The Nvidia GTC 2022 Keynote has brought the announcement of the latest RTX 4000 Series graphics cards, which operate on the new Lovelace architecture. The reveal of the next-generation Nvidia RTX 4000 series has everyone abuzz, alongside improved features such as DLSS 3 and 3rd-generation ray tracing. As of right now, the two GPUs in the RTX 4000 Series are the Nvidia RTX 4090 and the Nvidia RTX 4080, with the potential for the RTX 4060 and RTX 4070 to be released at a later date, though that has not been confirmed. One of the latest graphics cards to be announced as a part of the ongoing GTC 22 event, RTX 4080, is a direct successor to the successful last-gen RTX 3080 and will be available in two price variants depending on the memory capacity. The latest generation of GPUs from NVIDIA promises a lot through its powerful features.
RTX 3080, on the other hand, isn’t out of the picture just yet. It is still a worthy video card for those looking for a steal deal and who aren’t too enthusiastic about grabbing the best-in-class graphics cards. The long-standing supply chain turmoil has finally come to an end, leaving NVIDIA with more graphics cards to sell than ever. While previously announced GTX and RTX cards are now readily available at a fair price, many may choose to wait for the RTX 40 series, slated to enter the market in November. Below is an informative comparison between the RTX 3080 and RTX 4080 based on their revealed specifications, price points, and more. But with the release of new GPUs, it’s worth asking if the last generation of graphic cards is still worth your time. To see how the RTX 4080 compares against its predecessor, the RTX 3080, keep reading as we will be breaking it all down.
Price
The Nvidia RTX 3080 launched in September 2020 with a retail price of £649/$699. Since it has been out for two years, it can be found at varying prices, both more and less expensive than it was at launch, from third-party retailers like Overclockers or eBay. The Nvidia RTX 4080 will launch in November this year, though we don’t know the exact dates yet. Two configurations of the RTX 3080 will be available, with the 12GB variation starting at $899 and the 16GB variation costing $1199, making it the most expensive GPU when compared to its predecessor.
Specs
The Nvidia RTX 3080 features Ampere architecture which uses Samsung’s 8nm process and packs 26 billion transistors. It comes in two configurations, with the option of 10GB or 12GB GDDR6X memory, alongside 8704 or 8960 CUDA Cores, depending on the chosen configuration. The RTX 3080 (10GB) has a graphics card power of 320W, while the 12GB variant takes the graphics card power up to 350W. The Nvidia RTX 4080 also comes in two configurations. The first has 12GB GDDR6X memory, with a CUDA Core count of 7,680. The second option comes with 16GB GDDR6X memory and a CUDA Core count of 9,728, both using the next-generation Lovelace architecture.
Lovelace is built on TSMC’s 4nm process node, which packs 76 billion transistors, a massive increase on what can be found on Ampere cards. Generally, more transistors result in faster performance as the data is able to be processed and transferred at a faster rate. The RTX 4080 can consume less power than its predecessor, with the 16GB variation needing 320W, while the 12GB variation takes that down to 285W.
Performance
While we have reviewed the Nvidia RTX 3080, we haven’t had either configuration of the RTX 4080 in for testing yet, which means that we can’t definitively claim that one is more powerful than the other. However, we can use what we know about the RTX 3080 and what Nvidia is claiming about the RTX 4080 to gauge which GPU is most likely to come out on top. To start, we’re going to focus more on the most powerful version of the RTX 4080. This version of RTX 4080 has around 1000 more CUDA Cores than its predecessor, which suggests that it should be faster. CUDA Cores are developed by Nvidia and are essentially the GPU equivalent of CPU cores, with more Cores allowing for more parallel processing.
Plus, both versions of the RTX 4080 cards come with Lovelace architecture, featuring a 4nm node, which is a big upgrade from the previous generation. Nvidia has given us an idea of the power difference in the graph below. Plus, the RTX 4080 is privy to features that are exclusive to the RTX 4000 Series. Two of which are DLSS 3 and 3rd-generation ray tracing, which should ensure that gamers can experience enhanced visuals and higher frame rates with supported games.
Nvidia also claims that the RTX 4080 is between 2x and 4x faster than the RTX 3080 Ti, with the extent depending on which game you’re playing. Since the RTX 3080 isn’t quite as speedy as its Ti sibling, this implies that the RTX 4080 will be significantly more powerful than the RTX 3080. Until we get the RTX 4080 in for testing we don’t know the performance difference for specific games or workloads right now. Be sure to check in with Trusted Reviews soon, as we hope to review these new graphic cards over the coming months to see which ones offer the best performance power.
Architecture
For the new generation, there is an upgrade in the architecture. The 40 series will be based on the Ada architecture instead of Ampere. With Nvidia optimizing and improving the base architecture you can expect further performance for the same cards. Especially with the change in the foundry process between them. Looking at the chips, the RTX 3080 utilizes the GA102-200 GPU variant. Based on Samsungs 8nm process, it creates a die 628mm² in size. Managing to hold 28.3 billion transistors.
Whilst the RTX 4080 is rumored to feature an AD103-300 chip. Moving back to TSMC it now will be built on a 5nm process. So it will have a die size of 380mm² with an unknown transistor count. Although it’s a much smaller die size it is possible it might feature more transistors with a more compact size. Although it will also improve the efficiency of the architecture.
CUDA cores
CUDA cores are Nvidia’s parallel processor architecture. It allows for more efficient processing of graphics to bring to your screen. It is the Compute Unified Device Architecture and is a well-established system. With each generation further improving how well they work. So although we can compare the numbers on the cards, they may not work the same. And the final specification may change before the release of the card. The 3080 holds 8,704 CUDA cores and 8,960 in the 12GB model inside its chip. Whilst for now the RTX 4080 is going to have 9,728 in its 16GB model and 7,680 in its 12GB model.
TDP
Thermal design power is the total amount of energy the GPU will consume at full utilization. With more power-hungry cards running up your energy bill. This power is also converted to thermal energy. Which then is given off into your rig and room so it can be a consideration for your heating in Winter but also Summer.
Between these two cards, there is a massive increase in energy consumption. Although the 4080 was first rated at a higher level, it is now supposedly at a 340W board power.
VRAM
The video random access memory is there for a frame buffer and pixel storage. It enables a smoother experience with better framerate and latency. It’s especially useful for larger resolutions. As they have higher pixel counts to fill up many gigabytes. And requiring higher speeds to move that data. For the RTX 4080 vs 3080 comparisons, we see an increase in the VRAM. With 16GB of GDDR6X up from 10/12GB of GDDR6X memory. But also a standard 12GB model as standard. Along with the capacity, the clock is also increasing. Up by 4 Gbps to 23 Gbps with the newer memory chips. However, the memory bus is supposedly decreasing. Down to 256-bit from 320/384-bit. Which can lower the bandwidth overall for the newer card.
Conclusion
NVIDIA is presently delivering decent options for all kinds of gamers and creators. The RTX 40 series is yet to hit the market; hence, those looking for an immediate upgrade may find RTX 3080 (or the Ti) a better option. It is also almost $200 cheaper, and those looking for an overclocked version can also get favorable deals. That said, ones preferring monstrous graphical performance in video games and don’t mind spending the extra bucks should wait for the 4080 cards to drop. It offers a lot of advantages for live streaming and content creation, alongside improved features like Reflex and additional tool support.