Legacy PC, interested into Legacy support, Nouveau is not that good.
Some months ago NVK gets developed and starts to release for some GPUs. Nothing much is actually said or known tho.
Some pages show that “the GT 710 is working with NVK”, but I can’t figure out the details because I am “not in the loop”.
Your GPU is a fermi-2 architecture.
A quick search shows that NVK supports the maxwell architecture in the 700 series of cards – as well as some of the kepler chipsets. Nowhere can I find that it supports the fermi chipsets.
Here's a more detailed breakdown:
Maxwell (GMXXX): Some GTX 700 and 800 series, and most 900 series cards.
Pascal (GP10X): GTX 1000 series.
Volta (GV10X): TITAN V.
Turing (TU1xx): RTX 2000 series and GTX 16xx series.
Ada Lovelace (NV190/ADXXX): RTX 3000 and 4000 series.
Kepler (GKXXX): GeForce 600 and 700 series (in progress).
Blackwell (NV20/GB10X): RTX 5000 series (in progress).
In the linked message I showed the complete ~$ inxi -Fzxx output.
Inside it there was this:
The GK208B is noted as a “Kepler 2.0 chip”.
This said, and also to not re-link what I’ve already linked (a spanish page said “all kepler ones supported” and a social.treehouse.systems page showed the GT 710 running The Talos Principle) could it just be that I’ve an obscure version of the GT 710 which was not “listed in the supported hardware”?
Also, remember, I already knew about both those pages, which are months old, and newer pages (which I’ve linked) suggest that at least SOME sort of support is present.
That same inxi output shows the arch as Fermi-2
Are you saying that inxi lies and what you find on line supercedes whatever inxi may report.
This is such a confusing soup… [1]
The NVIDIA GK208B is a Kepler chip, as “GK” should mean.
Wikipedia here[4] said that “Fermi has been used for the 700 and 800 serie only for the mobile chips” (which may or may not be true, since it seems to contradict the other two pages).
Google’s AI spews out this when looking for “why does inxi Fzxx show the gt 710 as fermi?”, which may or may not be an hallucination (all the links used as “source” don’t say anything of what the AI said here) but at least SOUNDS plausible.
AI Overview
The NVIDIA GeForce GT 710, despite being based on the Kepler architecture (specifically the GK208B variant), is sometimes incorrectly identified as Fermi by inxi -Fzxx due to its relation to the older Fermi architecture. The GT 710 is a low-end card designed for basic tasks and media playback, and it shares some architectural similarities with the Fermi-based cards, leading to this misidentification in some system information tools.
Here's a more detailed explanation:
GT 710 Architecture:
The GT 710 is built on the 28nm process and uses the GK208B GPU, which is part of the Kepler family.
Fermi Connection:
While technically Kepler, the GT 710 has some design similarities with Fermi cards, particularly in its lower-end specifications and performance characteristics. This can cause confusion for system information tools that might not have the most up-to-date or specific architecture identification data.
inxi's Behavior:
The -Fzxx option in inxi is designed to provide detailed system information, but it may not always perfectly differentiate between similar architectures, especially when it comes to older or lower-end cards. It's likely that inxi is grouping the GT 710 with Fermi due to its architectural lineage and performance profile, even though it's technically a Kepler card.
Focus on Functionality:
For most users, the specific architecture (Fermi or Kepler) isn't as important as the fact that the GT 710 is a low-power, entry-level card suitable for basic tasks like web browsing, video playback, and light 2D applications.
inxi -Fzxx just shows the wrong info for some GT 710 (desktop) models lumping them all as arch: Fermi-2. I don’t know nor care how “1” and “2” differ, the point is that one is Fermi and the other is Kepler.inxi -Fzxx everything points out that the GPU’s chip is Kepler, except for the arch x entry, so I guess that that part is just wrong[5].I do not appreciate the abrasive tone when and where I am just and only talking facts and data. You are NOT Duke Baron inxi -Fzxx, so why are you taking this so personally?? inxi -Fzxx can just be wrong. ↩︎
I may be wrong on this, but there’s NOTHING to be ashamed about, and it’s not an important mistake (to guess wrong) on my part, here. ↩︎
Collabra is the lead developer. Dated 4 August, 2025: https://www.collabora.com/news-and-blog/news-and-events/mesa-25.2-brings-new-hardware-support-for-nouveau-users.html:
Starting with Mesa 25.2, NVK will now advertise support for Blackwell (RTX 50xx series) and Kepler (most GT and GTX 600 series, most GTX 700 series, and some GTX 800 series) GPUs. On Blackwell, NVK supports Vulkan 1.4 while Kepler is stuck at Vulkan 1.2 due to hardware limitations. With this, NVK should support the same set of desktop GPUs as NVIDIA’s proprietary driver ever has.
Fedora is using Mesa 25.1 and 25.3 is a release candidate, so we can home that 25.2 may soon appear.
That is very good to hear, thank you.
That said, check out my previous message above yours.
It comes out that some GPUs have been manufactured first with Fermi and then with Kepler, or at least that’s the next most probable event after the “people just wrote it down wrong”.
Some GT 710s (since we are talking about these) have GF chips, while others GK.
Mine is a GK, but gets written down as Fermi.