guys published this junk

Since when is a 5800X a budget CPU? I must have missed something. Its performance comfortably exceeds that of the CPUs installed in most gaming PCs today.

This guy out here literally with the mindset of "Imagine being so poor you could only afford a 5950X and a 3090 " , damn, I've never loved being poor as much as I do today

in short wtaf I don't like ai generated frames. or its equivalents

You're not alone, brother. I have to confess, I have never liked upscalers and I definitely do not like this frame generation thing. I will be skipping the launch models of this gen, probably going to grab something during the inevitable mid-gen refresh.

Other than that, it's a remarkable GPU. I'll be keeping a close eye on AMD's counteroffer, it looks like they have their work cut out for them.

 

[gffermari]

4090 shouldn't be tested in 1080p/1440p at the moment. Even in 4K, the card was kept behind by the cpu, even with 5800X3D(Hardware Unboxed).
The cpus clearly are ...30 years behind in technology compared to the gpus.

 

[Gica]

For haters: insert a link with a review in which the general impression was disappointment. I found it only with WoW!!!!

 

[Colddecked]

Since when is a 5800X a budget CPU? I must have missed something. Its performance comfortably exceeds that of the CPUs installed in most gaming PCs today.

This guy out here literally with the mindset of "Imagine being so poor you could only afford a 5950X and a 3090 " , damn, I've never loved being poor as much as I do today



You're not alone, brother. I have to confess, I have never liked upscalers and I definitely do not like this frame generation thing. I will be skipping the launch models of this gen, probably going to grab something during the inevitable mid-gen refresh.

Other than that, it's a remarkable GPU. I'll be keeping a close eye on AMD's counteroffer, it looks like they have their work cut out for them.

Since 12900 and 5800x3d came out, it's mid range now at best. Heck even in the 5000 series stack its in the middle.

 

[bobmeix]

Small typo on page 35: you can enable DLAA + FG

 

[Dr. Dro]

Since 12900 and 5800x3d came out, it's mid range now at best. Heck even in the 5000 series stack its in the middle.

I dunno man but I want some of what you guys are on calling it low-mid, but I want some, it must be the good stuff. You guys referring to it as if it were some sort of antique lmao

Vermeer remains one of the most competent gaming processors money can buy, as well as one of the most efficient processor architectures out there, and the 5800X remains a very high end processor for this purpose. I would go a step further on that, the 5700X is the CPU that most people looking to play video games, including those looking to build with an Ada GPU.

That particular processor is reliable, runs cool and has amazing performance, too. I'll just leave @Mussels' post on the Zen Garden do the talking:

Ryzen Owners Zen Garden

I had to take my B-Die to 1.55V to be stable at 3800 CL14 1T. Here are the settings I used. ClkDrvStr 40 and you may need to add AddrCmdSetup 56 if you get a bluescreen going into windows. RDWR can go to 7. It boots but when tested with memtest5 1usmus config give me errors on 6 and 12...

www.techpowerup.com

I don't think you will be losing out on a quality experience by using these with a Ryzen 5000 processor, or even one of the faster 3000 processors like a 3800XT or a 3950X.

 

[Why_Me]

View attachment 264998

+63% vs RTX3090 @ 3x 2nd hand market price in Europe (2K vs 700 Eur)
+88% vs RTX3080 @ 4x 2nd hand market price in Europe (2K vs 500 Eur)

Absolutely idiotic pricing in Europe. No thanks Ngreedia

Just about everything in the EU is priced insane. Do you ever think it might be the EU and not Nvidia?

 

[Wasteland]

As I said, I think MSRP is the only reasonable way to compare pricing. And +$50 on a $299 card is only +16.7%. For that you get +45% at 1080P and +67% at 4K. That's not hard to understand.

The MSRP of the 1060 (6 GB) was $250. The MSRP for the Founder's Edition was $300. This was back when NVIDIA charged a premium for the FE. I know because I bought a Zotac 1060 for $250 at launch.

I don't have a dog in this hunt, particularly, but there was a minor outcry about bad perf/price on Turing when it launched. I remember being deeply underwhelmed, myself. Later on the 16 series appeased people who felt that RT features weren't worth the premium, which I think was a sensible position at the time.

EDIT: Better link - https://www.gamersnexus.net/news-pc/2505-official-nvidia-gtx-1060-specs-price-release-date

 

[Dr. Dro]

Just about everything in the EU is priced insane. Do you ever think it might be the EU and not Nvidia?

It's NV alright. Prices are historically high even in America which gets dirt cheap hardware anyway. That said it is not out of line compared to the 3090's launch price, and honestly, anyone with a GA102-based GPU should not be running around with their hair on fire, it's just FOMO, really.

If you spent $800-$1600+ on a high end GPU, give it at least 3 years before you replace it. I guarantee anyone on a 3080+ is going to be comfortable gaming on ultra high settings for the next year if not more.

 

[Dimestore]

No latency is absolutely not reduced by DLSS's AI frame insertion. Given that it requires you to wait for the next frame to be rendered before creating the AI generated frame it will always carry a latency penalty until they remove that requirement.

This is why Nvidia require that Reflex be enabled with the frame insertion, to hide a portion of the latency penalty. If you disable frame insertion and enable reflex you will get lower latency.



I assume there's a sweet spot between 60 FPS and 144 FPS where it makes sense for some people to enable it. Above 144 FPS it doesn't make much sense to enable as smoothness is already very good and the latency penalty would outweigh any diminishing benefits above that frame-rate. Below 60 FPS and the latency penalty would be quite sizable, maybe for strategy games that aren't real time but even then it could still be annoying and those are the games least likely to benefit from additional motion smoothness.

Hopefully improvements come to the tech but until then I feel like it's going to be highly preferential as to whether it's worth enabling.

I don't see how it is possible to generate an intermediary frame between two frames using temporal (time-based) data and not require that the second frame be rendered before the interpose. Frame1 - Interframe - Frame2, in order to create interframe you must have the already rendered Frame1 and Frame2, and by the time you push Frame1 to the screen it is behind by a frame, giving the 1/3 extra latency. Is this incorrect?

 

[gffermari]

I dunno man but I want some of what you guys are on calling it low-mid, but I want some, it must be the good stuff. You guys referring to it as if it were some sort of antique lmao

Vermeer remains one of the most competent gaming processors money can buy, as well as one of the most efficient processor architectures out there, and the 5800X remains a very high end processor for this purpose. I would go a step further on that, the 5700X is the CPU that most people looking to play video games, including those looking to build with an Ada GPU.

We don't care if the 5000 are still capable or not or what most people have.
In reviews we care about real numbers.

It's the same when we test cpus and put a 4090 playing at 720p.
We don't care if it never happens. We only care about numbers and differences.

Yes a 5800X was a bad choice for GPU review but it is understandable. Every reviewer has to retest every gpu using the latest and greatest cpu when there is no always time for that.
Wizzard will do that, obviously, when 13900K arrives. And this time it would be great if we had gpu usage at some point, so we know if there is or not a cpu bottleneck.

 

[adilazimdegilx]

Um.... from the portion of the review conclusion that I've been criticizing (because it makes no sense from the data)?

I'll quote it again, for reference. Emphasis added.



I'm not critiquing the card. I'm critiquing a section of this review's conclusion that makes ZERO sense. The data in this review does not support these statements with respect to "classic" raster performance. The data in this review suggests that this card is right in the middle of the pack with regards to new GPU architectures launched over the past 10 years of GPU launches and is just fine.

This card's performance is excellent - as expected - without needing to use unwarranted hyperbole and unearned effusive praise in comparison to what previous architectures achieved at their respective launches. It sounds much more like marketing than product review.

I think w1z is correct with this one. But I agree it's not obvious as it's not explained clearly.
4090 does indeed make the highest jump between new generations since (at least) 6xx series. (I didn't check older generations but it probably goes until 8xxx series, TPU reviews only).

This 'gap' is between the fastest new generation card 'at launch' versus the fastest old generation card at that time.

So it's between 4090 and 3090ti and it's 45% at 4K.
It was close to this (43%) between 3090 (launched 1 week after 3080 but still counts I guess) and 2080ti.
39% between 2080ti and 1080ti.
Only 23% between 1080 and Titan X (1080ti released 1 year later).

 

[P4-630]

it's finally time to test the most powerful cards at 8k.

Waiting also 5K, 6K, 8K test. And PCI-E 3.0 vs PCI-E 4.0?

Here you go, 8K gaming tested in various games:

Nvidia GeForce RTX 4090 - Olifant in de porseleinkast

De Nvidia GeForce RTX 4090 Founders Edition is een gigantische videokaart die een flinke prestatietoename moet leveren. Is de kaart net zo snel als hij groot is?

tweakers.net

 

[metalslaw]

Doesn't HDMI 2.1 already max out at like 4k 240hz and 8k 144hz? I mean who would ever need more than that mate?

If I'm going to run triple screen 4k for a sim rig, I'm limited to a max of 97 hertz with 1.4a at 4:4:4.

With that number not being 120/144 hertz, this would also nullify freesync/g-sync use at 4k, as most screens only offer the magical 2.4x range required, at min to max hertz range of 50 hertz to 120 hertz, or 60 to 144.

This of course wouldn't be a problem if the gfx cards still had a 3 DP, 3 HDMI, arrangement (as I could then use 3x HDMI instead). But every 3090 is currently maxed at 2 hdmi ports (and 3 1.4a DP ports), so I'm stuck with 1.4a for triple screen, with no ability to run triple 4k 4:4:4 with g-sync/freesync on in that triple setup.

 

[Colddecked]

I dunno man but I want some of what you guys are on calling it low-mid, but I want some, it must be the good stuff. You guys referring to it as if it were some sort of antique lmao

Vermeer remains one of the most competent gaming processors money can buy, as well as one of the most efficient processor architectures out there, and the 5800X remains a very high end processor for this purpose. I would go a step further on that, the 5700X is the CPU that most people looking to play video games, including those looking to build with an Ada GPU.

That particular processor is reliable, runs cool and has amazing performance, too. I'll just leave @Mussels' post on the Zen Garden do the talking:

Ryzen Owners Zen Garden

I had to take my B-Die to 1.55V to be stable at 3800 CL14 1T. Here are the settings I used. ClkDrvStr 40 and you may need to add AddrCmdSetup 56 if you get a bluescreen going into windows. RDWR can go to 7. It boots but when tested with memtest5 1usmus config give me errors on 6 and 12...

www.techpowerup.com

I don't think you will be losing out on a quality experience by using these with a Ryzen 5000 processor, or even one of the faster 3000 processors like a 3800XT or a 3950X.

Just because its a mid level CPU doesn't mean it sucks. Its now a very good value. BUT it isn't the pinnacle of PC gaming, that's all.

 

[TheoneandonlyMrK]

If I'm going to run triple screen 4k for a sim rig, I'm limited to a max of 97 hertz with 1.4a at 4:4:4.

With that number not being 120/144 hertz, this would also nullify freesync/g-sync use at 4k, as most screens only offer the magical 2.4x range required, at min to max hertz range of 50 hertz to 120 hertz, or 60 to 144.

This of course wouldn't be a problem if the gfx cards still had a 3 DP, 3 HDMI, arrangement (as I could then use 3x HDMI instead). But every 3090 is currently maxed at 2 hdmi ports, so I'm stuck with 1.4a for triple screen, with virtually no ability to run triple 4k 4:4:4 with g-sync/freesync on in that triple setup.

I think the video output dire straights on these cards is on purpose, a lifetime limiter so to speak.

 

[antuk15]

So we get ~70% more performance on average with ~60% more shaders + 50% more clock speed.

Architecturally it's not scaling very well compared to previous generations.

 

[evernessince]

It is you who is trying to interject "market price" from 3-4 years ago. I am using MSRP. I'll keep on using MRSP. The reasoning for using that to compare has been explained ad nauseum not only here, but on countless other sites. If demand is too high, then the AIBs and retailers pocket the difference. If it's too low, then they take a hit. MSRP is the only thing we have, anything else is chaos and cherry picking.

My comment about paying $309 was just pointing out that I got a discount on the card by buying late in the cycle, since you seemed dead set on comparing discounted prices to new release prices, or high demand scalp prices to MSRP. Anyone can do that, just cherry pick launch vs mid cycle vs late cycle, crypto bust vs crypto mania. That's chaos.

As I said, I think MSRP is the only reasonable way to compare pricing. And +$50 on a $299 card is only +16.7%. For that you get +45% at 1080P and +67% at 4K. That's not hard to understand.



That's a laugh. I posted charts, I've been posting charts, I don't usually say stuff without proof when it comes to numbers because proof is so easy to find. It's 45% faster than 1060 at 1080P and 67% faster at 4K. I paid about $30 over what the 1660 Ti was selling for at the time (about 12% more) to get +20% over that card. The reason was likely because everyone said that the 2060 wasn't worth it, which drove up 1660 Ti demand, while the 2060 wasn't selling. So I got to take advantage of both bad information in circulation and its results. I don't do these things because fanboy gibberish like you imply, I do it because the math works out, and I already showed that it did.

Are you contesting the fact that AIB models were more expensive than founders edition cards? That's an argument certain to fail. That fact alone disproves the idea that using only MSRP while ignoring street price is misguided, let alone the 3000 serier's MSRP vs what people actually paid. It's one thing to use MSRP when people could actually purchase at said price like the 10xx series. That's completely fine. It's not fine to only use MSRP when the vast majority of people are not getting MSRP prices, whether that be from scalpers of Nvidia founders edition shenanigans. The goal of any price comparison is to compare prices people actually paying, it makes no sense to ignore other factors that will impact that.

You are using MSRP? Your $309 2060 KO comment says otherwise.

No, you seem to be using whatever numbers happen to support your point when you are clearly biased by your 2060 ownership.

MSRP has been plain misleading in many cases the last two generations. The 2000 was a terrible generation overall. If you can't agree on those two points you are just denying reality and we have nothing further to discuss. No amount of goal post moving will change that.

 

[erek]

The biggest issue of this generation is that other cards despite costing an arm and a leg are castrated too much, not to mention that NVIDIA had no qualms calling RTX 4070 an RTX 4080.

It even looks like there will be no RTX 4060/4050 this gen and Nvidia will continue selling the previous generation stock.

Won’t they rebrand the 3050 through 3060 to new generation parts?

 

[evernessince]

I don't see how it is possible to generate an intermediary frame between two frames using temporal (time-based) data and not require that the second frame be rendered before the interpose. Frame1 - Interframe - Frame2, in order to create interframe you must have the already rendered Frame1 and Frame2, and by the time you push Frame1 to the screen it is behind by a frame, giving the 1/3 extra latency. Is this incorrect?

DLSS 3.0 frame insertion does indeed wait for frame 2 in your example before generating the intermediary frame.

 

[cvaldes]

Won’t they rebrand the 3050 through 3060 to new generation parts?

One would presume so although no one here has a timetable. Clearly NVIDIA is trying to draw down inventory of assembled graphics cards in the channel as well as Ampere GPU chips.

Unfortunately for NVIDIA crypto mining demand has completely evaporated due to the crypto market crash and Ethereum's PoS merge so they don't have the option of selling off excess Ampere GPU chips in mining cards.

At some point, they will sell out of a particular Ampere GPU and then it'll be time for NVIDIA to decide whether it's worth transitioning the low end products to the Ada Lovelace generation. Or they can tell Samsung to fire up their foundry and churn out another batch of those Amperes.

From a marketing standpoint, I'm sure they would rather have their entire product line on the latest generation rather than straddle both Lovelace and Ampere.

But remember that two years ago when Ampere launched, they marketed the 3090 and 3080 while still selling 2060s. The 3060 and 3050 didn't come until later.

 

[QUANTUMPHYSICS]

As we can see from the MSI SUPRIM LIQUID X, (and even the Kingpin 3090Ti)

there's No good reason the cards have to be that size.

 

[N/A]

So we get ~70% more performance on average with ~60% more shaders + 50% more clock speed.

Architecturally it's not scaling very well compared to previous generations.

4090 is double the performance in games that are not CPU limited like resident evil in 4K gen over gen which means 3090 non Ti. And with frame insertion probably goes up to 3 times.
How is it supposed to scale when it uses the same memory type, it needs GDDR7 very badly. But no such luck. This behemoth is doomed to a kind of failure like the Fermi. GDDR6X is broken, it does very little, 3070 Ti is the same as 3070, no actual benefit from transferring data on 4 voltage levels. just a ginormous power loss.

 

[cvaldes]

As we can see from the MSI SUPRIM LIQUID X, (and even the Kingpin 3090Ti) View attachment 265077there's No good reason the cards have to be that size.

Water has a higher cooling capacity than air.

However you're also moving the heat from one place (the graphics card PCB) to another (the radiator) which is placed in a location that is better for heat dissipation.

You can't just add up cubic centimeters of cooling solutions and say "the stock cooler doesn't need to be that thick."