As expected (and a horrible timed fallout with EVGA), Nvidia went ahead to announce what is next for their GPU family of GeForce under the Lovelace architecture, based on Ada Lovelace, the historically recognized first programmer after helping via letter Charles Babbage to idea an mechanical general-purpose computer.
The Lovelace generation of GeForce as a graphical processor is an historic one because this time the AI and all the technical advantages are all metaverse in mind.
Yees, the metaverse seems to be the future for social Internet an Nvidia definitely wants to take a piece of the cake under the efforts of Meta and as member of the Metaverse Standards Forum, but employing their layer as the Omniverse which is giving the 3D push to the metaverse with emphasis that virtual world and real world collide together.
To physically have this possible, the Lovelace architecture or the 3rd generation RTX, includes 76 billions transistors built on a 4nm manufacturing process and a RacerX demo was showcased with heavy graphical details and ray tracing was processed with just ONE GPU.
Lovelace also includes a third generation RT Core that handles the Ray TRacing necessities of a graphical content or game and a third generation DLSS AI processor that has an AI entity to calculate a predictive new frame before a game scene changes.
With all this, Nvidia claims that gaming performance is twice as good compared to past generations.
Officially for 2023, we will have the “new GPU Heavyweight” RTX GeForce 4090 starting at $1599 and Nvidia just limited to say that it is from 2 to 4 times faster than the GeForce 3090 Ti thanks to its 16,384 CUDA cores, 22.234GHz an high as 2.52GHz with boost, 1,321 TEnsor Teraflops, 191 RT Teraflops and 83 shader Teraflops.
Nvidia recommends a 850 Watts PSU and a Ryzen 5900X for CPU if you are showing this GPU that in my opinion, is for professional users like animation studios or television stations.
The RTX GeForce 4090 will be released on October 12.
Surprisingly, a two variant will have the consumer high-tier RTX GeForce 4080 with one version with 12GB and another with 16GB, both are slated for a November release but no exact day was given.
The 12GB 4080 variant starts at $899 and includes 7,680 CUDA Cores, 639 Tensor-TFLOPs, 92 RT-TFLOPs, 40 Shader-TFLOPs, and GDDR6X memory which, Nvidia claims that as powerful or a little more than the GeForce 3090 Ti.
Finally the 16GB variant at $1,199, includes 9,728 CUDA Cores, 780 Tensor-TFLOPs, 113 RT-TFLOPs, 49 Shader-TFLOPs of power, and GDDR6X memory, making it twice as fast as the GeForce 3080 Ti with 10% less of power consumptions.
As you can notice and BIG difference from 2020 where GeForce 3090 and 3080 are close on an october launch and 3070 on a november launch, the RTX GeForce 4070 was nowhere to be seen and evidently is bound for 2023.
Finally and as expected, Ampere will still be in main line along Lovelace as announced by Nvidia with MSRP being reformatted and the only Ampere confirmed to have new prices is the GeForce 3060 with a MSRP of $328.
Reportedly this is all part of Nvidia's plan on an inusual scenario facing inventory overstock in a world with semiconductors shortage and too much to just write off as loss, so this remaining months of 2022 looks like a good time to built your PC depending on your needs.