According to prominent leakster, kopite7kimi (@kopite7kimi), the RTX 4090 could ship out with a base clock of 2235Mhz and boost clock of 2520MHz, plus a game clock of 2750MHz. To put those numbers into another context, the card’s supposed boost clock is almost 50% higher than those belonging to the RTX 3090, which is set at 1695MHz.
— kopite7kimi (@kopite7kimi) July 4, 2022 For that matter, the RTX 4090’s alleged base clock alone is approximately 60% higher than the RTX 3090’s, which is set at 1395MHz. Beyond that, the rest of the information posted by kopite is more or less on the ball with whatever they’ve previously posted about NVIDIA’s unreleased GPU. As a quick primer, the RTX 4090 is expected to lead the launch, alongside the rumoured and more powerful RTX 4090 Ti, use the AD102-300-A1 GPU, have at least 16384 CUDA cores, a 384-bit memory bus, 24 GDDR6X graphics memory running at 21Gbps, and have a bare minimum TDP requirement of 450W.
On another note, kopite’s post also reiterates the specifications for the RTX 4080 and RTX 4070. As before, the former is expected to be fitted with a slightly less but still as powerful AD103-300-A1, have 10240 CUDA cores, a 256-bit memory bus, 16GB GDDR6X graphics memory running at a speed of 21Gbps, along with a 420W TDP. For the RTX 4070, it is expected to run on a AD104 GPU, although which one exactly is yet to become public. As for the other details, the card should have 7168 CUDA cores, an oddly smaller 160-bit memory bus, 10GB GDDR6 graphics memory running at 18Gbps, and a 300W TDP. Based on these details, it also seems more or less like a confirmation that the RTX 4060, which isn’t mentioned here, may just have a higher TDP than the RTX 3070.
NVIDIA is expected to pull back the curtains on its GeForce RTX 40 Series later this year, starting in September through next year, one card at a time, from it’s flagship RTX 4090 and right through the RTX 4060. (Source: Videocardz)
