GeForce RTX 4000 (“Ada”): Return of the Titan with 48 GB and 900 watts?

Rumors already speak of Nvidia’s AD102 GPU alone up to 450 watts, now a new Titan with 48 GB GDDR6X and up to 900 watts is buzzing through the rumor mill. At the end of February there was already speculation about a TDP of up to 850 watts for the GeForce RTX 4090. Values ​​that the new Titan could achieve.

The rumor mill was mostly unanimous and certifies that Nvidia’s upcoming Ada generation has a TDP of 450 to 600 watts at the upper performance end, depending on the model.

The titan of the rumor mill

The usually well-informed Twitter user @ kopite7kimi would now like to have heard from my own sources about a test board that could be intended for a possible new Titan RTX. This should even be able to absorb up to 900 watts, but it is still unclear whether the model will appear like this – the test board is said to have three 16-pin power connectors, which could even provide 1,800 watts according to specifications and is therefore very likely far from the series.

Rumors about the largest Ada GPU AD102 (Those 1, Those 2)
  • TSMC N5 or N4 (custom process)
  • up to 18,432 CUDA cores
  • up to 48GB GDDR6X at 24Gbps @ 384-bit
  • up to 2× 16-pin (12+4-pin) PCIe 5.0 (“12VHPWR”)
  • bis zu 12 Graphics Processing Clusters
  • up to 144 streaming multiprocessors
  • up to 72 texture processing clusters
  • up to 600 mm² die size
  • up to 900 watts TDP

*) not officially confirmed!

A new Titan, the editors last tested the Nvidia Titan RTX of the Turing generation in March 2019, would also explain the new PCIe 5.0 connector (“12VHPWR”) and its 600 watts by itself.

More shaders, more cache, more consumption

The AD102 graphics processor, which is said to be the largest GPU in the GeForce RTX 4000 series, continues to follow a classic monolithic approach and is rumored to be used on the GeForce RTX 4090 and a possible Titan RTX. Until now it was thought that the GeForce RTX 4080 should also be powered by the AD102, but now the smaller AD103 is supposed to take over this role, while the GeForce RTX 4070 is said to be based on the AD104 graphics processor. The AD106 and AD107 then round off the generation down, they say.

The GeForce RTX 4080, meanwhile, is said to have 16 GB of GDDDR6X memory and a comparable TDP to the current GA102 GPU (up to 450 watts), while the GeForce RTX 4070 is said to have 12 GB of GDDR6 and consume around 300 watts.

Compared to the GeForce RTX series aka Ampere, the new graphics cards are also said to have a massively enlarged L2 cache – similar to AMD’s Infinity Cache.

GeForce RTX 3000 vs GeForce RTX 4000* L2 cache comparison

Showdown towards the end of the year

Nvidia’s GeForce RTX 4000 series should compete against Navi 31, 32 and 33 and RDNA 3 in multi-chip module design on the Radeon RX 7000 series for a big GPU showdown by the end of the year at the latest.

The new PCIe 5.0 connector with up to 600 watts and the new power supplies, some with well over 1,000 watts of power, are also a clear indication of where the next GPU generation should go.

more on the subject

Still to be enjoyed with caution

Even if the highest possible numbers are “worked” again and these are still rumors that should be treated with great caution, the GeForce RTX 4000 series aka Ada should have a significantly higher power consumption than Ampere.



Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.

On Key

Related Posts