Nvidia’s RTX 4090 may outperform RTX 3090

Nvidia’s RTX 4090 may outperform RTX 3090, The Nvidia GeForce RTX 4090 graphics card may deliver double the performance of its predecessor, according to recent rumor’s.

The rumoured specifications for the forthcoming RTX 4090 and RTX 4070 have leaked, and it appears like we’re in for a significant performance jump across the board. kopite7kimi, a well-known Twitter source for all things PC hardware, has some tasty news for you today.

The rumoured Nvidia GeForce RTX 4090 Ti and RTX 4080 were previously theorised about by Kopite7kimi, but this time the leaks are focused on the RTX 4090 and RTX 4070, exposing more of the lineup that will be available in a few months.

The most exciting piece of information comes from kopite7kimi’s tweet, which claims that the RTX 4090 will easily quadruple the performance of the RTX 3090. While it’s enjoyable to hypothesise, it’s much more thrilling when the speculation is backed up with the card’s speculated specs.

Indeed, the difference in specifications between the RTX 3090 and its Ada Lovelace cousin implies that the next-gen GPU might deliver similar results.

The specs of both the RTX 4090 and the RTX 4070

Apart from juicy details regarding the RTX 4090’s performance, kopite7kimi discussed the RTX 4090 and RTX 4070’s specifications. The key distinction between the two cards is that each is powered by a different GPU.

The RTX 4090, like the RTX 4090 Ti, has an AD102 GPU, but the RTX 4070 has an AD104 GPU. Despite the fact that the RTX 4090 and RTX 4090 Ti share the same GPU core, the non-Ti variant will have lower specs. Nvidia’s RTX 4090 may outperform RTX 3090

There will be 144 streaming multiprocessors in the AD102 GPU (SMs). That’s a 71 percent boost over the current top GA102 GPU, according to Wccftech.
There will be 18,432 CUDA cores on the next-generation GPU. The RTX 3090 Ti, Nvidia’s current top graphics card, has 10,752 CUDA cores, so we’re witnessing a significant rise in both SMs and cores.

Nvidia GeForce RTX 4090 is also rumored to come with 24GB of GDDR6X

As we go on to the card’s L2 cache size, the advantages continue. According to rumours, the AD102 will include up to 96MB of L2 cache. Because the GA102 only has 6MB, the Ada Lovelace GPU will have up to 16 times the memory of the current-gen Ampere equivalent.

The Nvidia GeForce RTX 4090 is expected to include 24GB of GDDR6X memory, 21Gbps clock rates, over 1.0TB/s bandwidth, and a 384-bit bus interface. Nvidia’s RTX 4090 may outperform RTX 3090

All of that performance, unsurprisingly, demands a lot of power, but we’ve already discussed the power needs of Nvidia’s RTX 4090. The base model’s estimated total board power (TBP) is now 600 watts, and that number will only rise for bespoke versions created by Nvidia’s partners. Given the card’s massive power consumption, we expect some of Nvidia’s partners to use hybrid cooling technologies to keep it steady.

Power Consumption of Nvivdia’s RTX 4090

In terms of power consumption, Nvidia’s GeForce RTX 4070 appears to take a more balanced approach. It is supposed to have 60 SMs, 7,680 CUDA cores, and a (still huge) 48MB L2 cache based on the AD104 GPU.

This is partnered with 12GB of GDDR6 memory and a 400GB/s bandwidth expectation, all of which is shared across a 192-bit bus. Wccftech predicts that this card will perform similarly to the RTX 3080 Ti, making the 300-watt power need much more realistic. Nvidia’s RTX 4090 may outperform RTX 3090

Nvidia is expected to introduce its new RTX 4000 “Ada Lovelace” GPU family later this year. As we get closer to the second half of 2022, Nvidia could start releasing teasers that corroborate or deny the leaks we’ve seen thus far.