After the last 344.75 NVIDIA driver update, I thought maybe we might not get any more updates until the New Year. Certainly I wasn't expecting to move from the R343 category of drivers to R346, but today NVIDIA has done just that. This is also one of the rare instances where NVIDIA has released a beta driver this year; the last official beta came back in June with 340.43, after which NVIDIA had six straight WHQL updates. You can find the drivers at the usual place.

I have to be clear, however: NVIDIA's driver numbers can often be something of a mystery, and this is a great example. One look at the full release notes (PDF) and I have to ask: why is this 347.09 instead of 344.80? NVIDIA might know, but I asked and they're basically not telling. The jump in numbering would usually suggest at least some new feature, but if it exists it isn't explicitly listed anywhere. More likely it's something that will come with a future update, but then why bump the number in advance?

I also like how this is part of the "Release 346" branch, but it comes with a 347 major revision (similar to how the Release 343 drivers started with 344 numbering). Of course, you can find 343.xx and 346.xx drivers for Linux, so that at least explains the main branch labeling somewhat.

The main reason for the driver release appears to be getting a Game Ready driver for Metal Gear Solid V: Ground Zeroes, which was released yesterday for PCs. This is also a Game Ready driver for Elite: Dangerous apparently, which might seem a bit odd as Elite: Dangerous was already listed back with the 344.65 update; then again, the game was in early access for Kickstarters before, where now it has officially launched.

Other than being Game Ready for those two titles, the only other changes mentioned are some 3D profile updates, a new profile for Project CARS (apparently for developers and testers, as that game isn't due for release for another three months), and a few miscellaneous bug fixes. We haven't had a chance to do any testing of the new drivers, but NVIDIA didn't mention performance changes so I wouldn't expect much.

I should also note that the AMD Omega Drivers came out almost two weeks back, and I have done some testing of those. We had planned for a launch day article but due to sickness that has not yet been completed. I can report that the Omega drivers appear to be an improvement in performance or at least status quo for all of the games I tested, and a few titles (BioShock Infinite in particular) show a rather large performance increase. We will hopefully have the full write up posted shortly, but if you haven't updated I have found no reason to hold off doing so.

Source: NVIDIA

Comments Locked


View All Comments

  • Phaedrux - Friday, December 19, 2014 - link

    These beta drivers enable 3D vision support for the Oculus Rift, which is a pretty huge deal. There is currently no head tracking, but it's a start. Inside the Rift the game is rendered as if it were on a giant wrap around movie screen in front of you. It's not as immersive as proper native Rift support, but it does add a lot of immersion to existing games. The effect is enabled whenever you turn on the 3D vision portion of the drivers and have the Rift set as the primary display. This has unfortunately also broken some games that had existing Rift support, such as Elite: Dangerous. Only solution is either to use Direct to Rift if possible, or roll back drivers.
  • mfenn - Friday, December 19, 2014 - link

    That's Metal Gear Solid V: Ground Zeroes, get in tune with your inner Kojima. ;)
  • JarredWalton - Friday, December 19, 2014 - link

    LOL... you're right. And the thing is, I saw that somewhere and thought someone was just not a native English speaker. Hmmm... Ground Zeroes. That sounds so wrong.
  • eanazag - Friday, December 19, 2014 - link

    OMEGA: um, clearly there is going to be testing on discrete GPUs. Will we see APU data also. I think it should look like A10, A8, A4. I'd like to see some bottom of the barrel APU numbers.

    I have been working on a friend's laptop; HP with an A4-3300M (dual core). The laptop is slow, but it has a regular HDD in it. 1GB of Windows updates took more than overnight. On the processors behalf, it was running at about 205 utilization. I take that as the hard drive being so slow that it isn't pushing the CPU. I was wondering what it would be like with an SSD in it.

    I typically work on laptops with i5's . I was curious how well this would game and I read on another site that the OMEGA drivers give a boost to APUs. For gaming on Intel I have a baseline of the HD4000 in Ivy Bridge; I found it tolerable. This is with StarCraft II, which taps both CPU and GPU pretty equally.
  • JarredWalton - Friday, December 19, 2014 - link

    I tested on desktop GPUs, as sadly I have no AMD APUs. If you don't include BioShock Infinite, most of the gains are marginal at best from what I've seen. There was a regression in performance from 14.9 to 14.11 Beta as well, and Omega at least mostly restores the lost performance (with Shadow of Mordor being one exception).
  • hfm - Friday, December 19, 2014 - link

    It's possible the revision bump is due to a code branch release and these betas have a feature disabled with a feature flag, but it's still there and built with the new revision of code.
  • JarredWalton - Friday, December 19, 2014 - link

    Yeah, someone (SH SOTN) just emailed me noting that OpenGL 4.5 is now in main driver branch... well, except this is a beta branch. Also, CUDA 7.0 might be in the drivers as well (again, from information from SH SOTN). My NVIDIA contact just told this was a "routine update (for now)". LOL -- so I know as much about the "for now" part as you.

    And if you're using HDMI, apparently the ability to go to full range colors is now fixed for RGB mode. I never noticed as I was on DisplayPort I think, though now I need to go test....
  • wurizen - Friday, December 19, 2014 - link

    this update didn't fix the displayport issue not detecting display upon cold boot or display getting fuzzy with my gtx 970. i am using DVI so it's fine. I am able to circumvent this issue. but, it's weird that nvidia has been mum about this. there is supposedly a hot fix (driver 344.80). But, haven't tried it, yet. My monitor, FYI, is an Asus PA238Q. A lot of people in various forums has been complaining about this. Seems like a defect in the hardware if Nvidia has been quiet. Usually means a big problem when the other party is all mum like. Like a mummy.
  • jwcalla - Friday, December 19, 2014 - link

    People have been saying that 344.80 fixes that, so I'm sure that change will get merged to the other branch soon.

    OpenGL 4.5 support was released in a driver on the day of its announcement.

    The 346 branch did make some changes to "dramatically improve OpenGL Framebuffer Object creation performance", so maybe that was a major code change that justified a new branch. Plus a lot more EGL work.
  • JarredWalton - Saturday, December 20, 2014 - link

    I have a GTX 970 setup that hasn't given me any issues with DisplayPort. Is the problem perhaps with the display (or the DP cable -- I *have* had issues there, though with AMD GPUs, not with NVIDIA), or maybe only certain cards are affected, like a 5% failure on a DisplayPort component or something? I do have to say that having only recently switched to using a DP connected monitor (after years running dual-link DVI with my 30" HP LP3065), I've encountered more oddities with DP in just a few months than I had in the previous seven or so years of running DL-DVI.

Log in

Don't have an account? Sign up now