Home > GPU

RTX 5090 vs RTX 3090 specs comparison – how do they compare?

How does the RTX 5090 stack up against the RTX 3090?
Last Updated on
RTX 5090 vs RTX 3090 specs comparison – how do they compare?
PC Guide is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Read More

The RTX 5090 has finally been released. It is one of the four cards announced during CES 2025, and it looks quite impressive with its beefed-up specifications. It is set to replace the RTX 4090 as the most powerful GPU, but we’re wondering what the difference is between this card and the RTX 3090.

We’ve tested and reviewed the RTX 3090 in-house, and it is a solid card for 4K gaming, especially when you leverage DLSS 3.5. Now that the 5090 has launched, we’ll take it through its paces in our testing lab just like we did with the RTX 3090 and update this guide with real-world benchmarks once our review goes live. So, for now, we’ll compare these cards based on our review of the 3090 and the available performance reviews and videos for the RTX 5090.

🚀 Save Up to $1,200 on the Samsung Galaxy S25!

Pre-order now and save big with trade-in and Samsung credit. Limited time only!

*Includes trade-in value + $300 Samsung credit.

Here’s how the RTX 5090 and RTX 3090 compare against each other so far.

NVIDIA GeForce RTX 5090 Founders Edition

Shop on Amazon

CHECK PRICE

NVIDIA GeForce RTX 3090 Founder’s Edition

Shop on Amazon

CHECK PRICE

RTX 5090 vs RTX 3090 – specs comparison

SpecificationsRTX 5090RTX 3090
Cores2176010,496
Base clock speed2.01 GHz1.40 GHz
Boost clock speed2.41 GHz1.70 GHz
Memory 32 GB GDDR724 GB GDDR6X
Memory interface 512-bit384-bit
Memory clock28 Gbps19.5 Gbps
Bandwidth1.79 TB/s936.2 GB/s
TDP575W350W
Power connector12V-2×61x 12-pin
Release dateJan 2025Sep 2020
Dimensions304 mm x 137 mm x 48 mm313 mm x 138 mm x 55 mm

As you can see in the table above, since both are flagship models in their respective generations, they offer top-tier specifications. Of course, being the latest card, the RTX 5090 takes the lead in everything, including VRAM, clock speeds, bandwidth, and architecture.

Memory

The most important feature of a GPU is its VRAM. This is a type of memory that can store graphical data like textures, and the higher the VRAM of a GPU, the easier it will be for you to adjust and increase the graphical settings of your games, especially at higher resolutions like 4K.

The RTX 5090 offers 32GB of GDDR7, the highest amount of VRAM currently offered by a gaming GPU. The memory type is also better and faster than what was offered in previous generations. The RTX 3090, on the other hand, offers 24GB of GDDR6X VRAM, the same as its successor, the RTX 4090.

This means that there’s a difference of 8GB between the two cards, which is massive. So, when it comes to raw performance, the RTX 3090 won’t be able to keep up with the latest flagship GPU. However, we do think that the 32GB VRAM on the 5090 is a bit overkill. We already had the RTX 4090 with 24GB of VRAM, which is more than enough for gaming. Nvidia should have instead focused on improving the VRAM on their mid-range models, like the RTX 5070.

Cores

CUDA cores are the backbone of the GPU, as they handle multiple tasks simultaneously without affecting the system’s overall performance. What this means is that a higher number of cores means that GPU can process and execute data much more quickly.

In terms of core count, the RTX 5090 has 21,760 cores, while the RTX 3090 has 10,496. This means that there’s a difference of 11,264 cores between the two. While the core count on the 3090 is pretty decent and can help it handle GPU-intensive tasks smoothly, the RTX 5090 leads this sector by a big margin. On top of that, it has the latest 4th-generation Ray Tracing cores and 5th-generation Tensor cores, which means each core on the GPU is much more powerful and efficient compared to the previous generations.

Clock speeds

With a base clock speed of 2.01 GHz and a boost clock speed of 2.41 GHz, the RTX 5090 offers much better clock speeds compared to the 3090, which operates at a base frequency of 1.40 GHz, capable of going up to 1.70 GHz. We don’t get to see such low clock speeds on GPUs anymore but do keep in mind that the 3090 is over four years old.

Higher clock speeds mean that a GPU can process more data per cycle, leading to better performance in tasks that rely a lot on the GPU. With such clock speeds, the 5090 can execute tasks at a much faster rate.

Size

The RTX 3090 was larger than the RTX 2080, and the sizes only increased from there. The RTX 3090 Founder’s Edition measured 313 mm x 138 mm x 55 mm (LxWxH), and Nvidia decided to stop this trend with the RTX 50-series GPUs as bigger and heavier cards cause significant stress on the PCIe slot and cause GPU sag.

The RTX 5090 is smaller in size than the RTX 3090 and the RTX 4090. It measures 304 mm x 137 mm x 48 mm (LxWxH), so only the width is comparable, but the length and height of the card saw a good reduction. A benefit of this is that the RTX 5090 is a 2-slot card while the RTX 3090 is a 3-slot card, meaning it takes less space inside the PC case and on the motherboard. You should have access to the secondary PCIe slots with the RTX 5090, where they would’ve been covered with a 3090.

Nvidia reduced the PCB size, which spanned the majority of the GPU. This meant the fans couldn’t pass air directly through the card as the PCB was in the way. However, with the RTX 5090, the PCB is smaller, giving the fans full access to pass the air, which should result in better cooling performance.

Performance

We’ve got the performance benchmarks for the RTX 3090 for various games running at 1080p, 1440p, and 4K. For the RTX 5090, we’ll use the available benchmarks till we’ve tested it ourselves in-house. The focus will be on native performance, as DLSS 4 would give the 5090 more of an advantage, while the 3090 only has DLSS 3.5 to work with.

We’ll use Cyberpunk 2077, a graphically demanding game and a mutual title that was run using both GPUs. The RTX 5090 managed 208 FPS at 1440p, while the RTX 3090 delivered 87 FPS. At 4K, we got 124 FPS from the next-gen flagship, while the RTX 3090 trailed behind with 44 FPS.

So, after one generation, the performance difference is leagues apart, and the RTX 3090 can’t match the 5090’s capabilities. If we were to put a number on it, the RTX 5090 would be 82% better at 1440p and 95% better at 4K than the RTX 3090. This is a performance difference that no amount of overclocking the GPU or AI upscaling can bridge.

However, being too good at something also has its issues, as there was a CPU bottleneck at 1440p, meaning the processor couldn’t keep up with the GPU. So, while you’re investing $2,000 or more in this GPU, you’ll have to spend more to pair it with one of the best CPUs for the RTX 5090, like the Ryzen 7 9800X3D. This will ensure that any titles running at 1440p wouldn’t be affected by a CPU bottleneck and playing games at 4K shouldn’t be an issue even if you pair it with a less powerful CPU.

Features

Nvidia has an upscaling technology called DLSS, which both the RTX 5090 and 3090 can access. This technology helps users boost the number of frames they’re getting in games, something that helps them get a performance boost even on weaker cards.

During CES 2025, Nvidia announced the latest version of this technology, DLSS 4. At the moment, RTX owners are currently using DLSS 3.5, but they’ll make a switch to the newer version once it launches alongside two RTX 50-series cards at the end of this month.

However, while both the RTX 5090 and 3090 will be able to access DLSS 4, one important feature called Multi Frame Generation will be exclusive to RTX 50-series cards. This feature will help generate three additional frames for every traditionally rendered frame, allowing users to get a significant boost in games.

In fact, Nvidia currently claims that the RTX 5070 will offer performance similar to the flagship RTX 4090 for just $549. While it can’t achieve this with just its specifications, we believe that it will be able to deliver the same performance with the help of DLSS 4.

Nvidia also shared a video on their YouTube channel, comparing different DLSS 4 technologies. While they’re getting only 27 FPS FPS with DLSS off, DLSS 3.5 makes the number jump to 142. On the other hand, DLSS 4 delivers over 100 more FPS, around 245.

Aside from that, if you are always worried about latency, Nvidia also revealed Reflex 2 during CES 2025. For those who don’t know, Reflex is a technology that helps reduce latency in games, leading to an overall smooth gameplay experience. Now, Reflex 2 will be paired with Frame Wrap, and together, they’ll work to reduce more latency than Reflex 1.

In the image below, Nvidia demonstrates how there was 56 ms latency in The Finals, which dropped down to 27 ms after turning on Reflex 1. However, Reflex 2 took it further down to just 14 ms, which is pretty impressive. If you’re a competitive gamer, this will benefit you greatly. But do keep in mind that Reflex 2 will first debut on RTX 50-series cards and make its way to other series later in the future.

Image source: Nvidia

So, in short, while the RTX 5090 already has a significant performance advantage over the RTX 3090 in terms of specifications, the two mentioned features will help it deliver even better performance.

RTX 5090 vs RTX 3090 – price comparison

As far as the price is concerned, the RTX 5090 will launch with an MSRP of $1,999 according to Nvidia. This will be, however, for the base model, meaning third-party variants will carry a higher price tag – somewhere between $2,200 and $2,400.

On the other hand, the RTX 3090 was launched with a price tag of $1,499. These days, you can get it for anywhere between $1,350 and $1,450, depending on the variant.

5076
RTX 50 series

Which Nvidia GeForce RTX 50 series GPU are you interested in the most?

Early impressions

Based on the information we have so far, the RTX 5090 will most likely be the most powerful gaming GPU on the market when it launches. It offers 32 GB of GDDR7 VRAM, the highest we’ve seen yet in a gaming GPU. On top of that, with DLSS 4 and Reflex 2, the 5090 won’t let any other card get anywhere near it. The RTX 3090, on the other hand, is still a decent option, but it is far behind the newest flagship GPU.

However, we do think that the 32 GB of memory on the RTX 5090 is a bit overkill. We already have the 4090 with 24 GB of VRAM, which is also more than what most games need. Nvidia should have focused on improving the VRAM on their budget and mid-range options instead.

In any case, once the review of our card is live, we’ll make sure to update this guide with real-world benchmarks.

About the Author

Abdul is a tech writer and Editor for PC Guide, specializing in all things tech, gaming, and hardware.