If you’ve got an RTX 4090, then you’re already in the upper echelons of PC gaming, however, this business wouldn’t work if you were ever made to feel content with your PC. So, you’re probably tempted by the newly announced RTX 5090.
Let’s pit the 5090 vs the 4090 and see just how much performance you stand to gain if you decide to upgrade, starting, as usual, with the specs.
GPU | RTX 4090 | RTX 5090 |
CUDA Cores | 16,384 | 21,760 |
Core Frequency | 2.23 GHz | 2.01 GHz |
Boost Core Frequency | 2.52 GHz | 2.41 GHz |
Memory Capacity / Type | 24 GB GDDR6X | 32 GB GDDR7 |
Memory Bandwidth | 1008 GB/s | 1792 GB/s |
Memory Interface | 384-bit | 512-bit |
Memory Clock | 21 Gbps | 28 Gbps |
TGP | 450W | 575 W |
Power Connector | 12V-2x6 | 12V-2x6 |
Release Date | October 12th, 2022 | January 2025 |
MSRP | $1,599 | $1,999 |
The story hasn't changed when it comes to video outputs, so those with very specific video output needs aren't locked into the 4090 over the 5090 or verse vica.
GPU | RTX 4090 | RTX 5090 |
Outputs | 1x HDMI 2.1a, 3x DP 2.1a | 1x HDMI 2.1b, 3x DP 2.1b |
Unlike the generational half-step represented by the SUPER series we got at the beginning of 2024, the 50-Series, and 5090 specifically show a larger jump in performance. Some of which can no-doubt be attributed to the GDDR7 memory, compared to the GDDR6X used in the 40-Series and SUPER variants.
The headline here is that GDDR7 allows for faster memory, and more of it. Up to 33% faster than than GDDR6X. However, just because the GDDR7 standard allows for way more capacity and speed, does not necessarily mean that any given GPU that utilizes this type of memory will hit those specific benchmarks.
However, what will certainly make a difference is that the RTX 5090 has 32GB of VRAM vs the 4090’s 24GB. This means that it can store significantly more and higher detailed textures, and can handle ray-tracing a more easily. If you yearn to trace those rays, then more VRAM will allow you to do so without tanking your frame rate quite as much.
We also see a significant difference in total CUDA core count, with the RRTX 5090 holding 21,760 of the little computational lads, which is over 5300 more than the 4090’s 16,384. So performance is certainly higher, but it all depends on the price. Dang. What a smooth segue we have crafted here together.
The RTX 4090 originally launched back in 2022 for the immodest price of $1599, and the $1999 price of the 5090 is a significant step up, though perhaps not as much of a price bump as people feared.
Unsurprisingly, the RTX 5090 has been designed to almost max out the 12V-2x6 standard with a TGP of 575W. significantly more than the 450W TGP of the 4090. If you’ve already got a 4090 and designed that build with a bit of breathing room with more powerful PSU than necessary, you might be alright. (You can use our PSU calculator to make sure, though).
If you’re using an older card and have found yourself on this page to decide between the 4090 and the 5090, then when it comes to selecting a PSU, you should probably aim for a 1000W power supply at minimum. A higher capacity PSU is no bad thing, however. It can provide some future proofing, and if you go with a CORSAIR PSU with Zero RPM mode, then you’re more likely to stay within the Zero RPM bracket for longer.
If you're concerned about how to power the 5090 or indeed the 4090, refer to the infographic below that explains the different ways you can power the latest GPUs with CORSAIR PSUs.
المنتجات في المقالة