You might be wondering why we’re comparing the RTX 4070 Ti and RTX 3080 when they aren’t even from the same class of GPU. But separating GPUs into different families is just a marketing strategy companies employ to upsell their products. And even though the 3080 was released a few years ago, it remains one of the most popular GPUs in the market. Of course, Nvidia claims that the RTX 4070 Ti, being part of the latest RTX 40 series, is a more powerful GPU. But is it really?
Today, we’ll make an in-depth comparison of the two GPUs in question. We’ll explore the differences in specs and performance. We’ll also share insights into their price range to see which one gives you the best bang for your buck.
Specification Comparison: RTX 4070 Ti vs RTX 3080
| GPU Specifications | RTX 4070 Ti | RTX 3080 |
| CUDA Cores | 7680 | 8704 |
| RT Cores | 60 | 68 |
| Tensor Cores | 240 | 272 |
| Architecture | Ada Lovelace | Ampere |
| Base Clock Speed | 2310 MHz | 1440 MHz |
| Boost Clock Speed | 2610 MHz | 1710 MHz |
| Processor Size | 4nm | 8nm |
| Memory Type | GDDR6X | GDDR6X |
| VRAM/Memory Size | 12GB | 10GB |
| Maximum Resolution | 7680 x 4320 | 7680 x 4320 |
| Recommended PSU Wattage | 750 Watts | 750 Watts |
| TGP/TDP | 285 Watts | 320-350 Watts |
| MSRP | $799 | $699 |
Looking at the spec chart, you’ll notice right away that while the 4070 Ti got major improvements in some areas, its CUDA cores, RT cores, and tensor cores seemingly got a downgrade. Does that make it less powerful? Well, numbers don’t always paint the complete picture.
You should also check out: RTX 4070 Ti vs RTX 3070 Ti
What’s New with the RTX 40-Series
There are a lot of improvements with the RTX 40-series, but the three that stand out the most are its use of a new architecture, improved RT cores and tensor cores, and support for DLSS 3, which previous series don’t have.
New Architecture
The RTX 40 series got a complete makeover with its new architecture. The Ada Lovelace is based on TSMC’s 4nm N4 technology, making it 53% smaller than the previous Ampere chip, which was based on Samsung’s 8nm process. The reduction in its node process size made it more power efficient. What’s even more impressive is that Nvidia was able to fit 76 billion transistors in this new architecture, almost tripling Ampere’s 28 billion transistors.
Improved RT Cores and Tensor Cores
The new architecture comes with a huge improvement in its tensor core and RT core. The fourth-gen tensor core is functionally similar to the ones first introduced on the Hopper H100 data center GPU but much faster. The Ada Lovelace tensor core uses the new FP8 Transformer Engine, which helps increase the GPU’s throughput speed by up to five times.
Ada’s third-gen RT core, the latest from Nvidia, offers two times faster ray-triangle intersection throughput performance. It also comes with two new hardware components – A displaced Micro-Mesh (DMM) Engine and Opacity Micromap (OMM) Engine. The DMM Engine speeds up the Bounding Volume Hierarchy (BVH) build time by as much as ten times. The OMM Engine, on the other hand, accelerates ray-tracing of alpha textures often used for particles, fences, and foliages.
DLSS 3.0 Support
Aside from the numerous hardware upgrades, Ada Lovelace also offers DLSS 3.0 support. This helps reduce latency while increasing frame rates, especially when running high-resolution applications like 4k games. DLSS was already a technological breakthrough when first introduced as it helps reduce the GPU workload by predicting the next pixel using AI.
But DLSS 3.0 takes this a step further with its use of Nvidia’s latest Frame Generation Technology, allowing the video card to predict the next frame instead of just pixels. This significantly improves how individual frames are redrawn, so you can expect better image rendering even for CPU and GPU-intensive games. DLSS 3.0 won’t be coming to previous RTX series GPUs. That’s because it’s dependent on the Optical Flow Accelerator component and the fourth-gen tensor cores on ADA GPUs to power it up.
Performance Difference: RTX 4070 Ti vs RTX 3080
At first glance, you’ll notice a big difference in the number of CUDA cores. Surprisingly, the RTX 4070 Ti has fewer CUDA, RT, and tensor cores. Of course, it’s normal to assume that it’s less powerful than the RTX 3080. But these CUDA cores benefit from SER (shader execution reordering), a process that rearranges the execution order of instructions within a shader program to improve parallelism so each CUDA core is optimally loaded. The 4070 Ti also offers the 4th-gen tensor cores and 3rd-gen RT cores, both of which are the latest addition to the RTX 40 series GPUs.
Plus, its processor size is just half the size of the 3080s, which accounts for the lower TGP of the 4070 Ti. It also got a massive boost in its clock speeds. With a base clock speed of 2310 MHz and a boost clock speed of 2610 MHz, you can expect top performance from this beast of a GPU. Plus, with its new and improved RT cores and tensor cores, you can play even the most graphics-intensive game at a high frame rate.
But that’s not to say the 3080 isn’t a good option. Although it’s a few years older than the 4070 Ti, it’s still a pretty powerful GPU with decent clock speeds and a good 10GB of VRAM. Its high number of CUDA cores also helps ensure you get the best performance possible while gaming or video editing.
Price Difference: RTX 4070 Ti vs RTX 3080
The RTX 3080 was released with an MSRP of $699, while the RTX 4070 Ti was $799. Both are on the pricey side – even at MSRP, which isn’t really much of a surprise given their premium features. If only their prices stayed that way. But no, they had to skyrocket out of control, especially back in 2021. If you see how much they cost now, you’d wish you had already bought one when they were first released.
Fortunately, the prices have somewhat come back to Earth. Of course, it’ll still be challenging to find a brand-new 3080 for less than $800. Well, there are refurbished GPUs selling for around $600. They may not have the same lifespan as a new GPU, but they should work just fine.
The same is true for the 4070 Ti. While its pricing is less volatile than the 3080, you’ll still hardly find it at its original price anymore. Most cards are priced between $800-$900, putting it in the same price range as the 3080. For almost the same price, you get higher performance and increased power with the 4070 Ti, making it a more sensible option.
RTX 4070 Ti on Amazon
RTX 3080 on Amazon
Power Consumption
Before purchasing a GPU, it’s very important you check how much power it consumes, as this will help you choose the right power supply for your system. The 3080 uses 350 Watts, while the 4070 Ti’s TGP is only 285 Watts. Both have a recommended PSU of 750 Watts because even if the 4070 Ti is 65 Watts less power-hungry, overclocking will spike the TGP up.
Normally, the more powerful a GPU is, the more energy it consumes – but not in this instance. The 4070 Ti is actually more power efficient, thanks to its smaller process node. This is a welcome change since we’ve noticed a pattern of increasing power consumption with each new model release in the previous Nvidia GPU series.
Benchmark Tests
Watch this video by Testing Games to see the benchmarks between the 4070 Ti and 3080 in different 4k games.
Is the RTX 4070 Ti Worth it?
The 4070 Ti is one powerful GPU. Its new architecture gives it better performance and faster clock speeds at only 285 Watts of power consumption. The only downside is that your wallet will be taking a serious hit. It’s a very pricey GPU. But then again, the 3080 is just as expensive. So if you must choose between the two, just go for the 4070 Ti. It’s also a great way to future-proof your system.
