During its CES 2022 keynote, NVIDIA announced the not-so-secret GeForce RTX 3090 Ti. Another completely unnecessary SKU that just goes on to show the present state of the GPU market (pretty much anything sells). It is based on the same GA102 die as the RTX 3090 and comes with a mild increment in shader count and faster GDDR6X memory. While the RTX 3090 packs a total of 10,496 cores across 82 SMs and 82 RT Cores, the Ti leverages the full-fledged GA102 die to increase the shader FP32 core count to 10,752 (across 82 SMs) and the RT Core count to 82.
Like the RTX 3090, its Ti variant also features 24GB of GDDR6X memory paired with a 384-bit bus. However, the memory clock has been bumped up to 21Gbps which is a new high for graphics memory, bringing the overall bandwidth up to 1,008 Gbps (on par with HBM2). The core clock is set to run at 1,560MHz at the base frequency and 1,860MHz under load.
The primary highlight of the RTX 3090 Ti, however, isn’t the performance though. It’s the TGP. The new Ampere flagship draws a whopping 450W under load, requiring a 1000W PSU for ideal operation. This is a large increase over the RTX 3090’s TGP of 350W for a smidge more frame rate. The PSU requirement of 1000W is also quite a step up over the 3090’s 750W recommendation. () 250W for 5-6% more performance is rather embarrassing.
NVIDIA RTX 3080 Ti 16GB Mobile GPU: Notebooks Starting at $2,499
NVIDIA also announced a new mobile flagship: The GeForce RTX 3080 Ti, unlike its desktop counterpart, is based on the mysterious GA103 die. With a shader (FP32) count of 7,424 across 58 SMs paired 16GB of GDDR6 memory via a 256-bit bus, you get a bandwidth of 512GB/s, the highest of any mobile GPU, and 2Gbps higher than the RTX 3080 mobile.