• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

**THE NVIDIA DRIVERS THREAD**

Link please somebody?

Look up :p

nVidia Vulkan Driver 364.91 Beta

Win 7/8/8.1 x32 :
https://developer.nvidia.com/win7win832bit

Win 7/8/8.1 x64 :
https://developer.nvidia.com/win7win864bit

Win 10 x32 :
https://developer.nvidia.com/win1032bit

Win 10 x64 :
https://developer.nvidia.com/win1064bit

https://developer.nvidia.com/vulkan-driver

DriverVer = 04/08/2016, 10.18.13.6491

Updated Vulkan API to 1.0.8
Improve pipeline creation performance and multi-threaded scaling
Increase our maximum bound descriptor sets from 4 to 8
Add support for asynchronous transfer queue
Improve VK_EXT_debug_report messages on incorrect API usage and shader compile failure
Fix VkImageFormatProperties maxMipLevels property
Reduce VkPhysicalDeviceLimits bufferImageGranularity requirement on GM20x GPUs
Improve Vulkan support on Optimus platforms
Fixes for minor driver and SPIR-V compiler bugs
 
One of the problems with Fermi (though it also affects first generation Kepler to a degree) the hardware is lacking certain functionality for running stuff in parallel and other instruction scheduling stuff that kills a lot of DX12 features and/or means emulating them without any performance gain from using them.

That isn't a problem with Fermi, it's an architecture from years before DX12 was considered. The problem is that at DX12 launch Nvidia stood up on stage proclaiming to the world that everything from Fermi upwards would support it. Rumours that the DX12 spec was lowered so that Fermi could be included and would be the base line level of DX12 support. Nvidia have spent the past what is it, 2+ years lying about Fermi's ability to run DX12 and since Windows 10 launch when Fermi DX12 drivers failed to materialise they didn't admit it couldn't, they lied, lied and lied again about delays and how support was coming shortly, then later, then end of the year, etc, etc.

Fermi doesn't support DX12, fine, Nvidia lying about it supporting DX12 is the issue. Let's just wait for some async drivers for the rest of the line up, because Nvidia has also delayed those but insists it supports it and it will come eventually.
 
That isn't a problem with Fermi, it's an architecture from years before DX12 was considered. The problem is that at DX12 launch Nvidia stood up on stage proclaiming to the world that everything from Fermi upwards would support it. Rumours that the DX12 spec was lowered so that Fermi could be included and would be the base line level of DX12 support. Nvidia have spent the past what is it, 2+ years lying about Fermi's ability to run DX12 and since Windows 10 launch when Fermi DX12 drivers failed to materialise they didn't admit it couldn't, they lied, lied and lied again about delays and how support was coming shortly, then later, then end of the year, etc, etc.

Fermi doesn't support DX12, fine, Nvidia lying about it supporting DX12 is the issue. Let's just wait for some async drivers for the rest of the line up, because Nvidia has also delayed those but insists it supports it and it will come eventually.

Getting proper DX12 functionality up and running on Fermi is a fair job - but not impossible - so I wouldn't rule out it coming eventually (though I wouldn't hold my breath) - I have to wonder why nVidia is dragging its heels over ASync though as there is nothing stopping 2nd generation Maxwell form making good use of it even though it needs more developer consideration to get the best out of it than it does on GCN. Not sure whether they are just doing it on purpose to mess with AMD or to try and drive sales of Pascal when it arrives :(
 
Nvidia have spent the past what is it, 2+ years lying about Fermi's ability to run DX12 and since Windows 10 launch when Fermi DX12 drivers failed to materialise they didn't admit it couldn't, they lied, lied and lied again about delays and how support was coming shortly, then later, then end of the year, etc, etc..

I though windows 10 had been out for 8 months ?, I better update my clocks to 2018!

It is sucky nvidia lied or just chose not to do it, but how many people relay is it going to affect the number of Windows 10 users running a Fermi GPU let alone one powerfull enought for it to matter (say a 480/580) is going to so small that nvidia probably though it not worth to cash to do it and then maintain the support.

At least with Vulkun this seems to be the case https://youtu.be/nGkpPp2tGSs?t=2784.

In short they (probably) can but wont
 
I though windows 10 had been out for 8 months ?, I better update my clocks to 2018!

It is sucky nvidia lied or just chose not to do it, but how many people relay is it going to affect the number of Windows 10 users running a Fermi GPU let alone one powerfull enought for it to matter (say a 480/580) is going to so small that nvidia probably though it not worth to cash to do it and then maintain the support.

At least with Vulkun this seems to be the case https://youtu.be/nGkpPp2tGSs?t=2784.

In short they (probably) can but wont

I got a GTX 580, which is capable to do 970-1000+ GPU clock and 3 gb vram. giving very decent performance overall.
For me no point to change it. I need the money somewhere else.
Another fact, Vulkan API even much easier implement and most likely to give better performance increase to Fermi than DX 12. Nvidia decided to cancel it......
Not to mention the official nvidia forums, where users are reporting plenty of dead cards since the 364.xx drivers.

It is "funny" , IT sites remain silents about this very serious "issues".
 
Last edited:
Not defending here

But not many companies support a 6 year old product. From software side.
You don't see it phones
You don't see it TVs
You don't see it in motherboards


Not saying its right at work we have MANY 450/550s for cuda work
 
Last edited:
Not defending here

But not many companies support a 6 year old product. From software side.
You don't see it phones
You don't see it TVs
You don't see it in motherboards


Not saying its right at work we have MANY 450/550s for cuda work

Personally im okay with that in most cases, my problem is when a company has promised something and then tries to make everyone forget that promise.. They should have kept their mouth shut from the beginning and since they didn't they should deliver on their promises.. goes for nvidia, amd or whatever company we could be talking about.
 
Personally im okay with that in most cases, my problem is when a company has promised something and then tries to make everyone forget that promise.. They should have kept their mouth shut from the beginning and since they didn't they should deliver on their promises.. goes for nvidia, amd or whatever company we could be talking about.

did Nvidia they actually say it in writing? or are people going off tech site said this? so must be true?
 
did Nvidia they actually say it in writing? or are people going off tech site said this? so must be true?

Yeah, I'd be very surprised if these promises were all official. Most old cards(ie, fermi based) are due to for upgrade anyway as they simply won't perform in latest games and why would any company invest more time and £ into adding these new features into old cards? Implementing DX12 features fully in any cards wont be done properly until at least the next gen IMO. If you want proper DX12 support buy a next gen card when they finally arrive.
 
Last edited:
did Nvidia they actually say it in writing? or are people going off tech site said this? so must be true?

Yeah, I'd be very surprised if these promises were all official.

Nv had a monumental PR stiffy@GDC with Forza running DX12 on TB:

NVIDIA will support the DX12 API on all the DX11-class GPUs it has shipped; these belong to the Fermi, Kepler and Maxwell architectural families.

https://blogs.nvidia.com/blog/2014/03/20/directx-12/

Which NVIDIA GPUs will support DX12?
NVIDIA will support the DX12 API on all DX11-class GPUs it has shipped; these belong to the Fermi, Kepler and Maxwell architectural families. At launch, DirectX 12 is supported by all Maxwell and Kepler GPUs. Fermi will receive DX12 support later this year (expected around the first wave of DX12 content).

http://nvidia.custhelp.com/app/answers/detail/a_id/3711/~/which-nvidia-gpus-will-support-dx12?

Nv Articles still live@Nv claiming Fermi DX12 support is coming:

https://blogs.nvidia.com/blog/2015/05/15/dx-12-game-ready-drivers/
 
did Nvidia they actually say it in writing? or are people going off tech site said this? so must be true?

Yes they have.. here is a link and a quote from said link. It doesn't get more official than nVidia's own website

https://blogs.nvidia.com/blog/2014/03/20/directx-12/
In addition, NVIDIA will match Microsoft OS support for DX12. Over 70% of gaming PCs are now DX11 based. NVIDIA will support the DX12 API on all the DX11-class GPUs it has shipped; these belong to the Fermi, Kepler and Maxwell architectural families.

EDIT: darn seems you beat me to the punch tommybhoy
 
Last edited:
So got a weird issue here, am going to report it to Nvidia unless you guys can think of why?

100% reproducible on 362.00

Working setup
GPUs 980s with SLI ON
Monitor 1 - Game X
Monitor 2 - Anything
Connections - Dual DP in top Card

so wanted to test 1 GPU on 1 Screen & run 2nd card for second screen as this used to give much better alt+tabbing results in games.

Not-Working
GPU 1 Connected to Monitor 1
GPU 2 Connected to Monitor 2
Connections as above
SLI connector unplugged!

Launch game on Primary Monitor
American truck sim - Both monitors flicker constantly UNLESS fullscreen = off

League of Legends - Both Monitors flicker constantly UNLESS fullscreen = off

WHY, this setup used to work on Windows 7,8 fine. Maybe W10 issue?
 
Back
Top Bottom