NVIDIA has been the top dog in the graphics card industry for a while now. And, while there has been plenty of competition from AMD in the mid-range and budget options, the flagship NVIDIA graphics cards are still miles ahead of anything else on the market. This is due to NVIDIA’s commitment to introducing new features while others play catch up.
One of the staple features of the newer NVIDIA cards has been the support for hardware real-time raytracing. The feature, while still in infancy according to many, will revolutionize the way games look. Nvidia also marketed Ray tracing as a feature that should improve overtime as more games start to support it.
The dominant opinion in the tech world was that ray tracing hardware would get better over the next generations, and it is the future of gaming. Keeping the ray tracing feature aside, Turing architecture also brought good performance bump over the past generation of video cards. We got improvements in budget cards due to which 1080p gaming got more mainstream.
The new generation of Nvidia cards also brought support for DLSS or Deep Learning Super Sampling. This antialiasing technique allows Nvidia to provide better antialiasing on budget cards with the help of AI and Machine learning.
To keep the competition on their side in every target price, team green has some exciting new features and designs in the works. The RTX 3000 series is about to launch soon, and some accurate leaks are popping out. The leaks for high-end devices have been around for a while now. What’s more, interesting is the company’s approach towards the new budget cards.
The 70 series of cards from Nvidia target users that need an acceptably good performance at a reasonable price. These cards provide you with enough horsepower to run the majority of new games at results lower but close to top-end cards. It is safe to say that this is the product where the price translates to performance in the highest percentage.
The recent leaks talk about the next generation of Nvidia’s mid-tier cards RTX 3070 series. kopite7kimi over at Twitter talks about the specifications of these cards. These specifications look quite authentic, considering the fact that the same user has predicted a lot of details about other GPUs from Nvidia in the past, which proved to be true.
The post says that Nvidia will be releasing two new cards equipped with the new GA104 GPU. The past generation used TA104 GPU in devices such as RTX 2080 SUPER, RTX 2080 & the RTX 2070 SUPER. The top-end models this time will have GA102 GPU, and the GA104 will only be used in two cards. These cards might be named RTX2070 and 2070Ti.
Talking about performance, the updated GPU will offer some upgrades over the last generation due to IPC improvements. Users can expect higher performance and better thermals in this new design of GPU. Here is a more detailed look at the internals of both speculated cards.
RTX 3070 Specifications:
The Nvidia Geforce RTX 3070 is the cheaper of two models so that it will feature a slightly less performing GPU GA104-300 with a rumored memory of 8 Gigabytes. The memory technology would be GDDR6.
Although that is an older standard, it can still outperform the previous-gen card with proper implementation. This technology can go up to 18Gbps in testing conditions. In the last years RTX 2080, we observed speeds up of up to 15.5 Gbps.
If the Ampere architecture takes that value to 16 or above, this 256-Bit card will then have an output bandwidth of 512 Gbps as compared to last year’s 448Gbps on Turing architecture cards. In terms of processing cores, this card will be similar to an RTX 2080 with 2944 CUDA cores.
RTX 3070 TI Specifications:
It is confusing what Nvidia has done with their naming schemes over the past years. TI variant in the history of products from NVIDIA are the cards that have similar hardware specifications but some minor improvements and performance bumps. The cards in the 70 series from Nvidia have had a TI variant in the past, but the last we saw one, it was in pascal generation the GTX 1070TI.
The Turing generation did not have a TI model because Nvidia decided to release a refresh of their existing cards with the name “Super”. We could then say that the RTX 2070 Super could be a TI version of the card. With that said, we still don’t know if this card will be 3070 Ti because there is no robust naming scheme by the company yet.
Talking internals, this variant will have the proper GA104-400-A1 GPU with a massive 3072 CUDA cores. It will have a similar 8 gigabytes of memory configuration. There will be two boards of the cards PG141 and PG142 The difference between the two would be memory type. PG141 will use GDDR6X memory technology while the other one will incorporate the usual GDDR6.
The GPU setup besides architecture bump looks similar to the last generation. The TU104 GPU inside the RTX 2080 SUPER, which also had 3072 CUDA cores. The main performance improvement in this card over the 2080 Super will then come from the new GPU architecture and the faster clock speed because both of the GPUs use the same core configuration.
About the card variants, we have seen Nvidia do things like this in the past. One explanation for the two cards could be that these are testing variants, and the Retail version of the device will have a single option. The company will check the GDDR6 and GDDR6X cards to determine which provides the best performance at a price.
By this time, you can already guess that this card in terms of performance will be close but better than the existing RTX 2080 Super. The speculated price of this card is about $500. This points out that users will get performance similar to the current 699$ cards from NVIDIA.
All these talks are still rumored, and ultimately only Nvidia can provide the actual price. However, if we keep past behaviors of the company in mind, a good guess will be a price of around 500 dollars.
We already have the first custom card design for Nvidia GeForce RTX 3080 by Asus. The whole lineup launch will take place in a couple of moths. With AMD competing to provide the best value for their devices, Nvidia will probably make some aggressive pricing decisions like last year.
AMD is currently setting up the prices of their video cards based on RDNA 2 architecture. These cards will launch in the Fourth Quarter of this year. Now that AMD is all set to provide the best prices, Nvidia will also focus more on providing more value for their cards.