Full TU104 GPU
NVIDIA is using the "full implentation" of the TU104 GPU with its new GeForce RTX 2080 SUPER, with the full 3072 CUDA cores and 48 RT cores versus the 2944 CUDA cores and 46 RT cores in the RTX 2080.
The TU104 GPU is clocked at up to 1650Mhz base and 1815MHz boost on the GeForce RTX 2080 SUPER, while the TU104 in full on the RTX 2080 SUPER offers up 11.2TFLOPs of compute performance versus the 10.1TFLOPs on the original RTX 2080. This results in Tensor FLOPs at 89 TFLOPs fersus 81 TFLOPs.
But it's the faster GDDR6 memory on the RTX 2080 SUPER that is exciting.
Fastest GDDR6 Ever
NVIDIA injected some faster GDDR6 memory into the RTX 2080 SUPER with its 8GB of GDDR6 clocked at 15.5Gbps, up from the 14Gbps on the regular RTX 2080. It has the same 256-bit memory bus which results in 496GB/sec of memory bandwidth compared to the 448GB/sec on the RTX 2080.
The new RTX 2080 SUPER Founders Edition has a higher TGP with 250W versus 210W on the regular RTX 2080.
- Page 1 [Introduction & Pricing]
- Page 2 [Specs: Turing GPU & Fastest GDDR6 Ever]
- Page 3 [Revised Look, Same Cooler]
- Page 4 [Detailed Look]
- Page 5 [Test System Specs]
- Page 6 [Benchmarks - Synthetic]
- Page 7 [Benchmarks - 1080p]
- Page 8 [Benchmarks - 1440p]
- Page 9 [Benchmarks - 4K]
- Page 10 [Thermals, Power, Noise]
- Page 11 [Overclocking]
- Page 12 [Performance Thoughts]
- Page 13 [What's Hot, What's Not]
- Page 14 [Final Thoughts]