The generational leap from the 20 series to 30 series GPUs is the highest we have ever seen in graphics cards evolution till now. While Nvidia says that the RTX 3070 beats the RTX 2080 Ti, RTX 3090 which is $200 more expensive than the 2080 Ti is said to be almost 2 times faster than the RTX 2080 Ti literally.
Here I have listed out the important and most significant differences between the two and how RTX 3090 has gone so ahead in gaming.
Turing vs Ampere
In my previous posts of comparison between the RTX 3070 vs RTX 2070 and RTX 3080 vs RTX 2080, I have already explained how much of a difference does the Ampere makes over the Turing architecture, I will briefly explain why the RTX 3090 performs much better than the 2080 Ti.
Though the Turing architecture was the first to introduce Ray-Tracing and DLSS(Deep Learning Super Sampling) which made games look much more realistic than ever. With Ray-Tracing, the shadows and lighting become dynamic and with DLSS, the game looks better but with significant improvement in performance.
DLSS uses Artificial Intelligence and there are several different types of operations that both Turing and Ampere architecture use for speeding up the AI processes. The Ampere architecture brings additional TF32 and FP64 precisions that speed up the AI processes up to 20 times helping the GPUs achieve much higher performance.
Turing was a kind of Beta testing or a worst-case scenario which at the time of launch of RTX 20 series graphics cards were known due to the slow performance of RTX 20 series GPUs including the 2080 Ti in games when the RTX was turned ON. RTX 2080 Ti wasn’t able to provide proper 60 fps even at 1080p resolution, so the whole aim to achieve great graphics with smoother gameplay was impossible.
Thanks to Ampere, RTX 30 series GPUs have turned the tables completely and not only the RTX 3090 is capable of providing much higher frame rates with Ray Tracing ON but even the RTX 3070 beats the RTX 2080 Ti in this task.
A new FE design
RTX 3090 features a completely different designed heatsink that has fans on opposite sides of the heatsink. Both works as exhaust and the hot air don’t dissipate in every direction like the RTX 2080 Ti FE. RTX 3090 uses a very dense and thick 3 slot design that makes it almost 1.5 times bigger than RTX 2080 Ti Founders Edition. This should help it in achieving cooler performance but the real-world tests are still not available at the time of writing this article.
Cuda Cores are over 9000!
Amazingly, the RTX 3090 brings over 9000 Cuda cores, the exact number is 10496 compared to the 4352 Cuda Cores on RTX 2080 Ti. Cuda Cores do numerous complex calculations for graphics processing. In simple words, the more they are, the faster your GPU will be.
Related:- Best Power Supplies for RTX 3090
Remarkably faster memory
Just upgrading the memory from GDDR5 to GDDR6 in GTX 1650 Super, the GPU made 10-15% boost in gaming performance over the original GTX 1650. This is because the memory speed and bandwidth of GDDR6 is higher than that of GDDR5.
On the RTX 3090, the memory is upgraded from all the corners. As compared to the 11GB on RTX 2080 Ti, RTX 3090 uses 24GB of unbelievable VRAM with GDDR6X standard. The memory interface is increased from 352-bit to 384-bit and memory bandwidth is increased from 616GB/s to 936.2GBs/.
While the RTX 2080 Ti uses a good 14Gbps GDDR6 VRAM, the RTX 3090 uses a comparatively much faster 19.5Gbps of GDDR6X memory. This alone does a great job in increasing the potential of RTX 3090 and not to mention, other specs that include Core clock, RT cores and Tensor cores are also increased by a good margin, the power consumption of RTX 3090 though is 100W higher.
|SPECS||RTX 2080 Ti||RTX 3090|
|VRAM||11GB GDDR6||24GB GDDR6X|
Pricing and Conclusion
I have already mentioned about pricing above. The actual cost of RTX 2080 Ti is around $1200 while the RTX 3090 is said to be available for around $1400. Though the difference is $200 only, the performance of both the cards has a different ratio. While the tests and reviews are currently going ON, we already know that the RTX 3090 is definitely faster but once we have enough data, I will add the Performance section as well.[box]