7 ms·

> It would be great for Nvidia to get real competition in this space. It seems obvious AMD will be a player, since they have similar hardware (GPUs) and with a

• AlphaSite 10y ago
  Their mixed GPU/CPU modules should be very interesting as well.
• frozenport 10y ago
  No, they have inferior performance, and all the same issues as Nvidia- in addition to requiring a port.
  • dharma1 10y ago
    inferior performance because of lack of performant DL libraries, like CuDNN, not because of the hardware itself
    • dogma1138 10y ago
      It's a mix of both, hardware is inferior for current use cases but current use cases are driven heavily by NVIDIA. NVIDIA knows how to do 2 things right; understanding that software matters, even more than hardware and deliver on that notion; and how manage developer relations. While AMD makes it easy to access information without NDA and sign ups NVIDIA blows it it if the water the moment you show even the slightest interest at signing up with them. NVIDIA at this point can probably spin off a pure consulting division for AI/ML and bring in more revenue than AMD in its entirety.
      • arcanus 10y ago
        I liked some of this point except the end, which is not correct. > NVIDIA at this point can probably spin off a pure consulting division for AI/ML and bring in more revenue than AMD in its entirety AMD had $3.99 Billion in total revenue in 2015. Nvidia had 5 Billion. Perhaps you mean profit? Amd was operating at a loss and nvidia was very profitable.
        • dogma1138 10y ago
          NVIDIA doesn't charge you for consulting services, they do expect you to buy pretty expensive hardware tho. They offer training and consulting services for "free" and paid services through their network of partners, to which NVIDIA provides services for free. If they spin off their in house knowledge and start charging for services and software instead of giving it out for "free", they might very well reach a 4bln $ a year revenue stream, that said a lot of people won't be happy about it...
      • dharma1 10y ago
        How is the hardware inferior for current use cases? The compute capabilities are pretty much the same for the latest NVidia/AMD cards in the same price range. It's a software thing - lack of optimised DL ops, particularly fast convolution kernels that is hurting the perf of AMD in deep learning
        • frozenport 10y ago
          No. AMD has worse floating point performance, and has poor support for half precision floating point values. Also gets beat on power consumption per flop.
          • dharma1 10y ago
            FLOPS/$ is better on AMD afaik. double speed fp16 is only on gp100 and I don't know if any framework supports fp16 training yet. But yes AMD need to add 2x fp16 too. The main reason people don't use AMD for deep learning is lack of fast libraries, and AMD's inability to provide a CuDNN equivalent optimised for their hardware
            • frozenport 10y ago
              But I don't think this is a software issue: clFFT runs faster on NVIDIA. Simple operations like SAXPY run faster on NVIDIA. Etc.
          • AlphaSite 10y ago
            Speaking of this, AMDs new Vega cards in theory should offer more fp16 than the P100.
• rasz_pl 10y ago
  No, AMD doesnt do software suppport. In words of their own developer: >>We don't happen to have the resources to pay someone else to do that for us.
  • spangry 10y ago
    Yeah, that whole ongoing saga is very unfortunate. Also very illustrative. Reading between the lines, my take is that AMD senior management are a bunch of generic MBA types optimising for higher profit in the present (Windows gaming market) by stealing from the future (HPC & server segments + growing niches like ML ---> Linux and open standards, in other words). It makes zero sense for them to place increasingly large bets on proprietary drivers when they're clearly not the market incumbent.
    • touristtam 10y ago
      It sounds what was happening when AMD, ATi were still seperate entities.