'%20x='0'%20y='0'%20height='100%25'%20width='100%25'%20%0A%20%20%20%20%20%20%20%20%20%20xlink%3Ahref='data:image/jpg;base64,/9j/2wBDAAYEBQYFBAYGBQYHBwYIChAKCgkJChQODwwQFxQYGBcUFhYaHSUfGhsjHBYWICwgIyYnKSopGR8tMC0oMCUoKSj/2wBDAQcHBwoIChMKChMoGhYaKCgoKCgoKCgoKCgoKCgoKCgoKCgoKCgoKCgoKCgoKCgoKCgoKCgoKCgoKCgoKCgoKCj/wgARCAAGAAoDASIAAhEBAxEB/8QAFgABAQEAAAAAAAAAAAAAAAAAAAYH/8QAFAEBAAAAAAAAAAAAAAAAAAAAAP/aAAwDAQACEAMQAAAAvtrD/8QAIRAAAgEDAwUAAAAAAAAAAAAAAQIDAAQFBxEUEhMVIZL/2gAIAQEAAT8AysWYl0+1ZXmsLny9zw3W5dezHvF81ikePF2aSuXkWFAzE7lj0j3X/8QAFBEBAAAAAAAAAAAAAAAAAAAAAP/aAAgBAgEBPwB//8QAFBEBAAAAAAAAAAAAAAAAAAAAAP/aAAgBAwEBPwB//9k='%3E%3C/image%3E%3C/svg%3E)
The Nvidia RTX 5070 and the RTX 4090 are some of the most capable GPUs in the market. However, in the launch event, Nvidia advertised the new 70-class video card to be as capable as the last-gen flagship, which raised a few eyebrows. However, post launch, we came to know that the new 5070 can match the BFGPU from the last generation only with multi-frame generation and DLSS turned on.
When put head to head at native resolution, how does the performance of the two GPUs compare? In this article, we aim to answer that question by looking at the real performance gaps between the cards.
The Nvidia RTX 5070 and RTX 4090 are pretty distant in terms of native rendering performance
While the RTX 5070 and 4090 can match each other with AI upscaling tech turned on, the cards are quite different in terms of the underlying hardware powering them.
Trending
Let's have a look at the specs sheets of either GPU before delving into the performance differences.
Specs comparison
The 5070 and 4090 are very different graphics cards in terms of their DNA and what they try to achieve. While the last-gen flagship is a high-end GPU with 16,384 CUDA cores, the fastest 24 GB GDDR6X video memory with over 1 TB/s bandwidth, and a whopping price tag of $1,599, the Blackwell offering is much more modest.
In terms of the underlying hardware, you get 6,144 CUDA cores, only slightly more than the last generation RTX 4070 Super. While the VRAM standard has been upgraded to GDDR7, it continues to be just 12 GB, which limits the memory bandwidth to just 536 GB/s.
The detailed specs sheets of either GPU are as follows:
Performance comparison
With AI upscaling turned off, the RTX 5070 takes a back seat, as expected. The performance gap is pretty large. If you're playing at 4K, it can make all the difference between playable and unplayable. The 70-class card can still reliably pull off 1440p without major performance hiccups, easing things for gamers.
Below is a look at the frame rate numbers achieved by either GPU in some of the latest video games. They have been sourced from the YouTube channel Testing Games.
On average, the RTX 5070 is about 40% to 77% slower than the RTX 4090 across various games. However, the GPU still holds its ground and delivers comfortable performance at native resolutions effortlessly. This makes it a worthwhile purchase for 1440p gamers.
With frame generation turned on, however, the story flips on its head. The 5070 comes out on top with DLSS set 4x (for every real frame, AI generates four more frames to effectively multiply the frame rates by 2x-3x). Below is the performance you can expect.
With frame generation turned on, the 5070 delivers about 5-6 FPS more than the last-gen flagship, which can be pretty surprising given that the native hardware powering the Blackwell card is way slower.
Read more: Nvidia RTX 5090 vs RTX 4090
Overall, both the RTX 5070 and RTX 4090 are some of the fastest GPUs in the market and are designed to handle the latest titles with little to no performance hiccups. However, the new 70-class card is much slower when it comes to raw rasterization performance.
Nevertheless, most gamers won't be able to tell the difference between DLSS-enhanced and native anyway (that's how good the AI has gotten). The major hindrance remains the small library that supports the latest DLSS 4 tech.
