Today Nvidia is allowing us to lift the lid on GeForce RTX 4080 performance figures and this huge, expensive, but ultimately extremely powerful – the second Ada Lovelace architecture GPU to be released behind the RTX 4090 – will set you back a minimum of $1,199. Let’s see the Comparision, Nvidia RTX 4080 Vs RTX 4090 Vs RTX 3090 Ti: Which Should You Buy?
| NVIDIA | RTX 4090 | RTX 3090Ti | RTX 4080 |
| Boost Frequency | 2.52GHz | 1.86GHz | 2.5GHz |
| Cuda Cores | 16384 | 10752 | 9728 |
| Tensor Cores | 512 (4th Gen) | 336 (3rd Gen) | 304 (4th Gen) |
| RT Cores | 128 (3rd Gen) | 84 (2nd Gen) | 76 (3rd Gen) |
| Memory Bus | 384-bit | 384-bit | 256-bit |
| Dedicated Memory | 24GB GDDR6X | 24GB GDDR6X | 16GB GDDR6X |
| L2 Cache Size | 73728 kb | 6144 kb | 65536 kb |
| TDP | 450 W | 450 W | 320 W |
| Launch Price | $1599 | $1999 | $1199 |
The specifications are interesting because in a lot of ways the RTX 4080 is a lesser card than the likes of the RT 3090 Ti – a narrower memory bus, less memory, fewer Cuda cores – but trounces it with a massive 2.5GHz peak boost frequency, next-generation Tensor and RT cores and an L2 cache ten times the size. There’s also DLSS 3.0 that for now is RTX 4000-only.
Is my case big enough?
This is a big graphics card and the same size as the RTX 4090 – at least in Founders Edition guise. it’s 5.4in wide, requires three PCI slots, and has plenty of clearance underneath for cooling. It’s also 12 inches long. Most standard cases will be fine, but it’s worth checking these requirements, especially with smaller cases.
Test system and benchmarks
The test system uses the Nvidia release driver for the RTX 4080 and the latest drivers as of November 10th and 14th for other Nvidia and AMD cards along with a fully up-to-date version of Windows 11. I used an Intel Core i9-12900K, an Asus ROG Strix Z690 Apex motherboard, Kingston 6,000MHz Fury DDR5 memory, a Kingston Fury Renegade SSD, and Thermaltake Toughpower GF3 1350W PSU.
Other GPUs were the RTX 3090 and RTX 3090 Ti OC, courtesy of Palit in the form of its GameRock models, and AMD RX 6950 XT.
Another interesting note is that there are two variants of the RTX 4080: one with 16GB of GDDR6X memory, which will retail for $1,199, and another with 12GB of GDDR6X memory that will cost $899. This would not be the first time Nvidia did this, as the company released two different versions of its RTX 3080 graphics cards – the launch model included 10GB of memory, while another version released nearly two years later included 12GB of memory.
But memory (and pricing) isn’t the only difference between the two RTX 4080 variants. The 16GB model has more RT, Tensor, and CUDA cores and requires at least 750W of power. In contrast, the 12GB model has a higher base and boost clock and requires at least 700W of power.
If anything the RTX 4080 has reaffirmed what a monstrously fast graphics card the RTX 4090 is, with it enjoying lofty leads in a lot of games. However, the RTX 4080 was also quicker than anticipated, sticking close to the flagship in a lot of benchmarks and often enjoying huge leads over the RTX 3090 Ti, leaving a huge gulf between it and the RTX 3080.
There isn’t much left to say other than this is a huge, powerful, and extremely expensive way to add a graphics card to your gaming PC, but with the addition of a reasonably quiet cooler, low temperatures, and a potentially very useful feature in DLSS 3.0, the RTX 4080 certainly won’t disappoint if you can afford it. Of course, it’s also worth waiting to see what AMD has in store with its RTX 7900 XTX that’s released soon.