In regards to Fermi, any update on whether Fermi will get Vulkan support? Your launch article said Vulkan was Kepler and up, but nVidia previously said at Siggraph they would support Vulkan on Fermi.
That's really funny, because they still have Fermi SKU's in production for OEM's to this day. Maybe they should have thought about canning Fermi production before discontinuing support for it. I just don't get why they are tapping out anything but Maxwell at this point...
Vulkan support past the last 2-3 generations doesn't make much sense anyway (and that stands for AMD, too). By the time we get titles actually built for Vulkan, it will be 2017/18 and by that time, said hardware just won't have the necessary horsepower.
Maybe some professional software could use Vulkan even on Fermi, but I don't know much about that.
Nvidia is way too silent about DX 12 and Fermi! Vulkan api can be easy implement also to Fermi. The excuse what they found is, "low" base to do it. Still at least 10,000-20,000 Fermi owners out there. One more time at least. I bet Nvidia doesnt want to give free performance increase also. I personally measured 50% boost on draw calls. That was achieved with the early driver. WDDM 2.0 also gave me massive 45% boost.
I'd rather they not divert their software engineers to work on the small install base of too cheap to upgrade consumers. I want them focus'd firmly on the people like me who upgrade regularly! The consumer that evangelizes their brand and gobbles up their products with the thickest profit margins.
They've also been working on Async compute for over 7 months now and haven't released it. If they started when the first Ashes benchmarking happened. Given they cannot finish the drivers they claim they can (I doubt Kepler and Maxwell can do Async compute) I see no reason for Fermi drivers to be started.
Their overall Dx12 efforts seem pretty sad compared to AMD given how much faster they typically have done things in DX11 and how Nvidia has far more resources than AMD.
Um yeah, stay away from this driver. I have 2 monitors, and now i'm getting TDR driver crashes while just browsing the web. Might be related to youtube/video acceleration.
"Despite the quick turnaround this is not a rushed released."
I'm not so sure about that. A lot of people have severe problems installing this, including me. I can't get my head around the fact that this even passed WHQL?! Is this WHQL merely a label you have to pay for these days without any further testing involved?
It's a dual monitor bug that only happens when installed though GFE. Nvidia has since pulled the driver from GFE, and has fixed the problem, and submitted the fixed driver to microsoft for WHQL Certification.. We are now just waiting for microsoft to certify.
If you don't want to wait, Nvidia has posted a Beta driver on Nvidia.com, that is the exact same driver waiting for certification.
That's what WHQL has been for a decade. All it means is MS did some basic testing for memory leaks and some other possible OS issues. The only difference WHQL makes to nvidia is if they paid for the testing or not. For video drivers always ignore whql and just get the latests beta. There is no stability differnece on average
I didn't have a chance to look at 364.47, but 364.51 that replaces it is actually marked as beta...
Either way, it's been posted to death that WHQL is merely a verification that the driver conforms to some APIs, not a verification of functionality (remember how many WHQLed drivers used to wreak havoc on VIsta?).
Question for Daniel Williams or anyone else in Anandtech that can answer this.
How can a company get a WHQL stamp for drivers that have so obvious problems? And this is not the first time Nvidia needs to come up quickly with a hotfix driver after a problematic WHQL one.
Could you please write an article about the WHQL stamp? To investigate if that certification does warranty a better more stable driver today or not.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
22 Comments
Back to Article
ltcommanderdata - Tuesday, March 8, 2016 - link
http://anandtech.com/show/10035/vulkan-10-releasedIn regards to Fermi, any update on whether Fermi will get Vulkan support? Your launch article said Vulkan was Kepler and up, but nVidia previously said at Siggraph they would support Vulkan on Fermi.
Page 50
http://on-demand.gputechconf.com/siggraph/2015/pre...
Mahigan - Tuesday, March 8, 2016 - link
NVIDIA have announced that Fermi will not be getting Vulkan support.Ryan Smith - Tuesday, March 8, 2016 - link
Correct. Vulkan will only be for Kepler and later.Samus - Wednesday, March 9, 2016 - link
That's really funny, because they still have Fermi SKU's in production for OEM's to this day. Maybe they should have thought about canning Fermi production before discontinuing support for it. I just don't get why they are tapping out anything but Maxwell at this point...Michael Bay - Thursday, March 10, 2016 - link
Because everything past Maxwell is legacy to them, no need to put resources there. Maxwell itself will be, come middle or end of the year.Alexvrb - Saturday, March 12, 2016 - link
Wow and people criticized AMD for reducing driver support "too soon" for older pre-GCN architectures.bug77 - Tuesday, March 8, 2016 - link
Vulkan support past the last 2-3 generations doesn't make much sense anyway (and that stands for AMD, too). By the time we get titles actually built for Vulkan, it will be 2017/18 and by that time, said hardware just won't have the necessary horsepower.Maybe some professional software could use Vulkan even on Fermi, but I don't know much about that.
siriq - Tuesday, March 8, 2016 - link
Nvidia is way too silent about DX 12 and Fermi! Vulkan api can be easy implement also to Fermi. The excuse what they found is, "low" base to do it. Still at least 10,000-20,000 Fermi owners out there. One more time at least. I bet Nvidia doesnt want to give free performance increase also. I personally measured 50% boost on draw calls. That was achieved with the early driver. WDDM 2.0 also gave me massive 45% boost.Sttm - Tuesday, March 8, 2016 - link
I'd rather they not divert their software engineers to work on the small install base of too cheap to upgrade consumers. I want them focus'd firmly on the people like me who upgrade regularly! The consumer that evangelizes their brand and gobbles up their products with the thickest profit margins.testbug00 - Wednesday, March 9, 2016 - link
They've also been working on Async compute for over 7 months now and haven't released it. If they started when the first Ashes benchmarking happened. Given they cannot finish the drivers they claim they can (I doubt Kepler and Maxwell can do Async compute) I see no reason for Fermi drivers to be started.Their overall Dx12 efforts seem pretty sad compared to AMD given how much faster they typically have done things in DX11 and how Nvidia has far more resources than AMD.
archon810 - Tuesday, March 8, 2016 - link
I'd recommend either staying away or getting your safe mode ready for this build https://www.reddit.com/r/pcmasterrace/comments/49d...The best strategy seems to be a full uninstall first, then a clean reinstall.
I was afflicted myself https://plus.google.com/+ArtemRussakovskii/posts/E...
DabuXian - Tuesday, March 8, 2016 - link
#artemsluckGigaplex - Tuesday, March 8, 2016 - link
My brother was affected too, and it refused to enter safe mode.hechacker1 - Tuesday, March 8, 2016 - link
Um yeah, stay away from this driver. I have 2 monitors, and now i'm getting TDR driver crashes while just browsing the web. Might be related to youtube/video acceleration.Breit - Tuesday, March 8, 2016 - link
"Despite the quick turnaround this is not a rushed released."I'm not so sure about that. A lot of people have severe problems installing this, including me. I can't get my head around the fact that this even passed WHQL?! Is this WHQL merely a label you have to pay for these days without any further testing involved?
bikal adhikari - Tuesday, March 8, 2016 - link
It seems to be exactly that! I am happy with my 353 drivers. Most stable I have found yet.jasonelmore - Tuesday, March 8, 2016 - link
It's a dual monitor bug that only happens when installed though GFE. Nvidia has since pulled the driver from GFE, and has fixed the problem, and submitted the fixed driver to microsoft for WHQL Certification.. We are now just waiting for microsoft to certify.If you don't want to wait, Nvidia has posted a Beta driver on Nvidia.com, that is the exact same driver waiting for certification.
Khenglish - Tuesday, March 8, 2016 - link
That's what WHQL has been for a decade. All it means is MS did some basic testing for memory leaks and some other possible OS issues. The only difference WHQL makes to nvidia is if they paid for the testing or not. For video drivers always ignore whql and just get the latests beta. There is no stability differnece on averagebug77 - Wednesday, March 9, 2016 - link
I didn't have a chance to look at 364.47, but 364.51 that replaces it is actually marked as beta...Either way, it's been posted to death that WHQL is merely a verification that the driver conforms to some APIs, not a verification of functionality (remember how many WHQLed drivers used to wreak havoc on VIsta?).
testbug00 - Wednesday, March 9, 2016 - link
WHQL means of is it compatibly with The Windows versions it says it is and Microsoft backs that claim up.It means absolutely nothing for stability.
ZeDestructor - Tuesday, March 8, 2016 - link
Installed flawlessly for me.. though as usual SLI got disabled and clocks were being set wrong until a reboot....yannigr2 - Wednesday, March 9, 2016 - link
Question for Daniel Williams or anyone else in Anandtech that can answer this.How can a company get a WHQL stamp for drivers that have so obvious problems? And this is not the first time Nvidia needs to come up quickly with a hotfix driver after a problematic WHQL one.
Could you please write an article about the WHQL stamp? To investigate if that certification does warranty a better more stable driver today or not.