ASUS ROG Strix NVIDIA GeForce RTX 3070 Gaming
Excellent aesthetic design
Great cooling solution
High boost clock speed
On the higher end of the price spectrum
The flagship against the new middle generation
WePC is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Prices subject to change. Learn more
Does the top consumer card of the last generation bring the fight to the current mid-level GPU? To find out we compare the RTX 2080 Ti vs 3070, where you might be able to save some money for equal performance. Picking which is the best graphics card for you.
The 2080 Ti was released in October 2018, as the runner-up choice behind the Titan. Whilst the 3070 came two years later, taking the middle step in the 30 series lineup. So is it is enough to take the fight to the last-gen.
Excellent aesthetic design
Great cooling solution
High boost clock speed
On the higher end of the price spectrum
Without doubt, the highest performing consumer-grade GPU
Real time ray-tracing
4k gaming at ultra settings
0dB Technology
Expensive
Limited amount of games with ray tracing
The 20 series is based on Nvidia’s Turing architecture. Compared to the 30 series, which has the Ampere architecture. This gives some fundamental differences between the two generations in how they are built and how they perform.
In particular, the RTX 2080 Ti uses the TU102-300A die. Built with TSMCs 12nm process it is 754mm2 in size with 18.6 billion transistors. While the RTX 3070 has the GA104-300 GPU. Made with Samsungs 8nm process, able to first 17.4 billion transistors in a 392mm2 die.
RTX 2080 Ti | RTX 3070 | |
---|---|---|
GPU | TU102-300A-K1-A1 | GA104-300-A1 |
Shaders | 4,352 | 5,888 |
RT cores | 68 | 46 |
TMUs | 272 | 184 |
Memory | 11GB GDDR6 | 8GB GDDR6 |
Base clock | 1350 MHz | 1500 MHz |
Boost clock | 1545 MHz | 1725 MHz |
Memory clock | 14 Gbps | 14 Gbps |
Card bus | PCIe 3.0 x16 | PCIe 4.0 x16 |
Memory bandwidth | 616.0 GB/s | 448.0 GB/s |
TDP | 250 W | 220 W |
Beyond architectural changes, there are also differences in specifications. These offer an insight into where some other differences may come from.
One of the key generational differences is the PCIe bus used. This is the interface between the graphics card and the motherboard. Allowing for more data with the higher iteration of that interface.
The 40 series uses the PCIe 4th generation instead of the 3rd gen you find in the 20 series. This doubles the total bandwidth available to the card. For this, it increases the speed from 15.74GB/s to 31.51 GB/s.
Thermal design power is a rating of the maximum amount of energy the GPU can consume at max utilization. Also converted into heat energy it can provide some information about how much of a heater it will be. Between the 2080 Ti vs 3070, the newer 30 card has 30 fewer Watts. Minimizing its usage and heat output.
The cycle rate measures the silicon in the die’s ability to process instructions. The higher number allows the GPU to process graphics much faster and out to the display. Although not the sole responsibility for performance it is a candidate for overclocking to extract more from it.
Between the two cards, the 3070 is clocked at much higher rates than the 2080 Ti. It has a base clock 150 MHz higher. As well as a 180 MHz higher boost clock. Giving the newer card the advantage in processing speed.
VRAM of a graphics card is the buffer and pixel storage means. It lowers the latency between the CPU and GPU and allows for higher resolutions to work better. Since they require more pixels they utilize more VRAM space.
Comparing the two GPUs, they utilize the same type of memory with GDDR6. As well as being clocked at the same rate of 14 Gbps. However, the 2080 Ti has a higher capacity and bandwidth. With 3GB more, and 168GB/s more likely from the higher bus.
They make up the GPU die and are specialized in working on particular functions. We compare the shaders, RT cores, and TMUs. These work on shadows and shading, real-time ray tracing, and creating 3D objects respectively.
As previously seen the 30 series takes a big lead in the shaders, but lacks in RT cores, and TMUs. In this case, the 3070 has 35% more shaders. Whilst the 2080 Ti has 48% more RT cores and TMUs.
Comparing the RTX 2080 Ti vs 3070 in 3DMarks graphics benchmarks we see the pure performance. The 20 series cards get a score of 14,695 with a price to performance of 14. Whilst the 30 series achieves 13,679 with a price to performance ratio of 27.
Comparing in-game performance, we see the 3070 does outperform the 2080 Ti. It achieves performance on average 3-6% higher than the 20 series card. Across all resolutions, it gives a good fight for being a lower SKU.
Comparing the launch price of the two GPUs we can see a significant difference. The 2080 Ti, cost $999/$899, whilst the 3070 launched for $499/£469. So nearly half the price for the newer card.
Although you’re unlikely to find either for that price currently. The market is recovering from a shortage with GPU prices inflated. So currently a 3070 costs $1100, or £666 in the UK. Whilst the 2080 Ti requires searching in the second-hand market, where it isn’t going to be cheap either.
Stylish, gaming aesthetics
Strong materials and build design
High-end cooling solution
On the expensive side
The fastest gaming GPU on the market
4k Gaming at ultra settings
Real-time ray tracing
Very Expensive
Not many games currently make use of ray tracing
Comparing the 2080 Ti vs 3070 we find the 3070 easily being the better offering. With cheaper asking prices but on par if not slightly better performance. It is the better solution.
However, you can always wait for prices to continue dropping. Or wait for the next launch of GPUs with Intel ARC Alchemist or RTX 40 series.
In 3DMark, the 3070 reaches a score 7% higher that the 2080 Ti. Whilst in gaming it also beats out the old generation by 3-6% across resolutions. So it’s only a single-digit percentage increase but at a lower cost.
Looking at the clock speeds, we see some changes too between the two cards. The 3070 has an 11% higher base clock, and a 12% higher boost clock. But the memory clock stays the same memory clock of 14 Gbps.