QA Consultants concludes that AMD has the most stable graphics driver in the industry.

Rroff · 17 Jul 2018 at 16:50

AlphaOne said:
Two points:

1-Most crashes on NVIDIA side happened on Qaudro cards, same for AMD (on Pro cards). Geforce and Radeon cards had comparable results.
2-The QA used the wrong driver for Quadros on Windows 1803. That's why Quadros had more crashes.

So they manipulated the results in favor of AMD, which isn't unusual given they were paid by AMD.

Ryan Smith:

AMD... lies??!?

V F · 17 Jul 2018 at 16:50

Zeed said:
Same goes for NV panel I look click set done. Looks simple and practical to ME.
then Again I would go back to Windows XP in a second if it was running as good as this W10 insider build. But no DX12 is a killer especialy with wow going in to DX12

There isn't much to set anyway. The in game settings takes care of that.

Unless one is playing some really ancient game.

Gerard · 17 Jul 2018 at 16:51

AlphaOne said:
Two points:

1-Most crashes on NVIDIA side happened on Qaudro cards, same for AMD (on Pro cards). Geforce and Radeon cards had comparable results.
2-The QA used the wrong driver for Quadros on Windows 1803. That's why Quadros had more crashes.

So they manipulated the results in favor of AMD, which isn't unusual given they were paid by AMD.

Ryan Smith:

Just because a driver was out at the time they did the tests doesn't mean they needed to use it, for all we know the tests were virtually completed by the point that driver came out.

And who exactly is Ryan Smith? You quoted his name like everyone's meant to know him. :confused:

AlphaOne · 17 Jul 2018 at 16:53

Gerard said:
Just because a driver was out at the time they did the tests doesn't mean they needed to use it, for all we know the tests were virtually completed by the point that driver came out.

NVIDIA Quadro drivers for 1803 were out the day after 1803 was released. So they definitely omitted this one out.

4K8KW10 · 17 Jul 2018 at 16:54

AlphaOne said:
NVIDIA Quadro drivers for 1803 were out the day after 1803 was released. So they definitely omitted this one out.

It doesn't matter because realistically, most people don't update the drivers.
You should test all driver versions.

Gerard · 17 Jul 2018 at 16:55

AlphaOne said:
NVIDIA Quadro drivers for 1803 were out the day after 1803 was released. So they definitely omitted this one out.

Does it mention when the tests were done? And would these quadro drivers have made any real difference? Not as if people are chomping at the bit with pro applications for new drivers.

4K8KW10 · 17 Jul 2018 at 16:57

ubersonic said:
Actually it was because Nvidia used to gimp their image quality to boost FPS.

Not even joking, around 2004 they lowered all the settings in the drivers, the new very high was the same as the old high, the new high was the same as the old medium, etc. This resulted in the dawn of the 3rd party drivers era with the leader being the Omegadrivers which increased Nvidia IQ to match ATi's (at the expense of FPS obviously).

I know:

Futuremark confirms nVidia is cheating in benchmark - Geek.com

AlphaOne · 17 Jul 2018 at 16:58

4K8KW10 said:
It doesn't matter because realistically, most people don't update the drivers.

It DOES matter, testing a specific driver for a specific windows version is basic science.

Gerard said:
Does it mention when the tests were done? And would these quadro drivers have made any real difference? Not as if people are chomping at the bit with pro applications for new drivers.

Yes it would, Mircrosoft updated Windows to version 1803. Then NVIDIA released a driver specifically for 1803. Using a driver that is NOT for 1803 will cause more bugs and crashes. It's like using a Windows 10 Vanilla NVIDIA driver for Windows 10 creator update = total disaster.

deuse · 17 Jul 2018 at 16:58

4K8KW10 said:
I know:

Futuremark confirms nVidia is cheating in benchmark - Geek.com

2003

Don't feed him

AlphaOne · 17 Jul 2018 at 17:00

4K8KW10 said:
I know:

Futuremark confirms nVidia is cheating in benchmark - Geek.com

Those stuff are VERY old things, AMD has had similar doings.

Smffy · 17 Jul 2018 at 17:33

Classic GPU section thread right here, my favourite bit was the revelation that commissioned is a fancy way of saying paid for.

Zeed · 17 Jul 2018 at 17:34

Tbh cant remember when i had driver crash.
Guess AMD is trying to shake xeap srivers legend

ubersonic · 17 Jul 2018 at 18:06

4K8KW10 said:
I know:

Futuremark confirms nVidia is cheating in benchmark - Geek.com

That's something completely different and also wrong, it was the dawn of application specific performance optimizations in drivers. Futuremark retracted the claim of cheating when they realised what was going on (it hadn't been seen before so they thought Nvidia had just programed the driver to dump quality when it detected 3Dmark or something) and it was adopted by AMD too.

Rroff · 17 Jul 2018 at 18:10

ubersonic said:
That's something completely different and also wrong, it was the dawn of application specific performance optimizations in drivers. Futuremark retracted the claim of cheating when they realised what was going on (it hadn't been seen before so they thought Nvidia had just programed the driver to dump quality when it detected 3Dmark or something) and it was adopted by AMD too.

Hah - look at some of his other quotes then go back to the actual origin of the quote - and the author actually debunks the claim 4K8KW10 is making in a later paragraph which he has omitted...

Such as these bits for instance:

Since then, however, more myth than stone has entered the food chain. A poorly designed GF2 reference board is enough to make GF4Ti's signal quality poor. How, you ask, I do not know. Maybe it has something to do with the perihelion of Mars almost coinciding with aphelion of Earth causing the aether to muddle the space-time continuum.

...

And all this was perpetuated from ATI/Matrox fanboys teaming up against the NVDragon in their holy crusade and a few hardware sites of ill repute using monitors which had insufficient input bandwidth. In the case of the monitor having too low bandwidth, all your video card magic is to no avail. Anything will look like crap. This is why all signal quality tests should be done with high bandwidth (at least 700MHz) oscilloscopes - I believe c't did such a test when Parhelia was released. The results may surprise the casual mouth breathing dogmatic.

Which were part of what he quoted in one of his posts but changed the context from what he was trying to portray so omitted.

4K8KW10 · 17 Jul 2018 at 18:19

That quote is pure nonsense and meaningless, that's why I wouldn't even care to read it again.

Gerard · 17 Jul 2018 at 18:22

AlphaOne said:
Yes it would, Mircrosoft updated Windows to version 1803. Then NVIDIA released a driver specifically for 1803. Using a driver that is NOT for 1803 will cause more bugs and crashes. It's like using a Windows 10 Vanilla NVIDIA driver for Windows 10 creator update = total disaster.

I woulda kinda doubt its QUITE on that level.

Rroff · 17 Jul 2018 at 18:23

Gerard said:
I woulda kinda doubt its QUITE on that level.

1803 release has a lot of issues including breaking things in many drivers including the nVidia drivers - given the timeframes involved there isn't really a good reason to test 1803 but not the fixed drivers - either stick with the older update and those drivers or update everything.

Satchfanuk · 17 Jul 2018 at 18:48

It's funny to see the reactions on this forum to threads like this, funny and sad.

Even if Jensen himself said AMD drivers were better, people on here would still clutch at any straw within reach to claim otherwise

Rroff · 17 Jul 2018 at 18:55

Satchfanuk said:
It's funny to see the reactions on this forum to threads like this, funny and sad.

Even if Jensen himself said AMD drivers were better, people on here would still clutch at any straw within reach to claim otherwise

This topic isn't (shouldn't) really be about better except in the sense of stability which is what this assessment looks at. But it is not exactly clutching at straws to point out there are some various obvious flaws with their methodology - when for instance one card that uses the same architecture and drivers as another is crashing persistently and bringing the average down while its siblings are working fine then that is something that needs to be looked at in more depth as it is obviously more than just the stability of the drivers themselves at play.

Satchfanuk · 17 Jul 2018 at 19:48

Rroff said:
This topic isn't (shouldn't) really be about better except in the sense of stability which is what this assessment looks at. But it is not exactly clutching at straws to point out there are some various obvious flaws with their methodology - when for instance one card that uses the same architecture and drivers as another is crashing persistently and bringing the average down while its siblings are working fine then that is something that needs to be looked at in more depth as it is obviously more than just the stability of the drivers themselves at play.

But don't drivers recognise the card as opposed to the architecture and operate accordingly? Therefore, couldn't it still be the drivers at fault?

I couldn't care less anyway, I just think its pathetic that the NV defence force jump on anything negative and tries to turn it round, regardless of whether the story is true or not.

And as if that behaviour wasn't bad enough, they also jump on anything positive about AMD and try to turn it into a negative, which is part of the reason AMD feel the need to commission things like this in the first place.

Forums like this and the people on them have perpetuated a myth about the quality of AMD graphics drivers going back over a decade, which is why comments to this day surface about them being dog waste.

Nvidia and AMD have both had issues with drivers over the years. Neither are perfect. The complexity of drivers and supporting numerous generations of GPU within them will always create problems.

All anyone should be concerned about it that their own personal experience is fit for purpose and that any problems they have are resolved in the next driver, or at worst, the one after that.

Regardless of who commissioned the report, it shouldn't be dismissed purely on the basis of who paid for it, which is what has happened in here.

That would be like dismissing Cancer Research UK saying cancer is bad because they want your money. Ludicrous to say the least.

Anyway, if there are flaws in the report, point them out, that's fair enough, but don't claim they're deliberately misleading to favour AMD because they paid for the report. That's just some conspiracy theory, foil hat BS of the highest order.

My thoughts, take them or shove them where the sun don't shine, whatever floats your boat

PS. This wasn't aimed at you Rroff, mostly generalising about the thread, you were just a convenient quote