We already know that the GeForce RTX 4080 gets the newer AD103-301 chip, which, unlike the default variant, does not have a comparator circuit in the chip’s voltage control system. But it doesn’t make any fundamental differences there, and production costs are said to have decreased by about a dollar. But she also got an improvement GeForce RTX 4090 (at least as far as the Founders Edition versions are concerned), but here it seems that the differences between the chips are in something else. In her case, we have a chip AD102-301-A1which has not been officially talked about any changes, however unofficially it seems that the chip has a reduced max voltage from 1.1V to 1.07V.
However, since we know that chip consumption increases with the square of the voltage, even such a small reduction in voltage could theoretically result in a consumption reduction of about 5% (which would correspond to about 20 W for the RTX 4090) provided that the same frequencies are achieved. But the question is whether the overclocking capabilities will not suffer from the reduced limit. But the first reports say that users have managed to overcome clocks of 3.0 GHz. The new cards also have a different PCB (16F4) and a new BIOS 95.02.26.00.02. Older cards with the original chip cannot be flashed to it.
:strip_exif()/i/2005545080.jpeg?f=meta "Microsoft invested hundreds of millions of dollars in supercomputer ChatGPT")
