Sarfraz
With the launch of RTX 3080, you might have wondered what improvements you would get if you upgrade from the RTX 2080/S to 3080. Generally speaking, most of the time people wouldn’t like to upgrade from one generation to a next-generation graphics card in the same tier as the improvements aren’t worthy of the price you pay.
But the RTX 3080 is a game-changer and has put the whole RTX 2000 series in a trash can. I am assuming that you already be knowing about how fast is the RTX 3000 series GPUs are but if you want to know what improvements did the RTX 3080 made over the RTX 2080/S, you should take a look at the most important changes Nvidia did below.
Architectural improvements
The biggest game-changer for the new-gen graphics cards is definitely the new Ampere architecture which makes the RTX 3080 blazingly fast and taking a long lead ahead of RTX 2080/S. While the Turing architecture of RTX 2080/S is definitely a good one over the architectures that were used in the graphics cards before, it did not do its job properly as expected.
You can say that Nvidia did some sort of Beta testing with the RTX 2000 series GPUs. Turing architecture made it easy for the RTX 2080/S to do Ray Tracing and some AI operations like DLSS(Deep Learning Super Sampling) which are a new trend in games.
While the GPU RTX 2080 was released in 2018, there are still not enough games that have support for Ray Tracing but surely we have some popular games like Shadow of the Tomb Raider, Battlefield V, CyberPunk 2077 which is about to be released, Doom Eternal etc. that do utilise this technique for real-time dynamic shadows and lightings.
Most of the gamers do want to play with “RTX-ON” but the RTX 2000 series GPUs that include a high-end card like RTX 2080/S is unable to provide sufficient performance even at 1080p resolution though you can play almost every game with this card at 1440p or 4K with 60 fps with RTX-OFF.
Ampere architecture now does this job very well with its 2nd generation RT cores and 3rd generation Tensor cores. The Ampere architecture is based on the 12nm technology unlike Turing which uses 14nm process and includes two new precisions which are the TF32 and FP64 for the acceleration of AI operations that now can speed up the processes up to 20 times which means faster and better performance in games using DLSS.
DLSS if you don’t know is an Nvidia RTX technology that uses AI which maintains the detail and also improves the image quality while increasing the frame rates significantly.
Related:- RTX 2080 Ti vs RTX 3090
Faster Memory
We already saw how Nvidia gained over 10-20% performance with the use of faster memory in already existing graphics cards in the GTX 1600 series. The GTX 1650 and GTX 1660 use GDDR5 memory but the GTX 1650 Super and GTX 1660 Super have GDDR6. GDDR6 has faster speed and more bandwidth than GDDR5 and this alone makes the GTX 1650 and 1660 Super GPUs faster without any increase in the cost.
The improvement now from GDDR6 to GDDR6X is even greater. While the GDDR6 memory comes with a speed of 12-14Gbps, GDDR6X runs at 19.5Gbps. Therefore naturally that is a significant improvement in the RTX 3080. Also, the memory is increased by 2GB, so now the RTX 3080 has 10GB VRAM while the RTX 2080 has 8GB.
The memory interface is also increased from 256-bit to 320-bit but currently, the memory bandwidth is officially not shown on Nvidia’s RTX 3080 page at the time of writing this article. Many sources indicate that it is 760GB/s which is almost 1.7X the bandwidth of RTX 2080.
Specs Comparison
| SPECS | RTX 2080 | RTX 3080 |
|---|---|---|
| Cuda Cores | 2944 | 8704 |
| VRAM | 8GB GDDR6 | 10GB GDDR6X |
| Memory Speed | 14Gbps | 19.5Gbps |
| Memory Interface | 256-bit | 320-bit |
| Bandwidth | 448GB/s | 760GB/s |
| Clock Speed | 1515/1710MHz | 1440/1710MHz |
| RT Cores | 46 | 68 |
| TDP | 215W | 320W |
The most increase Nvidia did with the RTX 3080 is the Cuda Cores. 2944 Cuda Cores on RTX 2080 vs 8704 on RTX 3080 is seriously unbelievable and surpasses the RTX 2080 Ti too. Core clocks are similar but the power consumption is a little bit higher sitting at 320W compared to 215W of RTX 2080. This should be expected with a lot of increase in specs numbers.
Design
If the Founders edition of both the GPUs is considered, they are completely different in design compared to the RTX 2070 vs RTX 3070 where things were somewhat similar. RTX 2080 uses a decent dense heatsink with 2x fans directly on top of the heatsink for cooling while the RTX 3080 uses fans on opposite sides.
The fans are actually more of push-type than intake which dissipates the heat in a particular direction from both sides. We currently don’t have the real temperature tests of the Founders Edition but if you are planning to buy a custom AIB RTX 3080, then you can expect some good designs for cooler operation.
Related:- Best CPU for RTX 3080
Performance
Source:- Nvidia.com
According to Nvidia, RTX 3080 is almost twice as fast as the RTX 2080 with RTX ON in games like Control, Minecraft and Borderlands 3. While these are ‘Their’ numbers but they did show comparison gameplay between the RTX 3080 and RTX 2080 Ti which is, of course, faster than the RTX 2080. Here the RTX 3080 is almost 40-50% faster than the 2080 Ti in most instances. The game they used was Doom Eternal with settings max at 4K resolution.
Pricing and Conclusion
The initial launch price of RTX 2080 was $699 and the launch price of RTX 3080 is also $699. For this price, it really doesn’t make sense to go with any of the RTX 20 series graphics cards especially the RTX 2080 Ti which gets beaten by the RTX 3070 too for almost $700 less.
Due to RTX 3080’s launch, the price of RTX 2080 Super which is faster than the RTX 2080 has dropped below $600 and may continue to fall due to the faster $500 RTX 2070.
[box]
[/box]
