3 ms·

It's possible Nvidia manufactures the higher-end cards and then, if they fail the high-end testing but pass low-end testing, are marked as the lower-end card.

by tedchs 14y ago

It's possible Nvidia manufactures the higher-end cards and then, if they fail the high-end testing but pass low-end testing, are marked as the lower-end card. There is precedent for this in many areas of computer hardware (CPU, RAM, hard drives). I would be wary of undoing this and expecting any level of reliability.

hfsktr 14y ago
That is a good point. Something I really wouldn't have even thought of before even though I knew about the hard drives part. In the post he says: "the GTX 690 and the Quadro K5000 (same GPU) and, get this... the K5000 is only single GPU and clocked some 25-30% slower then the gaming card, what a joke" so wouldn't it be downgrading the card to make it usable on linux? maybe not idk much about linux, drivers, hardware etc.
- GhotiFish 14y ago
  no, he's only altering how the drivers perceive the card. Frankly, throwing in resistors to modify how it get's identified by the computer is really a hardware hack for a software problem. The drivers are stupid, and they're deliberately choosing to not use card features.
  - colechristensen 14y ago
    When you buy a high end card, you're buying the development time required to enable advanced features, you are not buying hardware; the silicon simply doesn't cost that much to produce. Most people do not want or need these advanced features, nor do they want to pay for them. The alternative environment has everyone paying $2000 for the same card and much fewer sales vs. premium users paying for premium features, and everyone else getting what they need at a much lower cost.
    - makomk 14y ago
      They're not just spending the development time on advanced features. NVidia appear to have intentionally crippled stuff like data transfer speeds to and from the GPU on their consumer models (but only for OpenGL/OpenCL and DirectCompute - you still get full speed when doing the exact same thing through DirectX) on the basis that it mostly only screws professional users.

by tedchs 14y ago

hfsktr 14y ago

That is a good point. Something I really wouldn't have even thought of before even though I knew about the hard drives part. In the post he says: "the GTX 690 and the Quadro K5000 (same GPU) and, get this... the K5000 is only single GPU and clocked some 25-30% slower then the gaming card, what a joke" so wouldn't it be downgrading the card to make it usable on linux? maybe not idk much about linux, drivers, hardware etc.

GhotiFish 14y ago
no, he's only altering how the drivers perceive the card. Frankly, throwing in resistors to modify how it get's identified by the computer is really a hardware hack for a software problem. The drivers are stupid, and they're deliberately choosing to not use card features.
- colechristensen 14y ago
  When you buy a high end card, you're buying the development time required to enable advanced features, you are not buying hardware; the silicon simply doesn't cost that much to produce. Most people do not want or need these advanced features, nor do they want to pay for them. The alternative environment has everyone paying $2000 for the same card and much fewer sales vs. premium users paying for premium features, and everyone else getting what they need at a much lower cost.
  - makomk 14y ago
    They're not just spending the development time on advanced features. NVidia appear to have intentionally crippled stuff like data transfer speeds to and from the GPU on their consumer models (but only for OpenGL/OpenCL and DirectCompute - you still get full speed when doing the exact same thing through DirectX) on the basis that it mostly only screws professional users.