The latest market news, RTX 40 series graphics cards are getting closer and closer, multiple sources of information said that this year NVIDIA only plans to release a new core, which corresponds to only one new graphics card. The answer is self-explanatory. It is based on the AD102-300-A1 core RTX 4090, higher specifications of the AD102-45 core and relatively pro-AD104/AD106/AD107 core and corresponding products, all postponed to next year.
As for the timing of the RTX 4090, the fastest September release, October began to store for the market.
Previously the most controversy around RTX 4090 is power consumption, now the argument is that NV allows AIC to configure itself on 450/600W two-block TGP setting, quite spiritual.
In terms of price, the RTX 4090, which crosses the performance level of RTX 3090 Ti, will be cheaper, between $1499 (3090) ~ $1999 (3090 Ti). However, the RTX 30 Series has now broken the market price long after the fierce price cut, with the RTX 3090 already hovering near $1000 and the RTX 3090 Ti falling below $1500.
In this way, NVIDIA still attaches great importance to the RTX 30 Series inventory and is unwilling to allow any of the new cards that may have duplicate positioning to cause any interference with the old cards going into inventory.
Back to the specification level, RTX 4090 is expected to open 128 groups SM, integrated 16384 CUDA, 96MB L2 cache, matching 24GB G6X memory, etc., single precision floating point performance over 70T.
