Nvidia has come out swinging with its RTX 50-series GPU announcement at this year’s CES, but perhaps the biggest shock announcement was that the RTX 5070 is set to deliver the same performance as the god-tier RTX 4090, for a whole lot less too.
The first thought on our mind was just how close the RTX 5070 is to the RTX 4090. Thanks to CES 2025, we have some information on the new card that we can work with – namely, its specs. Of course, we’ll be updating this article with more in depth information once we put the RTX 5070 through its paces in our testing lab, so stay tuned.
For now, let’s take a look at how these two GPUs stack up on paper, and see if the RTX 5070 is gearing up to be everything Nvidia says it will be.
Spec comparison
Here’s a breakdown of each cards specs.
| Specifications | RTX 5070 | RTX 4090 |
| Cores | 6144 | 16384 |
| Base clock speed | 2.16 GHz | 2.23 GHz |
| Boost clock speed | 2.51 GHz | 2.52 GHz |
| Memory | 12GB GDDR7 | 24GB GDDR6X |
| Memory interface | 192-bit | 384-bit |
| Bandwidth | 532.6 GB/s | 1.01 TB/s |
| TDP | 250W | 450W |
Just at a glance, it is apparent there is a massive difference in hardware as the RTX 4090 has more beefed up specifications. While the RTX 5070 does feature the new GDDR7 memory type, other features of it don’t match up to that of the RTX 4090.
Memory
The RTX 4090 has 24GB of GDDR6X VRAM, while the RTX 5070 has 12GB of GDDR7 VRAM. In terms of capacity, the RTX 5070 is no match for the flagship GPU. That being said, GDDR7 has a per-pin bandwidth of 32 GB/s, while GDDR6X tops out at 24 GB/s per pin.
So, this is a difference of 8 GB/s but the RTX 4090 has double the VRAM, so it doesn’t seem like the RTX 5070 can directly compete with it purely based on hardware. Also, for those who want to max out their in-game settings at high resolutions, more VRAM is crucial which the RTX 4090 has plenty of.
Bandwidth
Another stark difference between these two graphics card is the RTX 5070 has a 532.6 GB/s bandwidth while the RTX 4090 has a 1.01 TB/s bandwidth. On top of that, the RTX 4090 has a 384-bit interface connection, while the RTX 5070 still has a 192-bit connection.
So, not only does the RTX 4090 have a much higher bandwidth, but it also has a superior memory interface width, meaning it can process more data than the RTX 5070 without running into issues. Bandwidth is an important factor in a GPU as it can determine its performance and efficiency.
Cores
The RTX 4090 has 16,384 CUDA cores, while the RTX 5070 maxes out at 6,144. This is a difference of 10,690 CUDA cores, which is massive. The RTX 4080 Super has 10,420 CUDA cores and this is more than that.
So, regarding cores, this is very one-sided. Even if the RTX 5070 has 4th-generation Ray Tracing cores and 5th-generation Tensor Cores, it can’t compete with the RTX 4090 in raw performance.
RTX 5070 vs. RTX 4090 – price comparison
The RTX 5070 has an MSRP of $549, while the RTX 4090 had an MSRP of $1,599. However, depending on the third-party manufacturer, it can cost anywhere from $2,000 to $2,500+. So, the price difference is huge, but you’re also getting much better performance.
Final word
Ultimately, on paper, it looks like the RTX 4090 has the edge, however, the RTX 5070’s Multi-Frame Generation (which is exclusive to the 50-series) is likely going to be the secret weapon here. We’re keen to test this card ourselves so we can get a more real-world idea of how it performs.
The RTX 5070 is set to release in February 2025, so be sure to check back in with us them to see which one comes out on top in our testing.
