The specifications of NVIDIA’s upcoming Ampere gaming flagship, the GeForce RTX 3080 Ti have been confirmed by a GPU-Z validation submission. There’s nothing new in the leak, it just confirms what we already knew and how close to the RTX 3090 the Ti will perform. The primary difference between the two will be in terms of the memory buffer. While the RTX 3090 has 24GB of GDDR6X memory, the RTX 3080 Ti has half as much, just 2GB more than the RTX 3080. As you can imagine, there will be a much smaller performance delta between the 3080 and 3080 Ti this time around.
| GPU | Core Count (FP32) | Memory | Bus Width | Pixel Fillrate | Bandwidth | Base Clock | Boost Clock | Memory Clock |
|---|---|---|---|---|---|---|---|---|
| RTX 3080 | 8704 | 10GB GDDR6X | 320-bit | 164.2 GPixel/s | 760.3 GB/s | 1440 MHz | 1710 MHz | 1188 MHz |
| RTX 3080 Ti | 10240 | 12GB GDDR6X | 384-bit | 186.5 GPixel/s | 912.4 GB/s | 1365 MHz | 1665 MHz | 1188 MHz |
| RTX 3090 | 10496 | 24GB GDDR6X | 384-bit | 189.8 GPixel/s | 936.2 GB/s | 1395 MHz | 1695 MHz | 1219 MHz |
Looking at the above specs, it’s pretty clear that the RTX 3080 Ti closer to the 3090 than the 3080 itself. The pixel fillrate is just a few figures lower: 186.5 GP/s vs 189.8 GP/s. In terms of bandwidth, the RTX 3080 Ti has a peak capability of peak 912.4 GB/s while the 3090 manages 936.2 GB/s. On the other hand, the RTX 3080 is limited to 760GB/s. The boost clocks are roughly the same while the RTX 3090 once again has a slight edge when it comes to the memory clock. Note that this is the unmultiplied memory clock. The effective clock is sixteen times as much.
