New rumor about Geforce RTX 4000: Maximum power limits of the GPUs leaked

New rumor about Geforce RTX 4000: Maximum power limits of the GPUs leaked


from Maximilian Hohm
The usually well-informed leaker kopite7kimi claims to have found out the power limits of the Ada Lovelace GPUs. These are immensely high and range from 260 to 800 watts in the desktop area, while the powerful mobile cards are said to consume 140 to 175 watts. Read more about this below.

Nvidia’s new Ada Lovelace graphics cards are said to continue to rely on monolithic dies, while AMD has probably chosen an MCM design for the large Radeon models. The first leaks about the graphics cards offered plenty of room for speculation and, above all, the power limits of the models seemed to change on a daily basis. Now the well-known leaker kopite7kimi wants to know the final maximum power limits of the Nvidia pixel accelerators for both desktop and mobile models.

AMD Radeon RX 7000: model with 16,384 FP32 shaders?

The flagship chip AD102, on which a rumored Geforce RTX 4090 (Ti) is said to be based, is said to offer a maximum power limit of absolutely untimely 800 watts and is only used in desktop graphics cards with this power consumption. The AD103, which is to be installed on a Geforce RTX 4080, should be one step below. The desktop model should draw level with the already quite wasteful Geforce RTX 3090 Ti with 450 watts. The mobile offshoot, on the other hand, should only get by with 175 watts, which would speak for greatly reduced clock rates, less memory and a smaller expansion of the AD103.

Interestingly, AD104, a chip attributed to the Geforce RTX 4070, is said to use the same power targets of AD103, although “only” 7,680 CUDA cores are expected here instead of 10,752 CUDA cores. A card from the upper middle class in the desktop area would then be allowed to consume 400 watts. The only chip that shouldn’t have absolutely insane power consumption is the AD106, which ends up at 260 watts in the desktop model and 140 watts for the mobile offshoot.

However, these numbers have not yet been finally confirmed by Nvidia and they are expressly the maximum possible power consumption of these GPUs, which do not necessarily reflect the Total Graphics Power (TGP) that a graphics card is allowed to consume. However, the same consumption between AD103 and AD104 is particularly surprising and may imply a smaller difference between the two SKUs than last, while an AD102 card could definitely be clearly ahead of both.

Source: kopite7kimi



Reference-www.pcgameshardware.de