NVIDIA GeForce RTX 4000 Series Release Date, Price, Specs

We don’t know much about the NVIDIA GeForce RTX 4000 series cards. Unfortunately, no one does at the time of writing this article.

However, there is some information we’ve managed to gather about this new generation of GPUs.

We also consider NVIDIA’s previous GPU lineup, the RTX 3000 series, and the performance improvements over the RTX 2000 series. This helps us visualize the potential of the RTX 4000 GPUs.

If you are interested in what the next generation of NVIDIA graphic cards will bring, this article contains the information you are looking for.

Let’s get right into it!

Related:AMD RDNA 3 Release Date, Price, And SpecsIntel Arc Alchemist Release Date, Price, Performance, Specs

Release Date

Currently, there is no official information regarding the release date of the following NVIDIA GPU lineup. So, what do we know so far?

Based on NVIDIA’s official architecture roadmap, Ampere Next (or Ada Lovelace) has a planned 2022 release. By 2024, they believe they will have Ampere Next Next ready, which is supposedly the RTX 5000 series.

However, there are also talks that Nvidia might skip the 4000 naming scheme altogether and jump to 5000. So, RTX 5000 could be Ada Lovelace which means Ampere Next Next will be RTX 6000.

Kopite7kimi’s latest rumors point towards early Q3 for the Ada Lovelace release date. This is a few months earlier than we originally expected.

Q3 early

— kopite7kimi (@kopite7kimi) May 15, 2022

An earlier release for Nvidia can be a huge win over AMD. Even if AMD comes out with an overall better GPU lineup, Nvidia will grab a large chunk of the marketshare.

NVIDIA’s Stopgap

It still feels early for the RTX 4000 series, considering the relatively recent 3070 Ti and 3080 Ti releases.

But wait, NVIDIA is planning more!

Circulating rumors hint towards RTX 3000 Super GPUs that will release at the end of 2021 or early 2022.

Let's make a summary.
3090S 10752 24GB G6X
3080S 8960 12GB G6X
3070S 5888 8G G6X
3060S 5632 12G G6
Although I doubt the specs of some of them and the name of 90S.
👀👀👀

— kopite7kimi (@kopite7kimi) September 22, 2021

The above leak by kopite7kimi shows the specs of the entire lineup. Highlights include increased CUDA cores, VRAM memory, and an upgrade to GDDR6X from GDDR6 non-X (excluding the 3060S).

There are some rumors that the RTX 3090 Ti’s production is at a halt because of a hardware problem. A hardware problem during production can cause a several-week or several-month delay. It could even completely suspend the release of the GPU.

Either way, these new Super or Ti variants of the RTX 3000 Series won’t be a significant upgrade from the current lineup. In fact, the upgrade will be minimal. So, if you are already on an RTX 3000 SKU, we recommend waiting for RTX 4000.

Specifications

As mentioned previously, no official info is available for the RTX 4000 series, but there have been some leaks. Currently, the only concrete information is its codename: Ada Lovelace.

The Ada Lovelace microarchitecture will be on a 4nm fabrication process, a giant leap from the 8nm Ampere (Samsung) chips. A smaller microarchitecture will result in two significant benefits: more performance for a lot less power.

Thanks to TSMC’s 4nm process, Nivida gets those benefits by cramming more transistors onto the chip.

Note that most of the first leaks we had on the RTX 4000 suggested that Nvidia will use TSMC’s 5nm process.

Leaks suggest an increase of more than 7500 CUDA cores from the GA102 (3090, 3080) chip. This includes almost double GPC (Graphics Processing Clusters), 7 to 12, and 30 more TPC (Texture Processing Clusters).

That’s a massive bump in specifications, considering the Ampere chip had an increase of just 1 GPC from the Turing 102 chip.

GA102, AD102, AD103, and AD104 (rumored) specification comparison table:

Specifications	RTX 3090	RTX 4090 Ti/Titan	RTX 4090	RTX 4080	RTX 4070
Architecture	Ampere	Ada Lovelace	Ada Lovelace	Ada Lovelace	Ada Lovelace
Process	Samsung 8nm	TSMC 5nm	TSMC 5nm	TSMC 5nm	TSMC 5nm
CUDA Cores	10496	18432	>18432	10752	7680
TFLOPs	35.6	66	>66	58	40
Memory	24GB GDDR6X	48GB GDDR6X	24GB GDDR6X	16GB GDDR6X	12GB GDDR6X
Memory Speed	24Gbps	24Gbps	21Gbps	21Gbps	21Gbps
L2 Cache	6MB	96MB	96MB	64MB	48MB
Total Graphics Power	350W	900W	450W?	350W	300W
Release Date	September 1st, 2020	TBA 2022	TBA 2022	TBA 2022	TBA 2022

We’re basing the specifications in the table above based on a leak from Twitter user 3DCenter_org.

So, nVidia's AD102 chip maybe is like:
12 GPC
72 TPC
144 SM
18'432 FP32 units
~66 TFlops FP32 power (on 1.8 GHz) https://t.co/A8OnUktE1s

— 3DCenter.org (@3DCenter_org) December 28, 2020

Interestingly, Kopite7kimi mentions that the RTX 4080 will use AD103 chips rather than the AD102 chip. That’s unexpected considering that the RTX 3080 shared the same chip (GA102) as the RTX 3090.

In February 2022, Nvidia went through a huge data breach which is unfortunate and we do not condone such attacks/hacks, however, this information is already out there, so here’s what we managed to find related to RTX 4000.

CUDA Cores and L2 Cache

Interestingly, 3D Center’s December leaks about the AD102 CUDA cores are actually correct. The leaks straight from Nvidia suggest 18,432 CUDA cores which is about 70% more than the RTX 3090. Already, those are some impressive numbers.

Additionally, Nvidia looks to increase L2 cache memory on all Ada Lovelace SKUs. Even the low-end GPUs (AD107) will get about 32MB of L2 cache. The RTX 4090 card will max out at 96MB of L2 cache.

The RTX 3060 has only 3MB of L2 memory which means that the future RTX 4060 will have 10 times bigger L2 cache. To make this change even crazier, this means that the 4060 will have more L2 than the RTX 3090.

Here’s a L2 cache table comparing RTX 3000 and RTX 4000 GPUs:

RTX 3000	L2 Cache	RTX 4000	L2 Cache
GA102	6MB	AD102	96MB
GA103	4MB	AD103	64MB
GA104	4MB	AD104	64MB
GA106	3MB	AD106	48MB
GA107	2MB	AD107	32MB

It looks like boosting cache size is a trend these days considering that both AMD and Intel increased their L2 and L3 cache sizes on their processors to boost gaming performance.

We’ll see whether Nvidia’s memory boost will improve the GPU’s ability to tackle games.

With these specifications in mind, it’s definitely possible that the RTX 4000 series graphics cards can be twice as fast as RTX 3000 cards. Maybe even faster.

Power Usage

There are multiple rumors that Ada Lovelace will have GPUs with TDP higher than 600W, but Moore’s Law Is Dead claims that these cards will not be able to draw more than 700W.

So, a realistic assumption is around 600W for an RTX 4090/4080.

These power usage-related assumptions are also supported by the fact that many power supply manufacturers are now switching over to 12-pin PCIe 5.0 power connectors with the capability to deliver 600W of power.

Someone told me the xx90 with 600W TGP has been confirmed.
I think it's too early to talk about it.

— kopite7kimi (@kopite7kimi) March 12, 2022

Previously, Kopite7kimi was pretty confident that the 4090 will be a 600W TGP GPU, but new leaks point to a lower TDP.

OK, let's do a new summary.
RTX 4090, AD102-300, 16128FP32, 21Gbps 24G GDDR6X, 450W, ~2×3090.
I am disappointed with RDNA3.
That's all.

— kopite7kimi (@kopite7kimi) May 16, 2022

If the RTX 4090 is truly 450W and outputs performance twice faster than the RTX 3090, this card can be truly competitive against Navi 31. We can only hope it will be priced accordingly.

While 450W is still pretty high power usage, it seems these changes are necessary. Even the switch to 4nm might not be enough to tackle AMD’s new MCM-based GPUs.

AMD’s Navi 31 SKUs will also have a higher power draw, but the limit may be around 400W which is still lower than Nvidia’s future flagship.

There are also rumors circulating about a new RTX Titan which could draw over 900W.

RTX 4080 will use AD103 chips, build with 16G GDDR6X, have a similar TGP to GA102.
RTX 4070 will use AD104 chips, build with 12G GDDR6, 300W.
Both of them haven't started testing yet, but soon do them.

— kopite7kimi (@kopite7kimi) April 27, 2022

Kopite7kimi shares that the RTX 4070 will have 300W TGP. While that’s still far away from the 4090 in terms of power draw, that’s still a lot of power for an xx70 card. The 3070 has a TDP of 220W which is miles better.

TFLOPS

Allegedly, AD102’s target frequency is 2.2GHz, which is theoretically around 66 TFLOPS (FP32). That is almost double in comparison to Ampere’s flagship (RTX 3090).

However, Greymon55 has another say about the specifications of Ada Lovelaces, which counters 3DCenter’s rumors. Back in 2021, Greymon55 already posted a leak with AD102 specifications and mentioned that the 3090 successor could have up to 92 TFLOPS.

If lovelace gets up to 2.5GHz, you'll get about 90t of FP32, and 3090ti is just 40t.😬

— Greymon55 (@greymon55) January 25, 2022

Now, the leaker doubles down on the idea and still believes that AD102 will have up to 90 TFLOPS compared to the 66 TFLOPS rumored by 3DCenter. That’s double compared to the 3090 or even the 3090Ti.

However, remember that TFLOPS isn’t the most accurate depiction of in-game performance.

Performance

There are currently no benchmarks open to the public, so we can’t really determine the performance of the RTX 4000 series, but we have a few rumors to share.

Greymon55 has again provided some insight. Based on their claims, these new GPUs should bring double the performance and double the power consumption.

Double performance, double power consumption, can you accept it?

— Greymon55 (@greymon55) November 5, 2021

Should this be acceptable? We welcome 2x performance, but doubling the power consumption can be a problem.

Only a small percentage of PC users have 1000W+ PSUs that can handle a 600W GPU.

However, it’s not all bad. Previously, most of us tech sites/reviewers assumed that Nvidia’s Ada Lovelace will be based on TSMC 5nm fabrication process.

But, it seems that we’ve all been wrong because RTX 4000 GPUs will actually use a 4nm node. Before this news, the favor was heavily shifted towards AMD’s RDNA 3, but the scales are finally tipping.

So, even with these huge power draw spikes, Ada Lovelace should be very competitive.

Price

It is impossible to guess how the pricing will look with the current info. However, since AMD’s RX 6000 series is on par with the RTX 3000 series in terms of performance (in some cases even better), NVIDIA will most likely stick with the exact pricing as Ampere.

So, the 4060 should be around $330 while the 4070 and 4080 for $500 and $700 respectively. We’re unsure about the 4090/Titan, though. Those two would cost more than $1000.

Unfortunately, there are some talks circulating that the jump to 4nm from TSMC could be a lot more expensive than we expected.

For example, the previously mentioned Twitter leak by kopite7kimi was a response to nerdtechgasm’s suggestion that 4nm NVIDIA is going to be very expensive for consumers.

Additionally, Nvidia is reportedly spending tens of billions of dollars to help the supply of RTX 4000 and future cards.

Scalping & Price Increases

If you have recently been searching for a new graphics card for your system, you have probably been disappointed several times. GPUs with a $500 MSRP are on retail stores like Amazon or Newegg for more than $1000.

The $700 RTX 3080 is frequently on sale by scalpers for $1500 on the second-hand market.

Such a level of scalping is possible due to the global shortage of semiconductor chips in 2021. The world’s largest semiconductor manufacturers, TSMC and Samsung, struggle to keep up with the demand.

Already this shortage continues in 2022 as it did in 2021 and 2020.

Right now, you might be wondering: will this kind of scalping affect the RTX 4000 series?

This is a great question and quite a concerning one, but when the semiconductor industry goes back to normal, we cannot know. Fortunately, TSMC has already begun investing to increase its manufacturing capacities. In addition, Samsung is also planning to build a new chip manufacturing factory in the US.

There are also rumors that the USA is planning to invest $37 billion in the chip-making industry.

We will keep you updated if there is any further information released about the future NVIDIA GPU lineup.