Still, because of the company’s history, its next-gen lineup is the most anticipated and subsequently the most covered. We’ve known for quite a while now that RTX 40-series will have significantly higher power requirements than the current-gen Ampere GPUs. Everything from the entry-level offerings, all the way up to the flagships, there’s a consistent uplift in TDPs across the board. Now, today, a new leak has come forward from @kopite7kimi who has shared the ‘true’ power limits for the Ada Lovelace GPUs. As expected, they’re around ~100W higher than previous-gen for each SKU. Interestingly, Ada Lovelace mobile GPUs are said to retain almost identical power requirements to their current-gen counterparts, despite desktop graphics cards being much more power-hungry.
— kopite7kimi (@kopite7kimi) June 18, 2022
Ada Lovelace power limit breakdown
According to the leaker, the flagship mobile GPU, AD103, is said to have a 175W power limit, the same as GA103 GPU that launched last year. Similarly, the next-in-line AD104 GPU also has a 175W power limit which is just 10W more than the GA104 SKU. That being said, the AD106 GPU carries a (roughly) 25W higher power limit than the previous-gen GA106 GPU. Moving over to desktop, the leaker brought forward a new power limit for the top-of-the-line AD102 GPU, which is now rumored be 800W. The interesting part is that the same leaker previously quoted 900W as the power limit for this exact same SKU. So, either one of these leaks is wrong. The most shocking entry in kopite’s list is the AD103 desktop GPU with a power limit of 450W. Compared to its predecessor, the GA103, the Ada Lovelace GPU has a 250W higher power limit. That’s a massive increase. To put that in even more perspective, the AD104 GPU is only rated 110W higher than its current-gen GA104 counterpart. Before you take all of these numbers for face value, I have to mention two important disclaimers. Firstly, TDP and power limit are not the same thing. TDP, or Thermal Design Parameter, is essentially the power requirement of your GPU. It indicates how much juice the card can pull from your power supply to function optimally. Power limit, on the other hand, is the hard cap that locks the GPU from going any higher in terms of wattage. This is done to save and preserve the life of the GPU so overclockers use modded BIOS to break past this limit in order to achieve record-breaking numbers. Therefore, the power limits shared by kopite won’t necessarily be the actual TDPs of the cards they’ll be in, such as the RTX 4090, RTX 4080, RTX 4070, and so on. That indicates the absolute maximum power that can be supplied to the board on factory settings. Still, even with this factor in mind, the RTX 40 series is getting a massive power, and subsequently performance, boost that will be interesting to compare to AMD’s more efficiency-heavy approach. In fact, AMD engineers have reportedly been heard touting that RDNA 3 will decimate NVIDIA in terms of efficiency across the entire product line. Whether that remains true or not, only time will tell.
