Naming scheme

• Professional cards include cards under their NVS, Quadro, Quadro RTX, GRID, and Tesla lineups.
• Consumer cards include their GeForce and Titan lineups.

Common misconceptions

There are multiple GPUs with the same name but part of a different architecture (therefore different support status length), and there are other cases to be careful of:

Desktop:

• GT730: Has a GF108 (Fermi) and a GK208 (Kepler) variant.
• GT625: Has a GF108 (Fermi) and a GK107/GK208 (Kepler) variant.
• GT640: Has a GF116 (Fermi) and a GK107/GK208 (Kepler) variant.
• GT645/620 is a Fermi card, even though all other 600 series are Kepler variants (with the exception of the above)
• GT705 has a GF119 (Fermi) and a GK208 (Kepler) variant.
• GTX745, GTX750 and GTX750 Ti are Maxwell, even though all other 700 series are Kepler variant (with the exception of the above)
• GT1030: Has a GK107 (Kepler) and a GP108 (Pascal) variant.
• All-in-one desktops are known to feature Mobile GPUs, which would make them fall under a different (and often shorter) support cycle.

Laptop:

• GT810M/GT820M are Fermi cards.
• GT825M/GT870M/GT880M are Kepler cards.
• GT920M is a Kepler card.
• Most cards in the 800M series have multiple variant with varying architectures (A card in this series can be Fermi, Kepler or Maxwell).

Identifying your GPU

Due to this confusing naming scheme, one should not look at just the model name when seeing their support status, but instead their architecture.

Windows

1. Download and run GPU-Z.
2. First identify if you have a consumer or a professional card. See the “Naming scheme” section above.
3. Next identify the card architecture. This will be the GPU textbox. You can cross-reference this with the support table at the top of this page.

Linux

1. Install the lshw package from your distribution’s repositories.
2. Run the command sudo lshw -C display, your GPU code is the product column.
3. Next identify the card architecture. This will be the GPU textbox. You can cross-reference this with the support table at the top of this page.

This GPU Code follows a similar pattern for most cards, for example we have GA102:

• G: This means generation
• A: This means it belongs to the Ampere generation.

Most GPU codes follow this same pattern, with the exceptions of TUxxx which means Turing architecture, and AD1xx which means Ada Lovelace architecture.

Driver Support Exceptions

• GKxxx (“Kepler”) Desktop GPUs are supported on Windows and Linux via the R470 legacy driver series till September 2024.
• Consumer GF1xx (“Fermi”) GPUs are supported on Linux via the R390 legacy driver series till the end of 2022.
• Not all Professional Fermi (GF1xx) GPUs are still supported on Windows, see the official GPU support list for specific models. On Linux there support is for all Fermi GPUs.