AMD vs NVIDIA Drivers....FIGHT!

I have said for awhile that Nvidia's drivers suck just as bad as AMD. Now since this was paid by AMD I am not too sure I believe it? But at the same time I can believe it.

Kyle, Maybe its time to a fair independent test yourself? What do you think?
 
Kyle, Maybe its time to a fair independent test yourself? What do you think?

I'm all for it but I can't imagine how Kyle would go about it.

Surely that level of testing (by [H] standards), would cause him to drift into madness....

header.jpg
 
I'm all for it but I can't imagine how Kyle would go about it.

Surely that level of testing, by [H] standards would cause him to drift into madness....

Yea I dunno, but he did do a freesync vs gsync comparison if I remember? But, I dont think people were happy with it.

Either way, drivers suck for both camps. just my 0.02c
 
Pardon my skepticism, but this is like expert witnesses in court cases: you can always find one that will testify in favor of your side.
If this study had shown NVidia winning, we never would have heard of it.

Not that I think NVidia drivers are better-- I have no recent experience with AMD drivers, and I haven't been gaming recently either.
They're shit, I hate my 1080ti from the stability of it's software, standpoint. Next card will not be Nvidia, I've had too many issues that result in complete system instability that actively prevents me from doing some hobbies I wish to take up.
 
Drivers is always the thing that amd get slammed for, and its really outdated thinking. This isn't the mid 2000's yet people seem to have (amd = bad drivers) embedded in their mind.

This is true, and a lot of it was especially pushed by the Linux crowd. I had a terrible time with drivers back in the day, and drivers for AMD on Linux were just utter crap for a long time. I stopped having major issues when the 6950s were out. AMD has made a lot of strides in that market, with Nvidia lagging more. Those were really good cards for me, I think I still have one sitting around somewhere. I switched back over to Nvidia mainly for performance reasons and what we were using at work (hard to beat Titan cards). Drivers for both sides have generally been decent for the most part lately. I do think AMD has done quite a bit to right the ship and start heading in the proper direction. I just wish they would try to compete in the high end gaming market again.
 
This isn't the mid 2000's yet people seem to have (amd = bad drivers) embedded in their mind.

For the record, I had a helluva lot more issues with my HD 5850 than I ever did with say, my 3850 or x1950 Pro.

I know my 9800 Pro was a pain in the ass.... What with using Omega drivers, Tray Tools, and Ultramon just to get my multi-monitor setup the way I liked it.

Plus in the 9800 days, this was an age where random sh*t happening was a common occurrence and formatting XP every 8-12 months was par for the course.
 
I'm just curious about the real-world validity of the software being used since that's out of my element. Being it's a MS thing they're testing with, I have my doubts and get the feeling it's about as representative as the gaming benchmarks suites are to actual gaming performance. Anyone have any experience/insight into Microsoft CRASH?

I did a 10 year stretch on AMD before switching back to NV about 9 months ago. IME they're pretty even right now. I came back to NV right when they were having those driver issues this past winter. Coincidentally I was having issues installing new AMD drivers right before I sold that card last fall (kept getting the old version number no matter what,) but I suspect they just had something wrong on their servers. I just rolled back but then sold the card so I never followed up on it.

On the NV side, I finally just updated to the latest a couple weeks ago (was still running a version from last fall yet before their problems started :cool:) and it's been rock solid stable. I had very intermittent driver crashes when playing video with the old drivers, but never during gaming. I also picked up a Gsync monitor shortly after the card and have been running it alongside my old 1080p monitor and haven't had any issues there, which honestly surprised me being different resolutions & refresh rates - I was anticipating problems that thankfully aren't there.

On the workstation side of things, I've always had better luck with Quadro cards. I think it's more dependent on the programs you're using though which is why I ask the question above. That said, I haven't used an AMD workstation card in quite few years at this point so I have no experience with their latest hardware & drivers on that end.
 
Nvidia doesn't provide working drivers at all for GTX 690 or Titan Z cards (last moderately stable ones were something like 38x, which Nvidia no longer offers), so it doesn't come as a huge shock that their driver team might be having some problems right now.
 
Where are the details of the errors/failures? Where is the details of the program/stress test that was running?

What version? What tests? This 'test report' is not really a valid report, its more of a summary. It is missing too much detail in regards to the software and configuration of said software.


Also I have seen crappy usb 3.0/3.1 drivers cause conflicts and BSOD's with non-gpu related drivers when usb devices flake out..

Where they using a USB KVM??



not enough details. feels like a PR stunt, not real scientific testing..


P.S. I have historically had more issues with AMD/ATI drivers than nvidia drivers, but agree that lately nvidia has fallen and AMD has gotten better about the drivers.


I read through the report, I would have liked to see the errors they were getting with some of these, especially when they had the hangups. I have been curious with the rate that Nvidia has been putting out "update" drivers specific for games if they were trying to get too specific with their drivers and potentially causing more issues with other applications. It seemed like there was just a lot of chance for compatibility issues with some of them.

Personally I have had a few issues, some of them stemming from drivers not working right for Titan cards. Overall, it hasn't been so bad though, nothing show stopping yet.

I'm just curious about the real-world validity of the software being used since that's out of my element. Being it's a MS thing they're testing with, I have my doubts and get the feeling it's about as representative as the gaming benchmarks suites are to actual gaming performance. Anyone have any experience/insight into Microsoft CRASH?

I did a 10 year stretch on AMD before switching back to NV about 9 months ago. IME they're pretty even right now. I came back to NV right when they were having those driver issues this past winter. Coincidentally I was having issues installing new AMD drivers right before I sold that card last fall (kept getting the old version number no matter what,) but I suspect they just had something wrong on their servers. I just rolled back but then sold the card so I never followed up on it.

On the NV side, I finally just updated to the latest a couple weeks ago (was still running a version from last fall yet before their problems started :cool:) and it's been rock solid stable. I had very intermittent driver crashes when playing video with the old drivers, but never during gaming. I also picked up a Gsync monitor shortly after the card and have been running it alongside my old 1080p monitor and haven't had any issues there, which honestly surprised me being different resolutions & refresh rates - I was anticipating problems that thankfully aren't there.

On the workstation side of things, I've always had better luck with Quadro cards. I think it's more dependent on the programs you're using though which is why I ask the question above. That said, I haven't used an AMD workstation card in quite few years at this point so I have no experience with their latest hardware & drivers on that end.
 
Yep. Ever hear of the i9 or 1080Ti?

Are you honestly going to counter with the Threadripper and Vega 64?



Never said anything about Vega 64. Its a dumpster fire. Vega 56 on the other hand is more compelling when the prices are at MSRP but does not even come close to a 1080ti. I was just pointing out you were flat out wrong about the CPU which this is off topic already anyways.

AMD drivers have been stellar this year. I dumped my 1070ti due to a dual screen hitching issue that wasn't fixed until after I sold the card. Its a shame because it was a good 1440p gaming card that sipped power.


EDIT ^^ Sorry Kyle, didn't see your note above until now.
 
That is really surprising to me. I've always had issues with AMD/ATI drivers. Of course, the last time I owned one was a Sapphire Radeon 4870X2. That one was far better than the one I owned before that, the Radeon 970. I tend to stay away from things for a while when I get bitten.
 
Well there's also the fact that AMD has been releasing the same (or very similar) gpu's for 7 years or something at this point.... Drivers should be pretty stable, lol.

The worst driver issues I had were with the nvidia 5900xt blue screening in unreal tournament, the radeon 8500 blue screening when changing resolutions in Halo PC, and the nvidia 8800 series constant blue screening in windows vista for the first few months. I haven't had those kind of problems in years with either vendor.

I was pretty ticked off about the 5900xt UT bug because nvidia wouldn't respond to support requests and I couldn't get anywhere with them, even though lots of other people had the problem too. I hope they've improved in that regard. With the Halo issue, ATI was super responsive, took all my feedback, reproductions, etc and fixed the issue ~catalyst 4.1. Then when I followed up and mentioned serious performance regressions I'd noticed (I had tested 12 versions or more of their drivers) ATI fixed those too in ~catalyst 4.3. It was funny because they really drummed up a press release about the huge performance improvements in that driver for their older cards, when in reality they were mostly restoring performance that had been lost. I think it was something like catalyst 3.0 or 3.1 where I noticed a particular scene was 50%-100% faster than the current driver.

I don't play much on day 1 though and don't really install beta drivers anymore either.
 
Nvidia has been going down hill for a while now when it came to drivers

The few times this year I updated my drivers for games I came to regret it

First was FarCry 5
Driver optimized for FC5 caused crashes, confirmed by Nvidia

Second was for Battletech, drivers didn't even install
Confirmed by Nvidia as well

And that's just the updates I personally cared about

I'll let other be the canaries from now on


Yeah, Nvidia used to be known for their excellent drivers drivers, but they have gone downhill.

It seems to all have started with the 372.54 which was buggy as hell. This also happened to be the first WDDM 2.1 release. I'm not quite sure why, but my take has been that drivers before 372.54 were generally pretty good, and drivers since then have been buggy as hell.

Not sure what changed. New driver coding team?
 
Actually the threadripper isn't near as powerful as the i9.

And the Vega64 doesn't touch the 1080Ti, let alone the Titan X.

Wrong and wrong. Enjoy your fanboy fantasies, I won't be responding again.
 
Actually the threadripper isn't near as powerful as the i9.

And the Vega64 doesn't touch the 1080Ti, let alone the Titan X.

Wrong and wrong. Enjoy your fanboy fantasies, I won't be responding again.

I love AMD and all, but the Vega 64 is more comparable to a standard 1080 in most titles than a TI.

There are a few narrow exceptions where Vulcan support and game optimization helps it close the gap a bit with the 1080ti, but in most titles it is more or less on the same level as a 1080.

As far as threadripper vs i9, I haven't done as much reading, but Kyle seems to like his threadripper.
 
Lol. you said competitive
I'd say that the 2700x competes pretty well with the 8700k.
2700x:
Cheaper
Competitive in gaming (doesn't beat the 8700k on average)
Soldered IHS.

8700k:
more expensive
higher clockspeed (therefore faster in games)
Takes a hit with the Meltdown patches.

probably equal if you factor in price/perf with an edge to the 2700x.

If gaming is all you ever do on your computer, than you'll be fine with a 8700k.
However, lots of people do more than just play games so, yeah a Threadripper would be better overall than your 8700k.
You could say the same thing about the Intel HEDT CPUs.
Gaming? 8700k > Intel HEDT anything.
Everything else? 8700k < HEDT AMD or Intel

Intels i9 stuff is a joke right now unless you delid or have a custom loop.
Intel's going to have a harder time when the 32 core Threadripper is out.
Sure the Intel 7980xe may be marginally faster than the 1950x but at what cost? you could buy 2 1950x's for the price of one 7980xe.

What I like about Ryzen is that I can get the 1700 or 2700 and overclock it to 1800x 2700x all core speeds.
Thats like making a 8600k as fast as a 8700k in everything, but sadly that can never happen due to no hyperthreading on the 8600k.

the word you were probably looking for was "faster in gaming"

Personally I'm on a budget and it was either a i5 8600k or a 1700.
Wasn't really that hard for me to pick since I knew that Intel was 100% guaranteed to make the socket obsolete by the time their next gen CPUs come out and I don't upgrade every year.
The 1700 was cheaper and faster than the 8600k in basically everything that I do except for Quake Live, but I don't need anymore FPS than 144 so.......................

Judging by your sig I'd say that you have the money upgrade to the fastest stuff whenever it comes out regardless of cost, so its Intel for you.
Pretty much everyone else looks at price/perf

Edit: kind of off topic, sorry.
back on topic, I like working with AMD's drivers vastly more than my Nvidia 850m laptop. Opening the panel or changing any setting the program just hangs for 10 sec everytime. AMD's stuff is fast and fluid and I use relive recording all the time. (No experience using Shadowplay)

I am in the process of building a $5000 gaming PC...could have bought anything I wanted, but still went with 2700x and Vega 64. Is it the fastest gaming PC I could have built with the money? No. Is it more than enough to max out my 75hz Freesync 1440p monitor? It sure is! By the time we have 144hz Freesync monitors, I will be ready for a new graphics card anyway.

More than anything else, I wanted STABILITY, UPGRADABILITY, and the ability to use Freesync both on my monitor and my upcoming Samsung QLED purchase. So even IF Nvidia had a card with TWICE the performance (it doesn't), it still wouldn't work well for the applications I have in mind.
 
They're shit, I hate my 1080ti from the stability of it's software, standpoint. Next card will not be Nvidia, I've had too many issues that result in complete system instability that actively prevents me from doing some hobbies I wish to take up.
Wow. What issues have you been having, and what hobbies are they that the driver is preventing you from doing?
 
Wow. What issues have you been having, and what hobbies are they that the driver is preventing you from doing?
Streaming, recording. Hell they even explode if I run some OGL applications and a d3d game at the same time, they're garbage, it's really frustrating.
 
I’m in for collecting a pool to retest with launch drivers of the Vega card vs. the launch drivers of the Pascal card.

Who’s with me?
 
AMD paid for this? Yea, this isn't a reliable test result.

no more reliable then the 100s-1000s game/software etc that Nv bias to skew results in their favor implicitly or implied.
a chunk of this time failiing miserably to hamper competitors benefit as well as their own consumers because they just can't "let AMD win" even if it benefits consumers on both sides of the proverbial fence.

this seems "about the same" but if they (the testers) were truly independent testing I put far more weight on their testing, 3rd party is 3rd party, commissioned or not, depends on how reliable their testing was and if their testing was not biased they just reported their finding as they discovered them to be.

that being said because it was commissioned by AMD if thier were "damning" information in the report AMD truly would be a "amazing company" to report it as it was given to them, because few companies/corporation/individuals want to show their faults ^.^

IMO AMD seems to WANT to know problems so they can address them in future product development, Intel and Nv want to know so they can spin stories or hide them under a shag carpet, Nv especially seems to go out of their way to find a "there is nothing to see here" BS act/tactics.

-----------------------------------
----------------

Shame there is no "longevity" testing of gpu/cpu like there is when it comes to HDD to allow consumers to maybe make the choice of steering clear of certain models or brands because they are much more likely to "die" an early death.

I have a feeling a "reference" design Radeon 8/10 would "live longer" than comparable Geforce either for the quality of the given product or overall stability long term when it came to product support (drivers etc)

maybe I am an ass or something (which is likely)
AMD seems to care far more about giving the best quality they can as much as possible (even if it costs them more to make happen)
Nv seems to care more about getting it on the shelf as cheaply as they can and maybe lasting warranty period nothing more.

the former is long term stability the latter is money now who gives a rats ass about long term, there are always unexpected cirumstances that can and do affect any electronic device, it is more about the "they knew exactly what they should have done and refused to do it anyways" that matters the most to me, as a consumer ^.^
 
Drivers is always the thing that amd get slammed for, and its really outdated thinking. This isn't the mid 2000's yet people seem to have (amd = bad drivers) embedded in their mind.

It doesn't need to be the mid 2000s. I've owned many brands of video card, and my current SOP is buy NVIDIA, buy the most card you can afford and do NOT bother with SLI, then if the drivers work leave them the fuck alone until I need to do something for which they don't work. REALLY be skeptical of driver updates once the marketing cycle for the next architecture hits.

Last AMD card I had was 2010/2011, but I've been there before. For me to jump off that bridge again, they have to bring some performance to the table, because they experience was extremely shitty.

With nvidia, sometimes stuff didn't behave well with a given game, and game support for SLI was iffy (usually due to how many man hours a developer was willing to commit to the technology for little return). However, if that happened, a little patience got me a new driver in short order. That driver was likely better than the last one overall. AT worse it was about the same. With every nvidia card I have purchased but the 5800, I've gotten to enjoy the product I paid for for damn near the entire time I owned it. The 5800 basically burnt itself up in the first 10 days of ownership (I was not alone in this) and I swapped it for team red at that time as nivida had thermal issues it wasn't going to escape in the near future.

Now AMD? Holy shit. Crossfire barely worked on any games. Those it "worked" on it worked like shit. Games support and driver support remained shit for the entire lifespan of the product I had. When something didn't work, I could wait for a driver. My hardware might or might not have been obsolete by the time one hit. It might or might not have fixed anything, and it might or might not have broken new stuff. Basically, from day one, one of my two crossfire cards was useless, and before I essentially decide to just throw them out, I had maybe two viable drivers for a single card setup that didn't have awful glitches in one or more game I liked. I might as well have flushed $1000 down the toilet. My previous foray onto team red was not nearly so bad, but lots of glitches and performance issues compared to nvidia.

Since Vista brought with it new handling of video drivers, I really haven't had too many crashing issues with either brand, both have their issues. But historically, team green has cost me less time not being able to use my stuff, gotten better performance in the games I play, and hasn't sat there promising me the drivers would come some day for essentially a products entire commercial offering.

Experiences like that are why you see a lot of people reflexively not trust AMD's driver support. Bad was beyond the pale and at points what I personally felt was fraudulent. They generated a fuck ton of ill will and bad blood. Their board partners at the time didn't help with that.
 
Doesn't surprise me. I love the GameWorks features that nVidia provides (when the devs implement them properly), but their drivers have given me some fits with rendering and clock speed bugs in the past.
 
Yeah, the Threadripper is more powerful than anything Intel offers, which is why they embarrassed themselves with that liquid cooled 28 core failbox. Just because it doesn't perform as well with OLD games doesn't make Intel better. Also, AMD uses solder, not toothpaste (TIM) on their chips, no need to de-lid.

And Vega64 is pretty damn powerful. Especially with games that are developed with it in mind. (it beats the 1080ti on the Battlefield V alpha) DX12 games generally run better on AMD. I will take (and took, as of last week!) a freesync monitor and Vega74 over a 1080ti any day of the week.

There is a Vega74??? Is that the new one coming out.

/s off
 
I respect the hardware AMD turns out. And though they have made improvements over the years, they still are not up to par with nVidia when it comes to drivers.
Agreed


I watched the video...

In other news, a recent study, funded by BP, indicates oil spills aren't as damaging to the environment as previously believed.



----------------------------


In the last year'ish, I've run Fury X in Crossfire, Vega 64 Liquid cooled, Vega 56 in crossfire, and a single 1080TI single and a pair in SLI in the same X99 board with a I7 6850K processor.

And I've gone back and forth between AMD and Nvidia before that too. My last cards have been Nvidia 670, AMD 285, AMD 380, Fury X, Fury X crossfire. Before the 670 cards I was nearly 15 years straight NVidia.

There is zero question in my mind that Nvidia has a seriously more dependable driver after my Vega experience.

I've had exactly one issue with my Nvidia 1080TI card since I got it. Kingdom Rush Orgins gets over 1000FPS and blinks in game.



My Fury X I sold last year about this time were also problem free, but I bought them about 1.5 years after launch and so the drivers had a chance to mature.



The Vega card drivers sucked fiercely at launch and the first three months after were trash -- so much so that when I sold them I wanted nothing to do with AMD video cards for the forseeable future.
My FS thread for my Vega cards (now long since sold) laid out my frustrations with AMD's Launch Vega. I tore off my AMD Loyalist supporter shirt right here. The emperor had no darn clothes when it came to launch Vega.
https://hardforum.com/threads/fs-three-rx-vegas-fair-prices.1946097/

  • Vega didn't support crossfire at launch which was a HUGE disappointment and was unannounced - just left up to buyers to find out on their own.
  • The Vega launch drivers were trash. Terrible frame rates, Freesync was unreliable between driver revisions on my three HP Omen 32 Freesync monitors (it must have cut on and off about 4 times between different driver release and kept breaking again once fixed. Sometimes it'd only work on one or two monitors, but not all three. Without Freesync after using it for the previous 9 months with the Fury X --- gameplay just felt...ughhhh.
  • Chrome browser paused up regularly, Facebook or any site with video thumbnails would lock up for seconds at a time when just browsing
  • I got red screens of death which were caused by the AMD Vega cards as best I could tell
  • JRIVER media player crashed constantly, or had jumpy/rubber band suspender type video playback with MadVR.
  • Eyefinity didn't work
  • Bezel correction with Eyefinity didn't work
  • Wattman settings were constantly resetting on their own (you had to use Wattman because MSI Afterburner didn't work)
  • PUBG would get 90FPS and then you'd get into a vehicle or into a fire fight and the game would slow down to 8FPS and I'd die.
  • Wolfenstein New Order (an AMD marketed title) would get 15-30 FPS in strange tunnel areas for no apparent reason.
  • Way too much hype an intentional mystery at Vega launch, Raja saying these things mine for 60, 70, 80,MHs (lie), We want to offer these AMD bundles to ensure gamers get their hands on our Vega (lie), We are holding back the launch to make sure we have enough to fulfill gamer demand (lie), We support AMD crossfire and plan to for the future (Marketing of RX580) (suddenly wasn't available at launch on Vega (lie), We want these cards to go to gamers and not miners (meanwhile releasing miner drivers that increase performance (lie). Sickening company PR spin, the worst I've seen on a product launch since the Viper 2000 (I also was suckered into). I don't trust AMD GPU Corporate at all anymore.


It was like I was beta testing a early engineering sample. The Vega 56 fans were louder than any GPU I've had in the last nearly 10 years. Just, truly a junk experience all around. It felt like I was pushed back in time to the early 2000's when the AMD products truly had bad drivers. (As mentioned, My Fury X driver experience was actually great, and I defended Fury X often on this forum --- but I bought those Fury X cards about a year after they launched - so the shifty Vega experience caught me wholly by surprise - and eliminated any good will I had towards AMD GPUs. Now I'll still recommend AMD CPU's with Ryzen and Threadripper, (though lets not pretend they aren't without their own problems. Don't believe me? Try to Play the newer game Carmageddon Max Damage on your Ryzen CPU? Or how about this?)

Never the less - I have personally used Ryzen without issue, and some of my friends very much like theirs - so I'll still recommend their CPUs --- but I won't recommend an AMD GPU for the forseeable future - they lost me as a customer and supporter for a spell.
-------------------------
 
Last edited:
"...we believe that AMD has the most stable graphics driver in the industry." (PDF - Conclusion)

I suppose Intel, with a near-equivalent market share (PassMark stats) to AMD, don't deserve consideration here. Or perhaps those numbers would be embarrassing for both parties involved? Surely there's more to "the industry" than red and green, or am I too naive? Would have loved to have seen them included.
 
"...we believe that AMD has the most stable graphics driver in the industry." (PDF - Conclusion)

I suppose Intel, with a near-equivalent market share (PassMark stats) to AMD, don't deserve consideration here. Or perhaps those numbers would be embarrassing for both parties involved? Surely there's more to "the industry" than red and green, or am I too naive? Would have loved to have seen them included.

What would they have included.... Larabee ?

I am pretty sure they where looking at discrete cards, which currently Intel doesn't make.

Yes integrated graphics would be its own thing and you would expect more stable. Would an Intel vs AMD integrated breakdown be interesting... sure however it wouldn't include Nvidia as they don't make any x86 integrated gpus.

I guess I'm saying right now there really isn't a 3 way race to speak of. There are 2, 2 way races featuring 3 players.
 
Back
Top