NVIDIA’s upcoming RTX 3060 Ti has also been confirmed recently, and its complete specifications have also been leaked. Now it seems that we still have a release date. According to MyDrivers.com (via Videocardz), the GA104-based RTX 3060 Ti will be released on November 17. We also heard the same information from our sources (more specifically, this is the embargo date), so it was not flagged as a rumor-although Jensen can and is likely to adjust the date a bit, just to make sense Leak the scene randomly.
NVIDIA RTX 3060 Ti will be released on November 17
NVIDIA GeForce RTX 3060 Ti will be equipped with GA104-200 GPU, with 4864 CUDA cores and 8GB GDDR6 memory, with a clock frequency of 14 Gbps. The bus width will be 256 bits and the total bandwidth will be 448 GB/s. Since the board will require less power, the TBP should be about 200W, while the clock of the reference design is about 180W. Compared with the RTX 2070’s 175W, this is an increase of 5W, which is in line with our expectation that the Samsung node will have a reasonable level of power efficiency compared to the TSMC node.
Remember, the reason why the performance of NVIDIA’s RTX 3090 and RTX 3080 cards cannot scale linearly is the bottleneck of the game engine and binary files, which cannot handle so many cores. We know that due to the benchmark performance of these cards in software such as vRAY and Octane (designed to handle a large number of graphics functions), the hardware and driver stack can be linearly expanded.This means that as you reduce the number of cores, performance will Do not Shrink linearly. In fact, I fully hope that the RTX 3060 Ti can beat the RTX 2080 clock in terms of clock! Assuming that the clock is slightly less than the RTX 2080 Ti, it should be easy to compare with the previous flagship product.
The RTX 3060 Ti series card has been on the market for a long time, and Videocardz has finally received confirmation.
This is the complete RTX lineup:
NVIDIA GeForce RTX 30 series “Ampere” graphics card specifications:
|Graphics card name||NVIDIA GeForce RTX 3060 Ti||NVIDIA GeForce RTX 3070||NVIDIA GeForce RTX 3070 Ti?||NVIDIA GeForce RTX 3080||NVIDIA GeForce RTX 3080 Ti?||NVIDIA GeForce RTX 3090|
|GPU name||Ampere GA104-200||Ampere GA104-300||Ampere GA102-150||Ampere GA102-200||Ampere GA102-250||Ampere GA102-300|
|Process node||Samsung 8nm||Samsung 8nm||Samsung 8nm||Samsung 8nm||Samsung 8nm||Samsung 8nm|
|Mold size||395.2 square millimeters||395.2 square millimeters||628.4 square millimeters||628.4 square millimeters||628.4 square millimeters||628.4 square millimeters|
|Transistor||17.4 billion||17.4 billion||28 billion||28 billion||28 billion||28 billion|
|TMU / ROP||152/80||184/96||232/80||272/96||312/104||328/112|
|Base clock||1410 MHz||1500 MHz||To be determined||1440 MHz||To be determined||1400 MHz|
|Boost clock||1665 MHz||1730 MHz||To be determined||1710 MHz||To be determined||1700 MHz|
|FP32 calculation||16.2 TFLOPs||20 tons||To be determined||30 tons||To be determined||36 racks|
|RT TFLOPs||32.4 TFLOPs||40 tons||To be determined||58 tons||To be determined||69 tons|
|Tensor TOP||To be determined||163 songs||To be determined||238 songs||To be determined||285 top|
|Memory Capacity||8 GB GDDR6||8 GB GDDR6||10 GB GDDR6X?||10 GB GDDR6X||12 GB GDDR6X?||24 GB GDDR6X|
|Memory bus||256 bit||256 bit||320 bits||320 bits||384 bits||384 bits|
|Memory speed||14 Gbps||14 Gbps||To be determined||19 Gbps||To be determined||19.5 Gbps|
|bandwidth||448 Gbps||448 Gbps||To be determined||760 Gbps||To be determined||936 Gbps|
|Three Gorges Project||180W?||220 watts||320W?||320W||320W||350 watts|
|Price (MSRP / FE)||$399 USD?||$499 USD||599 dollars?||US$699||899 dollars?||$1499 USD|
|Start (availability)||November 2020?||October 29||The fourth quarter of 2020?||September 17||The fourth quarter of 2020?||September 24|
AMD just recently released its Navi 22 series on October 28, which may be one of the reasons why NVIDIA decided to release Ti-based variants in the first place. Since the 3080 and 3090 cards are still out of stock, gamers are clearly eager for the next generation GPU. Oh, if you want to know about NVIDIA, then I provide you with some details.
It can be seen that the first few batches of Samsung’s 8nm node output are lower than expected and expensive. The company expects production (and wafer prices) to increase significantly in the coming months. In order to protect shareholders, the company has been cautious about the first few batches of orders, and has increased a large number of orders for subsequent batches. The bot program discarded the entire volume from the beginning, making this problem even worse. We expect large orders to begin to touch before the holiday season (assuming miners will of course not swallow everything).