Nvidia quietly updated the RTX 4090: Equipped with AD102-301 GPU
Two months ago, there were reports that Nvidia had begun offering two versions of the AD103-300 and AD103-301 for the GeForce RTX 4080. Galax was the first board partner to confirm the existence of the AD103-301 version. The version with the AD103-301 only had minor adjustments to its bill of materials (BOM), and there were differences in the PCB design, but there were no differences in performance or functionality. This is a normal operation that will help Nvidia’s partners reduce costs.
According to VideoCardz, some users have recently received the GeForce RTX 4090 Founder Edition and found that it is equipped with the AD102-301 GPU instead of the usual AD102-300, indicating that a new version has been released, just like the GeForce RTX 4080.
There is a new model ID in GPU-Z, and while the device PCI ID remains the same, the sub-ID has changed.
The maximum voltage of the AD102-301 has also been reduced from 1.1V to 1.07V, which can be confirmed using tools like MSI Afterburner and ASUS GPU Tweak. However, it is not clear whether the Founder Edition is restricted or if other similar products are the same.
It is expected that there will soon be comparison tests of versions with the AD102-300 and AD102-301 to see if there is any impact on performance. It is not just the GeForce RTX 4080 and RTX 4090; there have been reports that Nvidia’s upcoming GeForce RTX 4070 will also have two versions: the AD104-250, which requires a comparator circuit, and the AD104-251, which does not. The latter will have fewer components and is rumored to enter mass production several weeks later.
The GeForce RTX 4090 has 128 SMs, or 16,384 CUDA cores, with a boost clock of up to 2520MHz, 24GB of GDDR6X memory, a memory interface of 384 bits, and a memory speed of 21Gbps. Its default TGP is 450W, and the BIOS can set the maximum TGP to 660W.