NVIDIA took some leeway to create a GeForce RTX 4090 Ti

NVIDIA took some leeway to create a GeForce RTX 4090 Ti

NVIDIA took some leeway to create a GeForce RTX 4090 Ti

The GeForce RTX 4090 undoubtedly promises great performance, such as to clearly detach the RTX 3090 Ti as declared by NVIDIA in the past few hours. In the future, however, it could come a hypothetical GeForce RTX 4090 Tias the company has chosen not to implement a GPU with all the resources active inside it in the RTX 4090.

At the center of the card announced last night we find GPU AD102 with 16384 CUDA core, 512 Tensor core and 128 RT core. The 384-bit memory bus, on the other hand, connects the chip to 24GB of 21Gbps GDDR6X memory. The GPU in question for having a number of resources greater than 12.5% with respect to the data communicated. AD102 is shown in its complete form 18432 CUDA core, 576 Tensor core e 144 RT core. The RTX 6000 intended for the professional sector approaches these numbers.

Read This Now:   Apple: the new Magic Mouse could be even more magical and change shape

NVIDIA’s choice of opportunities clearly, and certainly not new: the production of the GPU (76.3 billion transistors against 28.3 billion of the GA102 of RTX 3090 Ti) with TSMC 4N process it does not allow to realize at this moment a large amount of fully functional chipsat least not in the amount that NVIDIA needs for a GPU that looks to creators and gamers.

AD102 has dimensions of 608.3 mm2, occupies a smaller area compared to the 628.4 mm2 of GA102, but integrates 170% more transistors thanks to the TSMC production process that allows a decidedly higher density, equal to about 2.8 times, compared to that of Samsung (8N) used for the Ampere GPUs.

How NVIDIA performed last generation with the GA102 of GeForce RTX 3090 and then of 3090 Ti? The GA102 in full form seen on the RTX 3090 Ti features 10752 CUDA cores, 336 Tensor cores and 84 RT cores, while the chip mounted on the RTX 3090 sees instead 10496 CUDA active cores, 328 Tensor cores and 82 RT cores.

Read This Now:   AMD is considering launching exclusive RDNA2 GPUs for mining and selling them directly

Between the 3090 and the 3090 Ti there is therefore a resource gap of only about 2.5%, an aspect that we have seen translate into a limited performance leap between the two solutions, while the TGP rose by 100W as a result of even more powerful memories. With the new generation, NVIDIA a wider margin of maneuver was maintainedwhich suggests that the rumors about the 600W of a hypothetical 4090 Ti might be spot on, considering the 450W TGP of the RTX 4090.


Notice: ob_end_flush(): failed to send buffer of zlib output compression (1) in /home/gamefeve/bitcoinminershashrate.com/wp-includes/functions.php on line 5373

Notice: ob_end_flush(): failed to send buffer of zlib output compression (1) in /home/gamefeve/bitcoinminershashrate.com/wp-includes/functions.php on line 5373