Fermi support?

I’m confused.

The current drivers support starts at GTX400 (Fermi and newer) and they have recently dropped the support for old GPUS, so if they don’t support Vulkan for some actual supported GPUS… There will be a different series of drivers for each one or just a more complex driver which will act differently with Fermi GPUS?
I think that will be more problematic for than just simply finish the work which is almost done. And if there aren’t technical problems. What’s the problem?

In relation with AMD. AMDGPU is far away to be ready, so changing to an AMD GPU is not an option, at least not now. We will see what happend in the future. I want more options so, AMD Hurry up! :p

But, like you, my Fermi is still working very well and it’s too son for changing my hardware.
Vulkan is just starting, I have time to think in other options and OpenGL is still be there, so I won’t buy a new GPU even if a won’t have support.

I am just kinda disappointed about Nvidia communications. I asked several times Nvidia rep at Nvidia main forum about the DX 12 and Vulkan support progress and no information has been given. Actually , Nvidia rep just fucking ignoring any request about this matter from anyone. All i think is, first: they have no idea at all what is going on. Second: They just don’t give a fk. Sorry for language but just getting feed up with this bull*t.

I have to agree. It would be nice to at least know whether or not certain Fermi GPUs will be supported in the future. As a developer, I’m spending most of my time working on a GeForce GT 635M, which still does a good job for a mobile GPU.
Practically, I’m forced to get a new notebook just because I don’t know whether or not the GPU will support Vulkan. And that is very disappointing, Nvidia.

I don’t accept the fact they said that Fermi was get support and well in the release of the drivers is something like: “Well we change our mind and you don’t care now, and well, we just get to that conclusion now.”
You don’t get your word back in this kind of matter like that.

To get the internet happy everyone was getting support but in the release date they simple ditch us. They talked lots of time after August about Vulkan but they simply hide that fact from us.
In my opinion the Fermi exclusion is important to at least say something about.

Disappointed and really angry because the Vulkan support was the decider fact in the deal of the used card that I buyed and if I buyed an used card was because I cannot afford an new in this time of my life.

In the end three cards in my house that got unsupported: “GTX 580 3Gb”, GTX 560 Ti 2Gb" and a “GTX 460 Top 1Gb”. Sad really sad.

So how about if we all keep “buzzing” the Nvidia rep at the official driver forum? Can’t just ignore us forever?
So far i was alone but if they see more users , maybe something gonna happen. Who knows?

Totally agree. Kick up some dust and make a fuss.

DX12 has been waiting for a while for Fermi support and still nothing.

The best thing is their Vulkan driver FAQ:


You’d think they maybe take the opportunity to clear up on the expectations they have raised and failed to deliver? You’d think that a frequently asked question might be “Where the hell is Fermi support?”

No! A frequently asked question is whether or not Vulkan support will be in the mainline driver in the future or if people will have to use and old branch that has Vulkan support added to it for all eternity. Sure! That’s what people will ask.

Why did they promise DirectX12 support for all DirectX11 cards back in early 2014:
Just to figure out that it would be better if these customers hand over some cash for a new card instead of providing DirectX12 support. And then 2 years later back paddle on the same promise again for Vulkan.

And NVIDIA doesn’t even bother to keep all their marketing material up to date, because their geforce.com website still says that Fermi supports DirectX12.

I think siriq is right that inside NVIDIA most people just don’t know what is going on, so nobody dares to step forward and give a real answer. Instead we get a somewhat official short answer somewhere in some part of the QA section of a webinar. Really great communications by NVIDIA.

Poor Cakefish. Little did he know…

Whatever has been said, I’m rather expecting them to actually support Fermi, even if they’re not saying a word about it.

At least this is what I want to believe when seeing this part of the FAQ :

  1. Will there be an official (non-beta) NVIDIA driver with Vulkan support?

Yes. […] These will be available to all users to run Vulkan content as part of our regular driver releases.

But actually, if your GPU doesn’t support Vulkan, you can still take a look at your CPU.
Currently I can get vulkaninfo running with the Intel ICD, which I got from Mesa’s vulkan branch. Since I just care about being able to test my applications, I’m fine with it.

If the NVIDIA devs has been able to provide an OpenGL 4.5 driver for Fermi, they are largely able to provide Vulkan.
Thus I’m pretty sure that, at the time being, it’s just a matter of priorities instead of an absolute yes/no. So instead of trolling NVIDIA on their own forums, I’d just trust them and see what happens.

I bet you 20 Dollars that there will no Fermi support in the first mainline driver that ships Vulkan. I’d love to be proven wrong. I have a Sandybridge i7, which has an iGPU that isn’t supported by the Anvil driver. Replacing it would be even more stupid, because at 4.7 Ghz it still hammers newer Intel CPUs at stock speeds and I’d have to throw away 16 GB of perfectly fine DDR3 RAM and a mainboard. Also Neil Trevett said that being able isn’t the issue so that’s not really reassuring that they will eventually provide a driver. If it’s not a yes or no decision why don’t they just say so? We already trusted them and saw what happened.

Obviously not few of us generating almost 3000 views about this question here. I wasn’t talking about trolling. All i ask, most of us or i would say all of us should set up and ask Nvidia about this matter. In the meanwhile , also ask about the DX 12 support as well. I have AMD FX CPU, so no other options for me to try Vulkan but with the promised Fermi Vulkan driver.

Wow now that you talk about it this topic has more views that most of the topics in other categories (DirectX and OpenGL for example), so please Nvidia don’t use that “install base” explanation.

We are not just a few and we realy have interest in get Vulkan support as you can see for the views count. In my opinion apart from been told that Fermi was get supported, we are proving that we deserve it.

I am not saying that we must get it in this beta drivers, i understand that the development of Vulkan has been a racing against time and understand if they said that Fermi support still not completed but we get it later, but let us down and ditch us this way is very sad. Please Nvidia don’t let us down again…
(WDDM 2.0 is here but DirectX 12 that was for the end of 2015 still nothing)

It’d be wise to hook up with the DX12 folks and explain to nvidia that we expect them to keep their word on supporting Fermi.

Maybe that means working the issue through some of the more popular tech journalists out there.

Same here. I have a Thinkpad T430s here, and it’s still in the warranty period (3 years). And I expect that this notebook will work the next 2-3 years. I wouldn’t bother if there would be a possibility to replace the GPU unit, but that’s not possible on a notebook, I would have to replace the whole notebook.

I agree. I also want to encourage the people who now generated 3300 Views to take the short time to register on the forum and speak up.

Agreed is not so difficult and is the only way to get the Nvidia attention.

I don’t have much knowledge about the ways of working of the Nvidia forum but we and DirectX12 guys have this forum, nvidia facebook, nvidia and vulkan twitter so we need to get their attention.

Okay, I have registered to voice my opinion. I have been watching this forum since before this discussion was created (it started with someone specifically asking about the 400/500 series support).

Like you guys, I read fermi would be supported in official nvidia documentation so I was satisfied with the cards I have (450 and 560ti) until the new generation of cards come out.

I wanted to develop in vulkan with my Fermi’s and try to utilize cross video card performance as vulkan has been advertised. I don’t need super fast cards for developing and I won’t buy anything new that is out right now as vr (htc vive) is imminently being released and will pretty much need the next generation of cards. I do plan on temporarily picking up a used 970 to satisfy my vr hunger by April, but again…I still want my fermi’s supported to vulkan up multiple cards. (Try having a card render main image and having other cards render mirror’s, etc… try different methods of mixing them together)

Just like other people, my laptops are also based on fermi.

When I go to make something in Vulkan, with the numbers from steam stats, this will reduce the amount of users that could run my program. And this is never a good thing. Especially from everything I can read, the Fermi cards can do it with “relatively” minimal effort.

As I will be buying the top of the line card in the new generation coming out this year, this Fermi support will influence my decision as my card isn’t even 5 years old yet. I would be more annoyed if I had bought the 590 (which I am thinking of the equivalent in the new generation).

I am disliking the silence from nvidia on this matter, especially in their own forums. I can only hope they come around and surprise us with support. As I do not run windows 10, and have little care for developing in DX12 I didn’t even know our cards couldn’t do it; I was blind and faithfully accepting what the nvidia website told me in it’s product line up.

On the slide 55 of this presentation they crossed Fermi out and on the following they say its not planned :’(

They used Unity stats. Seriously?


If you count up the 400 and 500 series cards plus GT 600 series cards that are actually rebrands. Fermi accounts for 9.72% of ALL users. This is in line with Unity stats that are stated on the slide (less than 10%). Now let us do the math. The last number of active Steam users was put at 125 Million. So NVIDIA decided to give 12.150.000 users the middle finger. And that’s only Steam not everyone who has a PC is active on Steam. Relative to NVIDIA only cards the share is 18%. So NVIDIA doesn’t want to support around 1/5th of their customers. Totally not worth it to support. Let’s rather make drivers for the 5 people on the planet that use Solaris. I was even nice enough not to include the 1.53% of GT 630 cards many of those are actually Fermi, too. An unknown percentage of Fermi cards is also part of the 12.78% other category, because individually they did not have enough of a share to be listed on their own. Sorry NVIDIA your customers aren’t morons you can’t fool us this easily. Given how small Vulkan drivers are supposed to be, because there is no heuristics needed anymore and that they coded the deviceids and features and limits for Fermi into the Vulkan driver this is really the cherry on top of the cake.

EDIT: I get a lot of requests over Google Drive for this spreadsheet. I have made it read-only so anyone can check that the numbers aren’t faked. If I forgot about a card please say it in this thread. You can check Wikipedia if the 600 series cards is actually Kepler (GK) or Fermi (GF).

What a company…

I can’t even say how disappointed I am with them. Ditching so many users because of reason’s! In the DX12 matter they even put it in their website!

If that attitude of them is to prevail, when I change my card (yes it’s going to take a while) is going to be for an AMD, just need good Linux drivers and bye bye liar’s!

I still don’t believe in that low, really low move by them! We really need to try calling them to the reason, we need to continuing insist because they simply can’t be so bad and so low.

Omg, wish I had seen this before posting. I take back what I said, that’s very disappointing.
Heck, even “10% of the install base” is huge. Following this reasoning, they shouldn’t even provide Linux drivers since, after all, it’s “only” 1% of the gamers install base, Huh ?
Come on NV. My GPU is barely 3-years old.

Leaves to party at Intel’s