Thanks to some detective work by ComputerBase, we’ve discovered that NVIDIA’s Ampere mobile offerings have as many as 28 variants. Yes, that’s right. 28 variants of just 3 GPUs. We’ve yet to see the budget-range RTX 3050 and the 3050 Ti, so expect that number to swell up to 30. Remember the GTX 1650?
| GPU | TGP (W) | FP32 Cores | Base clock (MHz) | Boost (MHz) | Bus Width (bits) | Memory Speed (Gbps) |
|---|---|---|---|---|---|---|
| RTX 3060 laptop GPU Max-Q | 60 | 3,840 | 817 | 1,282 | 192 | 12 |
| RTX 3060 laptop GPU Max-Q | 65 | 3,840 | 975 | 1,357 | 192 | 12 |
| RTX 3060 laptop GPU Max-Q | 70 | 3,840 | 1,050 | 1,402 | 192 | 12 |
| RTX 3060 laptop GPU | 80 | 3,840 | 900 | 1,425 | 192 | 14 |
| RTX 3060 laptop GPU | 85 | 3,840 | 1,035 | 1,485 | 192 | 14 |
| RTX 3060 laptop GPU | 90 | 3,840 | 1,163 | 1,530 | 192 | 14 |
| RTX 3060 laptop GPU | 95 | 3,840 | 1,215 | 1,567 | 192 | 14 |
| RTX 3060 laptop GPU | 100 | 3,840 | 1,267 | 1,605 | 192 | 14 |
| RTX 3060 laptop GPU | 105 | 3,840 | 1,305 | 1,642 | 192 | 14 |
| RTX 3060 laptop GPU | 110 | 3,840 | 1,342 | 1,680 | 192 | 14 |
| RTX 3060 laptop GPU | 115 | 3,840 | 1,387 | 1,702 | 192 | 14 |
| RTX 3060 Founders Edition | 170 | 3,584 | 1,320 | 1780 | 192 bits | 15 Gbps |
| RTX 3070 laptop GPU Max-Q | 80 | 5,120 | 780 | 1290 | 256 bit | 12 Gbps |
| RTX 3070 laptop GPU Max-Q | 85 | 5,120 | 855 | 1365 | 256 bit | 12 Gbps |
| RTX 3070 laptop GPU Max-Q | 90 | 5,120 | 930 | 1410 | 256 bit | 12 Gbps |
| RTX 3070 laptop GPU | 115 | 5,120 | 1.110 | 1,560 | 256 bit | 14 Gbps |
| RTX 3070 laptop GPU | 120 | 5,120 | 1,170 | 1,590 | 256 bit | 14 Gbps |
| RTX 3070 laptop GPU | 125 | 5,120 | 1,215 | 1,620 | 256 bit | 14 Gbps |
| RTX 3070 Founders Edition | 220 | 5,888 | 1,500 | 1730 | 192 bits | 15 Gbps |
| RTX 3080 laptop GPU Max-Q | 80 | 6,144 | 780 | 1245 | 256 bit | 12 Gbps |
| RTX 3080 laptop GPU Max-Q | 85 | 6,144 | 870 | 1320 | 256 bit | 12 Gbps |
| RTX 3080 laptop GPU Max-Q | 90 | 6,144 | 930 | 1365 | 256 bit | 12 Gbps |
| RTX 3080 laptop GPU | 115 | 6,144 | 1.110 | 1,545 | 256 bit | 14 Gbps |
| RTX 3080 laptop GPU | 120 | 6,144 | 1,155 | 1,575 | 256 bit | 14 Gbps |
| RTX 3080 laptop GPU | 125 | 6,144 | 1,185 | 1,605 | 256 bit | 14 Gbps |
| RTX 3080 laptop GPU | 130 | 6,144 | 1,230 | 1,635 | 256 bit | 14 Gbps |
| RTX 3080 laptop GPU | 135 | 6,144 | 1,260 | 1,665 | 256 bit | 14 Gbps |
| RTX 3080 laptop GPU | 140 | 6,144 | 1,275 | 1,665 | 256 bit | 14 Gbps |
| RTX 3080 laptop GPU | 145 | 6,144 | 1,320 | 1,695 | 256 bit | 14 Gbps |
| RTX 3080 laptop GPU | 150 | 6,144 | 1,350 | 1,710 | 256 bit | 14 Gbps |
| RTX 3080 Founders Edition | 320 | 8,704 | 1,440 | 1,710 | 320 bits | 19 Gbps |
Other than the standard Max-Q and Max-P (standard laptop version), each of these models has various TGP ranges, meaning OEMs can set them to any of the specified figures, rather than the one mentioned on NVIDIA’s website. For example, the RTX 3080 Max-P can have a TGP from 115W to 150W. As you can guess, that will have a drastic effect on the resultant performance due to the varying sustained boost clock. We’re looking at a delta ranging from 20-30% between the lowest-end and top-end models.
The reason for the varying TGPs is the difference in thermal solutions used by OEMs, but that should be listed on NVIDIA’s official site. Otherwise, you might end up buying a laptop that’s a good 20-25% slower than the marketed performance.
