In fact, as you can read from the message posted on Twitter, the board, which will use the PG136 / 139-SKU360 PCB, will be based on the AD103-300-A1 GPU, with 9,725 CUDA cores (or 76SM), accompanied by 16GB of GDDR6X with a speed of 21Gb / s on a 256-bit bus, for a maximum total bandwidth of 672GB / s. Unfortunately, the operating frequencies are not specified, but presumably we should expect clocks in the range of 2.0-3.0 GHz considering the use of TSMC's N4 production node. Finally, the TGP should be around 420W, while the score achieved in the 3DMark Time Spy Extreme test will be more or less than 15,000.
I'm not a chatterbox, but I have to make some updates. I hope you don't mind.
a possible RTX 4080,
PG136 / 139-SKU360
AD103-300-A1
9728FP32
256bit 16G 21Gbps GDDR6X
total power ~ 420W
TSE ~ 15000
Now I have completed the latest update for 4090, 4080 and 4070.
- kopite7kimi (@ kopite7kimi ) August 5, 2022
Ultimately, we are talking about a more powerful product than the current top of the Ampere range, namely the GeForce RTX 3090 Ti, by 35%. A great leap forward compared to the model it will replace (the RTX 3080). From the technological point of view, in addition to the greater computing capacity, the RTX 4080, as well as the rest of the RTX 40 family, will be able to rely on all the latest technologies of the company, such as the fourth generation Tensor cores, third generation RT cores. generation, NVENC Encoder and NVCDEC Decoder, as well as support for the latest API and exclusive features such as DLSS, Reflex, Broadcast, Resizable-BAR, Freestyle, Ansel, Highlights, Shadowplay and G-SYNC.