The Difference Between NVIDIA SLI And AMD Crossfire

Running two graphics cards in your computer can sometimes be a good way to boost your graphics power, but there are a few critical differences between NVIDIA and AMD when it comes to dual graphics. Linus from Techquickie breaks it down.

If you’re thinking about running two cards, you have one more factor to think about when deciding between NVIDIA and AMD, since they both handle it differently. I highly recommend watching the video to get the full gist, but it boils down to:

  • NVIDIA cards must be identical GPUs, while AMD cards are Crossfire-compatible with other cards in the same family and with different RAM configurations
  • Crossfire is available on more motherboards (and on cheaper ones)
  • Newer AMD cards do not require a connector — which isn’t a huge deal, but it looks cleaner and is one less link in the chain in which something can go wrong
  • AMD cards can crossfire a GPU with an onboard APU

In the end, NVIDIA’s SLI is a more constricting but consistent experience, while AMD is more flexible.

All that said, we still think single-GPU solutions are better than dual-GPU solutions, except perhaps in special situations (like multi-monitor gaming, for example). It’s up to you, though, and if you want to go with two GPUs, this video is a great primer on the difference between NVIDIA and AMD’s implementation.

AMD Crossfire vs NVIDIA SLI As Fast As Possible [Techquickie]


  • Interesting but missing some important information.

    1: Many users perceive an intolerable micro-stuttering effect when 2 GPUs are linked. If you are such a user, you pretty much have to go to 3 GPUs.

    2: CUDA (Nvidia) is still ahead of OpenGL (AMD), though the gap is closing.

    3: Nvidia drivers are superior to AMD’s and always has been.

    Now to fend off the fanboi accusations;

    I just built three new machines, two extremely high-end. All use AMD GPUs.

    In my estimation Nvidia provided better-value high-end gaming cards than AMD up until about 2004, at which time AMD passed Nvidia. For the past decade Nvidia have kept their market share by clocking inferior architecture high enough to post competitive specs, which they achieve by clocking high and using powerful coolers. Unfortunately the cooling solutions never quite seem to be beefy enough, and the board and coolers degrade quickly, truncating the card’s lifetime.

    This is a shame, because I genuinely prefer Nvidia’s support system.

    • I’m sorry but I think I need to say something here.

      Firstly, micro-stuttering is caused by frame-latency. Adding more GPU’s is not a fix for it, and it often makes the problem even worse. SLI, historically, has never had any major issues with micro-stutter, however AMD’s Crossfire technology has been rife with it. That being said, AMD’s recent driver updates have all but eliminated the problem on AMD cards, bringing them on par with Nvidia’s SLI.

      Secondly, CUDA and OpenGL are completely different things. They aren’t even remotely similar in what they are or what they do.

      CUDA is a type of unified architecture found onboard Nvidia GPU’s that can be used for purposes other than graphics, such as powering their PhysX technology and performing complex calculations in non-gaming related situations.

      OpenGL is a graphics API, similar to DirectX, though usually more efficient. OpenGL is not limited to AMD cards, Nvidia can and do use it just as well as AMD do. In fact Nvidia’s OpenGL driver support is actually better than AMD’s. This is why Nvidia cards are far better to use on Linux based operating systems (OpenGL is open source and thus is used widely on Linux).

      AMD and Nvidia both primarily use DirectX, however they both also work with OpenGL. Mantle only works on AMD’s, and is a rival API to OpenGL and DirectX, but it’s currently only adopted by a limited amount of games (primarily games on the Frostbite 3 and CryEngine engines).

      To summarize, CUDA is a type of hardware, OpenGL is a piece of software, a graphics API. Completely unrelated.

      As for the drivers, I agree. In the past Nvidia’s drivers have always been superior to AMD, though AMD did basically fire their entire driver team and recently their driver releases have been more frequent and have solved a lot of problems with Crossfire.

      Nvidia’s architecture is actually superior to AMD’s for the most part. The equivalent Nvidia card will almost always outperform it’s AMD counterparts, and they usually run cooler than AMD’s too, so I’m not really sure where you got that info from. EVGA are easily the best aftermarket brand of them all, their cards are extremely high quality, extremely well cooled, come with a lifetime warranty (7 years), and they last forever. I know a guy still using his original GTX 285’s from January 2009.

  • I used SLI many, many years ago, was an absolute pain in the ass. All too often it caused more problem than it solved. At the time I was thinking it was a fairly “cheap” upgrade for my PC, I could pick up an older model card to match my existing card and SLI it to get a performance boost for little cash. It gave a little bit of a performance boost, but if I had of just paid $100 more and got a new comparable tier card, it would have doubled it instead. Not worth the money in my mind.

Show more comments

Log in to comment on this story!