NVIDIA RTX 4080/4090 May Leverage GDDR7 Memory w/ Power Draw of at Least 450-550W [Rumor]

Some additional info regarding NVIDIA’s next-gen RTX 40 series “Ada Lovelace” graphics has surfaced. While much of it is essentially a rehash of what we already knew, the report from MLID claims that the top-die AD102 is being built to support GDDR7 memory speeds, and possibly GDDR7 memory itself. While the chances of the GeForce RTX 4090 or 4080 launching with GDDR7 memory are slim, NVIDIA might release a mid-cycle “Super” refresh featuring the next-gen memory. Expect super-fast GDDR6/GDDR6X memory on the RTX 4080 and 4090 either way.

The other juicy tidbit is with respect to the power target of the top-end RTX 40 series cards. Turns out, NVIDIA was initially planning for a TBP of 400-450W as is the norm with these large dies. However, in a bid to outperform (or at least level with) AMD’s chiplet based Radeon RX 7900 XT and 7800 XT (RDNA 3) graphics cards, it was pushed up to 600W. You can expect the GPUs to draw at least 450-550W now, meaning a power supply of 1000W is necessary for the RTX 4090.

Finally, there’s the matter of the launch dates. According to MLID, the RTX 4080/4090 will be launched in September or October. However, this may be a paper launch, with the actual hard launch planned for the holiday season.

The remaining data pertains to the specifications which are more or less as we have reported over the past week: