There's still a lot to come before the release but what can you expect from the two
WePC is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Prices subject to change. Learn more
Now the RTX 4000 series is in full swing, we look at what the RTX 4090 vs 3090 has to offer. The initial flagships of their generations, they are powerful and brought great innovations and performance with them.
They still have a lot to offer even if they do get overtaken. And so the RTX 4090 came out on the 12th of October 2022. Bringing forth the new Ada generation and the improvements that it brought with it.
Whilst the 3090 was announced and released in September 2020. Although these came with a stock shortage and scalping. This led to very high pricing and so it wasn’t for a while that you could get any cards.
So with that, we look at how they compare with each other.
Check availability with our RTX 4090 stock tracker and where to buy
Excellent air cooling solution
Pushes performance above the rest of the selection
Updated design cooling and aesthetically focused
A more expensive choice over the rest of the picks and will cost a lot to run
Powerful cooling system keeps temps low
High clock frequency boosts performance above base level
Incredible build quality and design makes it a top pick from the crowd
On the higher end of the 3090 price spectrum
A high price to pay for the card with the current generation
Shop on Amazon
CHECK PRICEThe two cards are from two different generations. With an architectural and foundry change again for Nvidia. Changing the core processing at the heart of the card.
The 3090 is based on the Ampere architecture, with the GA102-300 GPU variant. Created with Samsungs 8nm process it creates a die of 628mm² in size with 28.3 billion transistors.
Whereas the RTX 4090 graphics card is on the Ada/Lovelace architecture with the AD102-300 chip. Which itself has been moved back to TSMC and its 5nm process. With a die size of 608mm² and 76.3 billion transistors
Although the newer chip may be smaller, it does have a more compact build. With its smaller process, it can fit so much more in. Leading to its improved build and performance.
| RTX 4090 | RTX 3090 | |
|---|---|---|
| GPU | AD102-300 | GA102-30 |
| GPU process | TSMC 5nm | Samsung 8nm |
| CUDA cores | 16,384 | 10,496 |
| Base clock | 2,235 MHz | 1,395 MHz |
| Boost clock | 2,520 MHz | 1,695 MHz |
| Memory | 24GB GDDR6X | 24GB GDDR6X |
| Memory clock | 21 Gbps | 19.5 Gbps |
| Memory bus | 384-bit | 384-bit |
| TDP | 450W | 350W |
These are the core specs that we can use to compare the key features. As well as informing you why they are important.
CUDA stands for Compute Unified Device Architecture. These are parallel processors on Nvidia graphics cards working on creating the final image.
They are responsible for all the processing aimed at the graphics card. Along with count number, there can be changes in efficiency improving graphics performance.
The older Ampere card features 10,496 cores. Whilst the Ada card has 16,384. A significant increase likely leads to improved performance.
Also known as the clock rate of the processor, it’s the speed of processing. With how quickly the graphics and instructions get done through the card.
It is how quickly the silicon crystals flip that determines the stat. The spec can be played with and is often subject to overclocking. But does lead to increased temperatures and power providing those extra frames.
The new RTX cards have a vastly improved frequency. Comparing to the Radeon cards already on the TSMC.
And we see the FE RTX 4090 achieve a base clock of 2,235 MHz and boost up to 2,520 MHz. With many custom cards pushing that even further.
Whereas the 3090 is rather slow in comparison. With a base clock of 1,395 MHz and a boost clock of 1,695 MHz. A near 1 GHz improvement that is lightning quick.
Video memory is the RAM of the graphics card. It is utilized as a frame buffer and pixel store for improving framerates.
It’s most important for higher resolutions. Reaching 4k and beyond will need more pixels and therefore capacity and speed to keep it going.
Between the two the VRAM is kept the same. With 24GB of GDDR6X as well as using a 384-bit memory bus.
Where they differ is the memory clock. There the new gen card improves the speed by over 2 Gbps. Reaching 21 Gbps over 19.5, further increasing the quickness of the Ada cards.
This pushes the bandwidth to over 1GHz at 1,008 MHz, instead of the 936.2 MHz of the 3090.
The thermal design power of a card is the total energy consumed under full utilization. It is the maximum it will consume at any point.
This also determines how much heat the card gives off. Coming from the work the chip does in such a small area.
So comparing the two we see a large increase in power consumption. There is a likely 100W difference between the two GPUs.
With the 4090 the one reaching 450W. Closer to the consumption of the 3090 Ti than the 3090.
This has led to much larger cooling solutions like 4 slot cards. So it is important to consider where you’re putting it or consider water cooling your card. But also requires a more powerful PSU to run.
It will also feature the new Gen 5 connector like the 3090 Ti. Which will require using the adapter or a new gen PSU.
Below is the data from our ROG Strix OC RTX 4090 review where we compare it to the 3090. In which you can see how much the new gen improves over the old.
In particular, ray tracing and AI performance show generational improvements. Along with across-the-board increases in gaming and rendering.
There are significant improvements over the 30 series card across the board. Although some more CPU-bottlenecked games like CSGO may not show it as that isn’t as much of a difference, ranging from 4-14%.
Otherwise, there are performance increases of 20-90% in the other games. With DLSS 3 providing even bigger opportunities such as over double the performance in Flight Sim and Cyberpunk.
Now one of the big features of the new generation is the expansion in size. Whilst the 3090 was seen as chonk in its own right, the 4090 has improved upon that turning into a behemoth of a card.
So when it comes to the Founder’s Edition reference card, the 4090 measures 304 x 137 x 61mm. Compared to 313 x 138 x 61mm of the 3090. So it does cut down on the size with the cooling efficiency improvements.
However, where they differ is in the custom solutions. And for the likes of the ROG Strix the RTX 4090 comes in at 357.6 x 149.3 x 70.1mm. Whilst the RTX 3090 was only 319 x 140 x 58mm. Pushing the sizing up to four-slot territory.
The 3090 launched with an MSRP of $1,499USD/£1,399. Whilst the 4090 has increased that price by $100 and reached an MSRP of $1,599. This isn’t as much of a jump as seen on the 4080 and might be more justified if you need the power.
So for the performance improvements, it’s a fairly good deal. And we have seen the 4090 stock lacking slightly as people buy it up. Unlike the 4080 that seems to not grab as much interest.
Excellent cooling to give the card plenty of overhead
Minimizes the card size to just 2 slots instead of usual 3.5
A more premium design
Have to consider where to mount radiator
More moving parts that could fail
Competitively priced
2nd Gen Ray Tracing Cores
Extremely efficient cooling solution
One of the quieter 3090 AIBs
Not the most attractive RTX 3090
Shop on Amazon
CHECK PRICEOverall, we see that in fact, the 4090 is a lot better than the 3090. Although it does come in at an extra cost if you want to go for it. So f you can find good pricing it is definitely a good competition. Especially as even RDNA 3 is not expected to fight at that level.
Although as the older card, the 3090 can definitely be found for cheaper. So if the performance is enough for you and the price is right then there is no reason not to opt for the older card. As it itself is a great pick for its time.
The RTX 4090 has a 60% faster base clock and a 49% higher boost clock than the RTX 3090.
Yes, it is. The 3090 was too much for what the average user need when it came out. With so much VRAM and with a high cost it wasn’t on everyone’s radar. But with the new-gen coming and lower prices, it can achieve anything you want. With plenty of VRAM and cores, it can play anything even in 4k and ray tracing.
Yes, its MSRP is set to $100 more than the 3090. But the price difference will be much higher with the age difference leading to the price cuts on the 30 series cards.
The RTX 4090 is a huge card and has increased over the last generation. So looking at the Founder’s Edition we do in fact see the 4090 did get smaller. It measures 304mm in length, 137mm in width, and 61mm in height (triple-slot). Whilst the 3090 is 336mm in length, 140mm in width, and 61mm in height. Showing the improved airflow and cooling of the new card.
Where you do see bigger sizing is in the custom cards. So if you compare the ASUS ROG Strix models you do in fact see a big change. The 4090 Strix measuring in at 358mm in length, 149mm with, and 70 mm in height (quad-slot). Whilst the 4090 Strix comes in at 319mm in length, 140 mm in width, and 58 mm in height.
Trusted Source
WePC’s mission is to be the most trusted site in tech. Our editorial content is 100% independent and we put every product we review through a rigorous testing process before telling you exactly what we think. We won’t recommend anything we wouldn’t use ourselves. Read more
