AMD Pulls Graphics Driver After 'Anti-Lag+' Triggers Counter-Strike 2 Bans - Sla...

›

• 1 hidden comment

• • and then the charge back happens if they don't giv

    and then the charge back happens if they don't give you an full refund + return shipping costs

  • Re:

    AMD is trying to improve gaming performance-- one interpretation is that Valve's code sucks (or it couldn't BE optimized), and their anti-cheat software is actually objecting to AMD improving their performance.

    • Re:

      Maybe the basic issue is the graphics API doesn't provide the timing feedback that would be necessary to implement this optimization.

• Re:

  Anything can trip anti-cheat if it tries hard enough. AMD's driver seems to be trying pretty hard so it's not surprising.

  This feature injects code into the game. A core functionality of anti-cheats is to try to determine if game code has been augmented or modified.

  • Re:

    It doesnt inject into the game, it hooks on system api, which is pretty normal, but also something some anti-cheat engines check for.

• No shit it triggered the anti cheat system. I think this falls into the realm of what the hell were they thinking. You don't modify game code you work with the game developer to add those updates to their application themselves.
  
  It's a cool idea but I don't see how this could ever work in the real world. Kind of frustrating because it means that it probably can't be applied to single player games. I guess AMD could leave it in there but then they have to worry about getting caught by the anti cheat software and getting there users banned.
  
  Also 10 milliseconds? Coming from the shmup scene that seems crazy high

  • You don't modify game code you work with the game developer to add those updates to their application themselves
    
    Hate to break it to you, but you are flat out wrong. A whole lot of "performance improvements" that seem to come with every minor video card driver update is doing exactly that... fixing broken code from the developer. Shit, even Windows itself does this for compatibility reasons

• In this case, it's not the App that is being modified, it's the library code for the graphics APIs (Direct3D, etc) being detoured.

  Other fun fact, the Steam Overlay itself is also an example of detouring graphics APIs. But it's one that game makers must allow and never flag as cheating.

  • Re:

    I think this post covers what you're talking about. https://steamcommunity.com/groups/SteamClientBeta/discussions/3/3817418065215221719/ [steamcommunity.com]

    For Valve's steam overlay to work in all games, even those not designed for it, they have to make alterations to the API that games use. You could argue that Valve shouldn't be doing these kinds of things and the counter argument is that it saves developers the trouble of implementing all the features the steam overlay provides.

• Re:

  Ring 0 code can do what it wants.

  While I believe Linus has done a lot of work on himself, so much so that I doubt he'd be hunting anyone down, he has a lot more control over kernel drivers and what they can do, being a gatekeeper on what gets into the mainline kernel and what doesn't. With that said, there would be nothing technically stopping a driver on Linux or anywhere else running on x86 from doing what this AMD stuff is doing.

  • Re:

    The majority of the driver is Ring 3. The only part that is Ring 0 is the miniport. That's why NVidia drivers can be updated without a reboot (AMD is too lazy to make their code properly re-entrant).

    It's not going to be the driver itself that is hooking the DLLs, it'll be some non-driver related process that is bundled with the driver, similar to the Geforce Experience stuff.

    • Re:

      Maybe so, but if you found something that you couldn't do that way, what would be easier than to write something into the Ring 0 component to allow you to bypass all limitations. My reply was to the statement "shouldn't be able to". Well, to stop this you'd have to rewrite how driver code executes on Windows. Maybe put it in Ring 1 and enforce some limitations, but watch how quickly stuff would break.

      • Re:

        Are you stupid?
        
        The part of the driver doing this is in Ring 3.
        
        The part of the driver in ring 0... must be there its literally just the miniport driver that directly talks with the hardware and that is all its allowed to do (implement the miniport API).

        • Re:

          Up through NT3.51 the graphics driver was separate from the rest of the system in Windows. They changed it in NT4 in order to pursue greater performance and the rest is history. Whether it was a wise decision or not is debatable, but it certainly has had negative implications for both stability and security. Most blue screens seem to be graphics driver implosions.

          • Re: Windows Driver model is broken

            True only for xddm drivers wddm drivers only have the mini port in kernel mode.... The rest is in user land.

• Re:

  You're saying what Windows drivers can do here Linux ones can't, but you also claim the main safeguard against that is that they wouldn't get merged into the kernel, which suggests you're not so sure. I suspect the difference in capabilities is not as big between Windows and Linux as you think. Windows has WHQL which provides a similar protection as long as you use WHQL approved drivers. That said I don't know if AMD's feature here would pass or not.

  • 1 hidden comment

• Re:

  >unauthorized game mods are utterly broken
  
  No. Anything I deliberately install is something my computer should allow me to install, whether it's interfering with a different program or not.

  • Re:

    Perhaps your computer should allow it, but if you are competing over a network, should the network allow it?

    That's what seems (to me) to be the argument. It's like requiring a standardized baseball in officially recognized competitions. What you do in a pick-up game doesn't matter.

• Re:

  The fact a driver can access any app in memory and MODIFY ITS CODE is absolutely broken. I'm fairly sure that if you tried to get a driver like that in the Linux kernel, Linus would personally hunt you down. I haven't found any way of doing what AMD is doing here on Windows in Linux or Mac.

  They wrote a virus, pretty much any security software should ring alarm bells when this kind of access happens.

  You do realize it's possible to do this in Linux right? In fact, debuggers modify application code all the tim

  • Re:

    Or perhaps games shouldn't have root access rights. This sounds like administrator access rights are being abused to gain an unfair advantage. Competitions should be between things that are "standardized" in particular ways, and what those ways are should depend on what the competition is about. If you're playing a network game, presumably you're competing about your skill as a player. If that's the case, this definitely sounds like cheating....even if someone else is doing it for you.

• Re:

  You left out the reason Valve is wrong my dude.

  Was it Valve was wrong because details of how VAC works that anyone developing driver software would need to know weren't publicly available?
  Was Valve's wrong philosophically in the design of VAC that demanded a kind of performance parity that these AMD drivers now will not adhere too?

  Also, you just going to not address the issues mentioned in the piece the drivers have with games that weren't developed by Valve? If you don't people might think unfairly of you

  • Re:

    Good point. My only modification is that after verifying that it was AMD that issued the software, not as a "cheat" but as an "optimization", and that AMD backed off when informed, that the bans of affected players be reversed. It's not their fault.

    While it's not quite "update to the latest drivers, get banned", in that per the article you had to go in and turn it on, a checkbox in your driver configuration that tells you that it'll "improve performance" isn't far off.

    The only thing would be Nvidia rollin

  • Re:

    If this decision stands, the next step would be exclusive titles that artificially prevented from running at all on NVIDIA or AMD hardware.

    
    
    To answer your question: I was very clear, they are wrong because players using OEM drivers for Commercial off-the-shelf hardware are in not cheating. If Valve has a problem with AMD, they need to take it up directly instead of fighting proxy wars by penalizing players that did nothing wrong.
    
    I do not own an AMD graphic card and have not played CS for decades.

    • Re:

      "I didn't cheat! It's this wallhack.exe I ran on my computer that's cheating!"

      AMD did retract the driver release, but now they do have to put the onus on gamers not to circulate it underground and use it.

      • Re:

        Why would gamers want to circulate it? It's not like it's a wallhack level cheat, and it's obvious to the anti-cheat software, so they'd get banned.

    • Re:

      Well... apparently you don't remember this [theregister.com].

• Re:

  wat...
  
  Valve isn't picking a fight here. AMD wrote a driver that does the same thing that cheat programs do (modifies running game code). Valve detects when modifications happen for cheat detection - it doesn't know or care if that modification is for legitimate or illicit purpose.

  • Re:

    Well, they can probably whitelist "approved" modifications, but that would require AMD communicating with Valve before releasing the drivers into the wild.

  • now what if say an windows OS update tripped the c

    now what if say an windows OS update tripped the cheat flag?
    Or say an DirectX update added speed up code that modifies running game code?

  • Re:

    "Modifies running game code" is a meaningless statement when applied to the hardware platform. Your CPU modifies running game code, one machine code instruction at a time, all the time even when idle. That why we represent code as a state machine when formally analyzing it.
    
    You need to look into what it does - is there modification of player inputs or their timing, is there modification of what player is allowed to see at that specific time?

    • Re:

      That is not how CPUs work. They are not modifying anything on their own.

  • Re: Valve is wrong

    Does Valve's anti-cheat voodoo simply use reflection to get the currently executing assembly and ban the user if the assembly isn't theirs?

• Optical mouse

  I remember long time ago when I was still playing CS a new type of optical mouse with a faster sensor came out. I don't remember technical details, but Logitech figured out how to get position updates from a sensor more frequently when all other hardware used fixed rate. I think it was original MX, but I might be wrong. That was a huge advantage over existing hardware, so everyone switched to using that mouse. Then competition came up with something comparable. Nobody thought of using a better mouse as chea

  • Re:

    Because the mouse doesn't change or intercept your input in any way. It just sent better inputs to begin with. That is not what is going on with AMD. They are intercepting code they don't have permission to intercept and changing it around. If your inputs are being captured and changed in anyway before being sent to the server, you are cheating - whether you realize it or not.

    AMD sucks ass and makes very stupid decisions constantly.

• Re:

  I don't understand what their basis is. It's like saying a faster cpu or 144hz monitor is cheating.

  • Re:

    Not at all even close to similar.

  • Read the summary!

    A faster CPU or 144hz monitor doesn't modify game code to achieve a faster refresh rate. I know this is/. but it would probably be helpful to at least read TFS so you can at least understand where the point of contention lies.

    • Re:

      It reduces latency, same as the reported driver hack.

    • Re:

      It is the same. Gamer A knows that if he buys a certain card, it reduces apparently latency. Why do they give a shit how it's done?

    • Re:

      This doesn't modify game code either, it intercepts system API call to align scheduling of drawing.

    • Re:

      Arguably those things do give the player an unfair advantage though. That's the problem with PC gaming, it's not a level playing field in many games.

      Other games like Street Fighter limit the frame rate to 60Hz for this reason. Even there though, there are input devices that are debatably cheating. You can play on keyboard, or you can get a "joystick" that has the joystick part replaced with buttons. Moves that were previously slow, like rotating the joystick 270 degrees, are suddenly fast because you just s

• Re:

  AMD is hooking into the games Win32 API calls and intercepting them. Of course, it's going to trip up anti-cheat because those systems are literally designed to prevent external programs from tampering with the game in that manner. That's why NVIDIA's solution, Reflex, actually requires the game developer themselves to implement support.

  • Re:

    yeah, that's what an OS does, it responds to API calls as good as possible, always trying to make the user experience better. The anticheat stuff is broken DRM shit by design, period

    • Re: Valve is wrong

      This is an online only game where the players expect cheating to be prevented and/or punished.

      Doing 100% of the cheat detection on the server is unrealistic.

      Don't like it, don't play it.

      • Re:

        "Doing 100% of the cheat detection on the server is unrealistic."

        And I call big fat bullshit on that.

        Case in point, Phantasy Star Online 2 aka PSO2, a japanese 3rd/1st view MMO ARPG.
        The game was available on PC/PS4/Vita/Switch and free to play.
        Except the anemic switch was incapable of playing the game.
        Thus, SEGA was running the game on servers and streaming it back to switch players. FOR FREE.
        If SEGA can compute and stream the whole freaking game FOR FREE PLAYERS, don't you tell me FPS game makers can't do

        • Re:

          You seem to be suggesting that the cheat detection is just a performance thing? It's not.

          There are some forms of cheating that can be detected at the server level, and they probably are. You fired more ammo than you've got? You defied gravity in ways that the game doesn't allow? You took fatal amounts of damage but never entered the code path for death on the client side? Your aim is absolutely 100% flawless? Those sorts of things they can probably detect at the server level.

          But there's a bunch of chea

• Re:

  Have you looked at the video card prices recently? It is a lot more than "few $100". Top of the line card today easily cost $1500+. Insane.

  • Re:

    I was think $400 card vs $1000 card or something but yes video cards are crazy expensive.

• Re:

  Based on Valve's response it suggests to me they did not try to work with Valve on this, otherwise Valve could have added an exception into VAC (or advised AMD to not enable the feature for CS2). My guess is AMD tested it briefly in CS2, and pushed it live before the VAC ban hit the test account.

  • Re:

    Yup, VAC bans don't trigger instantly, they come in waves to make it harder to know if a new hack was detectable or not. So this new "optimization" may have looked fine to AMD's developers while still being flagged on Valve's end and the ban sent out weeks later.

    • Re:

      Perhaps this is a really stupid question, but would those who are working on cheats be able to determine that anyways? It seems like something that could be deduced rather easily.

• Re:

  This isn't a Valve/Steam issue. It's an AMD is fucking stupid as fuck issue. Don't buy AMD and don't use their shit drivers, which they've never been good at, and it's not an issue at all.

  • Re:

    Right because Valve should get to decide what operating system stack I run on my system.

    DirectX is a API, and API is contract, as long the interface presented to their software does what is advertised, its NONE OF THEIR BUSINESS HOW!

    They want to make rules against aim bots and automation - fine with me but if they can't distinguish between aim-bot and shim to enable some efficiencies in a video driver, its their shit that is broken not the video driver.

    • Re:

      Which makes it clear that you have no understanding of how VAC works, nor how drivers work, nor why this driver specifically is an issue. This is AMD's fault for introducing tech they didn't verify nor work with the publishers before releasing. In order for AMD to release this to the wild, they first need to make sure Anti-Cheats don't get triggered. To make sure anti-cheats don't get triggered, AMD would need to give a heads up to the publishers who use anti-cheats to make sure the anti-cheat software is u

      • Re:

        nope, the anti cheat, just like any DRM is crap and needs to be scrapped or fixed to get on with the times. this shit is hindering progress, period

        • Re:

          Well, again, to get fixed, they need to know what needs fixing. And if AMD would have given them a heads up, perhaps they could have fixed it ahead of time. But AMD didn't because AMD is fucking stupid.

          I would love to scrap anti-cheats all together, but as long as there are highly competitive games there will also be cheaters.

          There are only three reasons to want anti-cheat gone:
          -You are a cheater
          -You got fucked over by anti-cheat when you were 100% innocent (not including this driver release)
          -The anti-cheat

          • Re:

            Fuck, I added a fourth reason at the last second and forgot to go back and change it.

          • Re:

            -Anti-cheat should be server side but publishers are fuckers and like to stick it to the players for maximum profits.

            • Re:

              That doesn't work, because all the input would look the same. Server side can work against wall hacks and such, but wouldn't be able to differentiate between human and completely automated input - aka a bot. You can rely on players to report suspicious bot-like activity, but that creates new issues. False reports that have to be analyzed. Slow response times to take affirmative actions.

              Unfortunately, it needs to be some combination of the two.

      • Re:

        I am not clueless at all you are just shilling for Valve for some reason.

        If value is relying on stupid things like checksums not changing, entry points at specific memory addresses, and detecting other side effects of called code vs actually doing behavioral analysis of inputs and game play - their system does not work.

        That is litterally all there is to it, you are defending values crappy kludge just because they labeled it anti-cheat, instead of the "horseshit-it-is-really-our-computer-not-yours-DRM-shovel

• Re:

  Since AMD driver support is generally better in a lot of ways than NVIDIA's driver support, and since its hardware functionality is fine except for what are still edge cases like AI and ray tracing, you'd have to provide a better argument for buying NVIDIA.