“I’m not a talker”, so the leaker kopite7kimi writes on Twitter before putting pen to paper. new alleged technical specifications of the GeForce RTX 4080 by NVIDIA, coming to market by the end of the year. In the past few hours, the same leaker had retouched the information relating to the RTX 4070 upwards.
Unfortunately, the “job” of the leaker is not easy, often those who have sources within the producers or along the supply chain come across different information, not only because companies can change projects – for the most diverse reasons – but also because it is not uncommon for NVIDIA and AMD to spread false information to mislead the competition. Of course, and I hope there are some readers of very old date, the leak situation seems slightly improved compared to the times of Fudo.
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
But apart from this disgression, we come to the new specifications. According to the leaker, the GeForce RTX 4080 may have 9728 CUDA core and 16 GB of GDDR6X memory at 21 Gbps on a 256-bit bus. There is also talk of a TGP of 420W and a score that in 3DMark Time Spy Extreme should be around 15000 points. What changes compared to the previous information? Basically the number of CUDA cores, less than the 10240 indicated above.
The score indicated in the test of the 3DMark suite would still remain high, well above the approximately 11200-11300 of the current top of the NVIDIA range, the GeForce RTX 3090 Ti, even if clearly the comparison will be made with the AMD RDNA 3 video cards likely to arrive. starting in November.
In summarythe three flagships of the new series of NVIDIA arriving by the end of the year should be configured as follows, according to the latest rumors to be taken, as always, with caution:
- GeForce RTX 4090: 16384 CUDA core, 24 GB of GDDR6X memory at 21 Gbps on 384-bit bus. The TGP should be 450W.
- GeForce RTX 4080: 9728 CUDA core, 16 GB of GDDR6X memory at 21 Gbps on 256-bit bus. The TGP should be 420W.
- GeForce RTX 4070: 7680 CUDA core, 12 GB of GDDR6X memory at 21 Gbps on 192-bit bus. The TGP should be 300W.