Gadgetory

- Is Nvidia slowing down the performance of their graphics cards with driver updates? I was recently notified of another YouTuber's video called Testing Games, who tested out a variety of different games side by side as well with different driver sets, 399.24 versus 416.16 in particular. And they found that there was a massive difference. When they updated the driver to 416.16, and it was a massive difference in a negative type of way, meaning they lost performance on their GTX 1060. So, today we've got a GTX 1060 right here on the test bench, a 6 gigabyte model, and we're gonna try and replicate the results of the testing gains replicator, in at least a few of the titles, and see what we can find. And if there is some funny buggers going on, then we'll definitely report about it, to you guys. (upbeat music) - Is brought to you by Squarespace Tech Yeah city viewers are getting 10% off. And this is great because Squarespace offer a complete package where you can just build, manage, and maintain a website all year round, without crazy overheads and crazy charges that other companies will charge you to get off the ground. If you're a small business, this service is very crucial 'cause you get a professional-looking website that looks the part and plays the part for only a fraction of the cost of what the other guys can do. Welcome back to Tech Yeah City. From time to time companies like Nvidia, Intel, and even AMD have done questionable things in the past that have raised concerns. And in regards to Nvidia directly, over the recent years, they have had the 3.5 gigabyte V-Ramp fiasco in the GTX 970, I reported on that. They also had telemetry baked into the G Force experience which I didn't agree with personally. These were thing that I didn't like, and I've raised them to my viewers. Of course, if there was a big issue like this within video Nvidia performance. There's a few things that come to mind, and the first thing is, this is actually highly illegal. They would not only get a class action lawsuit, the government would be on their back with a thing called plan of feigned obsolescence, which again, is illegal. So basically for Nvidia there is a lot at stake if they get caught, essentially nerfing the cards via drivers. But anyway, let's gets onto the video itself from Testing Games. He or she, they either do some very consistent testing. Side by side, I must admit, the work they do looks pretty impressive. However, in this case, we decided to install on the exact same setup, with the exact same version of Windows, and I do put an emphasis on that, the exact same version of Windows 'cause it's gonna lead into something. When we installed the same driver set that we're using. We found absolutely no difference. At first we started off with Assassins Creed Origins, where Testing Games had the inbuilt bench mark, they showed the settings. We used ultra preset settings, and we found we had 52 FPS versus 52 FPS. So this is on the same instance. We then went back on DDU, that's display drive uninstaller. Completely ridded the 399.24 driver, and then we installed the 416.16 driver and there was no difference. We then moved on to the Witcher 3. We noticed virtually no difference, however there is a inbuilt bench mark in this game. So we had to sort of replicate an apples to apples comparison here. And again, there was no disownable difference. Nothing worth noting on. Then we tried to do Kingdom Come Deliverance and there was, again no difference in this particular game. And then last we decided to move onto project Cars 2 and do the same test, And yet again, we notice no difference going from these drivers. Now some people will say, "Okay, but the drivers in videos going back and they fix the mistakes and then rerelease them to the public." That might be great, they might be doing that. I have no idea. But the drivers that we used here have been on our NAS since they've been released. So we're using the original drivers that were released day 1 from a Nvidia and so they have been stored on our NAS and they haven't been tampered with at all, unless a Nvidia's infiltrated my house and somehow put the new drivers on. And then manage to go back and modify the dates on them, and do some other crazy stuff. And some how made performance different. But point in case here, when we're going over Testing Games results, they're showing some massive differences. In fact, they're in the likes of around 10% in some of the titles that they are showing in Assassins Creed Origin, for example, they've got a big boost in FPS with the 399.24 driver versus the 416.16. And I was sitting here scratching my head like, Well there has to be an explanation for this. So I started digging a little deeper, and on Reddit there was a user LockinJP, and he put it down to an actual specter and meltdown patch that came in with a Windows update. And he tested before and after. And he noticed quite a big difference in games when he applied this update versus manually disabling it. So we decide to manually to disable this patch, and then retest the game. And Assassins Creed Origin then did get a boost from 52FPS on our end, up to 56. So this may be the smoking gun that testing games was trying to refer to, but the problem here is that in the title of the video they put in Nvidia slowest performance, with a driver update in Nvidia. And this is the problem, I think Nvidia has nothing to do with this. They have no control over Intell and Windows implementing updates of spectrum meltdown. And this looks like it is the exact reason for these performance slowdowns. It just happen to be maybe he tested the games in the 399.24 before the Windows 1809 update, or before he got the spectrum meltdown update patch applied. And then he went back and tested those games again, and he noticed a huge difference just off the top of his memory. And he is like, "Damn, I've got this one. This is it. Nvidia, you're gonna be exposed." And he didn't go back to double check his results, and then especially test them on the same scenario. And that's the problem. When you do bench marks, a lot of the time you have to test things as they are at the moment. I can't go back and pull up old bench marks from two years ago on a GTX 1060, and now compare them to an RTX 2070, for example. That would be stupid, cause over time a lot of things have changed. Not just in the likes of Windows and Intell micro updates, but also drivers themselves. Games actually do improve with driver updates. Take for instance like a game like Skum. When I was testing this with a GTX 1070 TI, when I was comparing that against the LTX 2080, I initially had the 2017 driver installed. But when I installed the 2018 driver that was the latest update driver, we noticed the boost over 10% in frames. So driver updates, generally over time help performance especially when it comes to certain games. Some games it can make an even bigger difference. But in this case, Nvidia hasn't done any wrong doing. And as much as people would like to jump on the band wagon with their pitch forks, and have it out for Nvidia. And I know they've done some stuff, very questionable in the past. This time around they're not at fault, I've never found any direct evidence that Nvidia have been nerfing their cars with driver updates. And again, it's an illegal practice if one were to do this kind of thing, that would simply stop support all together. And that seems very weird the moment where apparently a massive glot of ten series cars that were made and never got sold because the demand for crypto currency mining faded away. And so with that, they now have to sell these ten series cars. They would be nerfing the cars if they want to sell them. It doesn't make any sense at all. And on the flip side, I've heard rumor of Nvidia deliberately capping the RTX 2000 series cars so they can sell the ten series cars, and when stocks low on them, they can then up the performance of the RTX 2000 series cars. I mean ultimately who knows, you would have to get a driver for the RTX 2000 series car which is before the day one driver. Which I can't do. However, ultimately back to the case on hand when it comes to doing videos and driver updates, and nerfing performance. If there is any funny buggers going on, I'll always let you guys know as soon as I can. This time around, Nvidia has done nothing wrong. And I'd like to also send an open message to Testing Games. To please change his title around, and go back and reinstall Windows and check the results on those two drivers he has. With a same day install with all the patches applied, and then do another before and after. That would be a really good follow up video. Cause I just don't think his results are right. And also Steve from __?___ has said in the past, It's a reason why when we do these test. When you grab your cars and use CPU's, we always test at the same time. So there are no possibilities or at least the possibility sort of variances of like Windows specter meltdown updates happening are put to a minimum. Also if you want a link for a program that can disable and enable meltdown specter patches easily, then I'll put the link in the description below. I do recommend doing the Redge Edit method though that way you defiantly know that it is enabled and disabled, and you can test these things yourself. See if you notice any differences. Anyway guys, I hope you enjoyed today's video. If you did, be sure to hit that like button, and also big thanks to today's video sponsor, Squarespace. Just sponsoring this one out. And I'll catch you in another tech video very soon. Peace out for now. Bye. (upbeat music)