Third-party audit reveals AMD drivers are the most stable for gamers

Discussion in 'AMD Flavor' started by gerardfraser, Jul 16, 2018.

  1. gerardfraser

    gerardfraser Gawd

    Messages:
    625
    Joined:
    Feb 23, 2009
    Link to report
    https://www.amd.com/system/files/documents/graphics-driver-quality.pdf

    In May 2018, AMD commissioned QA Consultants to independently evaluate the stability of several of the latest graphics drivers from both AMD and NVIDIA. A pool of 6 AMD and 6 NVIDIA graphics accelerators were subjected to 12 days of 24-hour stress testing using CRASH from Microsoft’s Windows Hardware Lab Kit (HLK). Gaming and Workstation products from both graphics vendors were equally represented. The aggregate of AMD products passed 93% of scheduled tests whereas the aggregate of NVIDIA products passed 82% of scheduled tests (See Figure 1).

    Untitled.jpg
     
    Last edited: Jul 16, 2018
  2. Pieter3dnow

    Pieter3dnow [H]ardness Supreme

    Messages:
    5,593
    Joined:
    Jul 29, 2009
    If Nvidia paid them then Nvidia would have won :)
     
  3. defaultluser

    defaultluser I B Dum

    Messages:
    11,724
    Joined:
    Jan 14, 2006
    I'm going to re post myself from Front Page News,because this is just a Press Release with a bunch of interpreted numbers.

    This "thorough" test doesn't account for any hardware failures, which is the first thing I thought of when I saw that string of errors for the GTX 1060 and Quadro P600 and WX 3100. "The test compared 12 GPU's, six from AMD and six from NVIDIA, running on 12 identical machines."

    1 = insignificant sample size, especially if you want to isolate the drivers from the hardware here.
    IF AN ERROR HAPPENS CONSISTENTLY ACROSS MULTIPLE COPIES OF THE SAME GPUS, then you have a data point. But a single observation is just an observation (could be hardware, could be software). But running that many GPUs would cost money, and why would you want to spend that on a PR piece?

    They also don't tell you which programs were run, which also screams biased selection to make the results go one way or another.

    If you want to be taken seriously as a "study," you need a larger sample size than just twelve GPUs, and a completely identified "secret sauce" of test applications (so you can determine if these programs they run are even useful at all).

    If Nvidia's drivers are true shit then I welcome the evidence, and will probably buy AMD next. But this PR is not the way to do it.
     
    Last edited: Jul 17, 2018
    Chimpee and Araxie like this.
  4. Pieter3dnow

    Pieter3dnow [H]ardness Supreme

    Messages:
    5,593
    Joined:
    Jul 29, 2009
    You mean you missed the ones that damaged cards and were pulled fairly soon after?
     
    testicle likes this.
  5. defaultluser

    defaultluser I B Dum

    Messages:
    11,724
    Joined:
    Jan 14, 2006

    https://wccftech.com/nvidia-geforce-drivers-burning-issues-gpus/

    Its just the usual case of internet witchhunters overreacting and assuming the worst because they got a BSOD, and they just assume it broke the entire card and burned their house down. It makes for a better story to tell the guys :D

    Do you just assume that nobody on The Internets is a Compulsive Liar? The guys I grew up with would make up better shit than this "Nvidia burned my card".

    Yes, they were very crappy driver release, but no, there was no permanent damage. They havn't had a release that bad in two and a half years, so they've taken that crappy release to heart.

    But I will, definitely believe they're no better than AMD at this point. All I'm waiting for here is for this paid third-party to actually provide a real study, BECAUSE UP UNTIL WE HAVE REAL DATA IT'S ALL JUST OBSERVATIONS YOU'RE PROVIDING HERE.

    -- Multiple graphics cards to remove hardware issues from the equation (let's you toss outliers, and assume 95% of the good cards will have the same results)

    - A list of applications tested. It's pointless if you're just testing Jim Bob's Casual Steam Game with zero sales, but who the fuck knows what they''re testing?
     
    Last edited: Jul 17, 2018
  6. Pieter3dnow

    Pieter3dnow [H]ardness Supreme

    Messages:
    5,593
    Joined:
    Jul 29, 2009
    This video has a bunch of links I'll post some of them

    https://hardforum.com/threads/nvidia-196-75-drivers-can-kill-your-gpu.1500710/
    https://www.zdnet.com/article/warning-nvidia-196-75-drivers-can-kill-your-graphics-card/

    https://www.fudzilla.com/news/graph...rce-364-72-drivers-come-with-plenty-of-issues
    https://www.tomshardware.com/news/nvidia-driver-364-47-pulled,31370.html




    If the internet is filled with liars what does that say about you?
     
  7. defaultluser

    defaultluser I B Dum

    Messages:
    11,724
    Joined:
    Jan 14, 2006
    No, I just didn't think you'd have to go digging eight fucking years to make your point.

    I seriously thought you were bitching about the release from three years back, where the Internets freaked out and thought this was the same thing. It was not, just an unstable driver.

    Yes, Nvidia made a release that damaged cards. It's been awhile. You think you'd cut them some slack for not burning cards for eight years now?

    The fact that there was no Class Action for this really leads me to believe that the number of damaged cards were very small. People didn't hesitate to file a lawsuit when the real specs of the GTX 970 3.5GB were revealed, and that actually paid-out. If this had actually affected thousands, it would have gotten the same amount of press and much higher payout.
     
    Last edited: Jul 17, 2018
  8. Decibel

    Decibel 2[H]4U

    Messages:
    3,760
    Joined:
    Nov 10, 2000
    Pieter is Peiter. You should know by now that if he isn't being paid by AMD, he should be. We all have our biases; some are just more readily apparent.

    I thought he was talking about the 320.18s. I forgot that the internet claims that it has happened more than once.

    http://modcrash.com/nvidia-display-driver-damaging-gpus/
     
  9. sirmonkey1985

    sirmonkey1985 [H]ard|DCer of the Month - July 2010

    Messages:
    20,316
    Joined:
    Sep 13, 2008
    it did, back during the 8000 series when it was most commonly killing 8800GT's due to a bug in the fan controller on the drivers and the single slot cooler not being able to passively cool the card. just because some one might have their bias doesn't mean they're always wrong..

    as far as the 320.18 shit i don't know anything about that since i rarely update my drivers unless i have an issue with a specific game so i probably skipped those ones.
     
    Last edited: Jul 17, 2018
  10. defaultluser

    defaultluser I B Dum

    Messages:
    11,724
    Joined:
    Jan 14, 2006

    Even that has turned out to be more storytelling. Everyone in the Overclockers.net thread that actually had a detailed story to tell about their experience all restored their cards to proper working order with a complete driver uninstall, or in some cases a reinstall of Windows.

    This post summarizes the entire thread, and is on Page 62 of 64.

    https://www.overclock.net/forum/379...lay-driver-damaging-gpus-62.html#post20179262

    People were just annoyed that hey had to reinstall Windows, so they made shit up about the card being broken, or they just don't know how to troubleshoot a system that suddenly won't boot, and just assumed it was dead. Blame it on the dog?

    Yeah, for a few it broke their Windows install, but the card ain't broke.
     
    Last edited: Jul 17, 2018
  11. horrorshow

    horrorshow [H]ardness Supreme

    Messages:
    4,227
    Joined:
    Dec 14, 2007
    Boo freakin hoo....

    Back in the early 2000's, windows broke all. de. tiome.

    I literally used to keep a fresh windows install imaged (without any drivers installed yet), on a secondary drive.
     
    RAutrey likes this.
  12. Pieter3dnow

    Pieter3dnow [H]ardness Supreme

    Messages:
    5,593
    Joined:
    Jul 29, 2009
    I'm getting paid all of a sudden or need to get paid, did you see my first response?
    If it did not happen I can not quote it from other sources it is just that all we hear driver better this driver better that and so on and so forth.
    People been living in a Nvidia bubble for to long ...
    Nvidia pulled the drivers for a reason. Litigation in America needs to be a slam dunk before lawyers will burn themselves in court on Nvidia ...
     
    ltron likes this.
  13. spine

    spine 2[H]4U

    Messages:
    2,328
    Joined:
    Feb 4, 2003
  14. defaultluser

    defaultluser I B Dum

    Messages:
    11,724
    Joined:
    Jan 14, 2006
    Hey, I personally experienced the wonders of the Quack3 fiasco on my 8500.

    I had a lot better experience with the drivers for my 4850, until a later release broke Borderlands unless I turned off the performance optimizations (for all games).

    So if I wanted to play BF:BC2 at full performance, I had to ditch any desire to replay through Borderlands. Playing driver games like this was one of the reasons I went back to Nvidia, although I'm sure they've gotten better since then, right?

    Oh wait, that still didn't happen. Remember this shit marred the FIRST FUCKING YEAR of GCN's existence?

    https://techreport.com/review/24218/a-driver-update-to-reduce-radeon-frame-times

    It took AMD six months to acknowledge the problem seen in many reviews, and another six months to fix it across-the-board.

    So now you guys pinky-swear to us that AMD's driver issues are finally a thing of the past, now that they've been using the same slightly-tweaked GCN arch for six years? Took them long enough.
     
    Last edited: Jul 19, 2018
  15. Pieter3dnow

    Pieter3dnow [H]ardness Supreme

    Messages:
    5,593
    Joined:
    Jul 29, 2009
    You failed to click the link on the first post?
    You made the comment about me posting stuff from a good while back and that was not okay ?
     
  16. RAutrey

    RAutrey [H]ard|Gawd

    Messages:
    1,571
    Joined:
    Jul 25, 2002
    Driver issues with GPUs are cyclic. Both AMD and nvidia have had big problems in the past and both will in the future. If you look across the past 20+ years it's really pretty close to a 50/50 deal.
     
  17. Pieter3dnow

    Pieter3dnow [H]ardness Supreme

    Messages:
    5,593
    Joined:
    Jul 29, 2009
    That is rather obvious :) But what is not that every time when AMD cards/drivers are discussed the amount of Nvidia cheerleaders claiming AMD drivers suck without anything but their word for it is annoying as hell.
     
    Last edited: Jul 21, 2018
  18. Nasgul

    Nasgul [H]Lite

    Messages:
    110
    Joined:
    Jun 11, 2005
    So what you're saying is that is a made up thing. ATI/AMD driver issues don't "exist" because the AMD cheerleading squad says is a "made up thing by nVidia cheerleaders' "word".

    I should've listened to them then, but it's too late, Riva TNT2 late.

    And as mentioned, nVidia didn't pay this time, it was............?

    In May 2018, AMD commissioned QA Consultants to independently evaluate the stability of several of the latest graphics drivers from both AMD and NVIDIA.

    The Song Remains The Same - Led Zeppelin
     
    Last edited: Jul 22, 2018
  19. defaultluser

    defaultluser I B Dum

    Messages:
    11,724
    Joined:
    Jan 14, 2006

    That's because when Nvidia has serious performance issues or serious driver issue, it takes them from about a week to a month to fix. And they don't release new architectures with buggy drivers.

    When AMD has serious driver issues (performance or stability) It takes them six months to a year to fix it (Vega release driver stability, Tahiti release frame time consistency are just two most-recent examples).
     
  20. Pieter3dnow

    Pieter3dnow [H]ardness Supreme

    Messages:
    5,593
    Joined:
    Jul 29, 2009
    Read 2nd post .
    You quoted me agreeing with the person that said it is about 50/50 ? And yet you call me a cheerleader you just proven what I said is right about Nvidia cheer leading... You guys don't care about the issues or the discussion around it it is all about smearing and personal attacks.
    Because the truth is that both will have problems from time to time but one is not worse then the other. Except for those ones that actually screwed with clocks/fan but those were not from AMD. Why don't you find Nvidia dpc latency issues :) See if they fixed that already (yes they did according to them...).
     
  21. noko

    noko 2[H]4U

    Messages:
    3,754
    Joined:
    Apr 14, 2010
    I have both Nvidia cards and AMD cards, problems seems cyclic on the drivers. I do not think Nvidia has superior drivers, they have releases a number of shit drivers, the same with AMD. Lately AMD CFX is beating out SLI in games I play. For example FC5 AMD CFX works near perfect (once in a great while some flashing I notice but rare), Nvidia - was not smooth. Depends on the program, updates etc. Interface wise, Nvidia is in the dark ages while AMD is very slick with good features from monitoring, overclocking etc. Right now I would give AMD drivers the thumb up over Nvidia.
     
    Pieter3dnow likes this.