Nvidia will reportedly be taking a no holds barred approach with its upcoming AD102 “Ada Lovelace” die. We’ve previously reported on the insane power requirements that are rumored for Nvidia’s next-gen GPUs, but Moore’s Law is Dead on YouTube says they are confirmed based on discussing the topic with system integrators and OEMs. These companies are already planning upcoming systems based on the RTX 4090 GPU, which is due to launch in Q3. In addition to cranking its consumer GPUs to 11, the company will also be raising the power limit for professional cards too.
The popular YouTuber says although Nvidia was originally targeting an insane 850W Total Board Power (TBP), it’s now aiming for its flagship board to be 600W. This would be inline with the recently unveiled ATX 3.0 power connector for GPUs, which supplies exactly that amount of power through a single cable. It should be noted that 600W is that Nvidia is targeting for its reference design, a.k.a the Founder’s Edition. Partner boards will probably go even higher due to overclocking and more advanced cooling mechanisms. According to their sources, some partner boards might be in the 700W range. This is the same situation we are seeing currently with the RTX 3090 Ti. Nvidia lists its board power as 450W, but some ultra-thicc partner boards are pushing that number up above 500W. This is also why it’s not outlandish to think Nvidia will be hitting 600W for next-gen; it’s practically already there with Ampere. The RTX 4080 is rumored to be around 450W, but it could be as high at 500W.
Nvidia is reportedly cranking up the power to make absolutely certain it beats AMD’s RDNA3 GPUs. It will not be taking any risks whatsoever this time around, as it is anticipating AMD to come out swinging. It has good reason to worry too, as its monolithic AD102 die will be going up against AMD’s first multi-chip module (MCM) design. As we reported previously, the flagship RDNA3 GPU will reportedly feature seven chiplets, so it should be a force to be reckoned with.
Moore’s Law says system integrators are already preparing for the arrival of these monster GPUs. They have begun looking at bigger cases and are also stockpiling high-wattage power supplies. Suffice to say, for the next-gen a 1,000W power supply will be the bare minimum. Hybrid cooling solutions might also be more common, which sport an air-and-liquid cooling mechanism.
It’s also bumping up the power on its workstation cards. They are rumored to be named RTX L6000 or RTX L8000, and will run somewhere between 320W and 375W. The current flagship A6000 GPU has a TBP of just 300W, so bumping that up to 375W is a big change. Moore’s Law says Nvidia is testing these GPUs with the current 8-pin connector. However, due to its higher power demands, it’s also testing the new 16-pin connector as well.
Now Read:
- Nvidia Says Its RTX 30-Series GPUs are ‘Restocked and Reloaded’
- Asus to Lower Nvidia GPU Prices up to 25 Percent
- Nvidia Reportedly Prepping Two New RTX 3050s, One with Just 4GB of VRAM
from ExtremeTechExtremeTech https://ift.tt/YcMtXuA
ليست هناك تعليقات:
إرسال تعليق