No announcement yet.

HD 3870 already outdated for OpenCL. Am I na´ve for expecting otherwise?

  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Originally posted by Veerappan View Post
    Yeah, you and Quaridarium are right that the double precision floating performance of the Fermi-based GTX cards is crippled, but at the same time, I haven't seen any consumer-level benchmarks that use fp64 either.

    Given that double-precision floating point wasn't even a requirement of the OpenCL 1.0 spec (it was an optional extension), I haven't placed much importance on it when weighing purchasing decisions.
    its not only openCL... openGL4.1 also have "double-precision floating point " for graphic if you really use openGL4.1 highdefinition graphic then the nvidia cards is crippled.
    most of the amd lowend and mid cards also cripple this feature.
    its only 5850 and up but 6870,67xx, 66xx, and so one not but 6970,6950 support that.

    this makes it easy for openCL or full(openGL4.1) if don't buy an 5850/5870 or 6970/6950 nvidia is better in OpenCL right now.
    vram matters to if you can get a 2gb version its just better than 1 gb in openCl..


    • #17

      The current AMD Stream SDK explicitly lists the 4xxx series as supported. The previous SDK listed 4870 as supported but 4890 as not formally supported. 3870 - well it is older than OpenCL, so YMMV.


      • #18
        The 4xxx is supported, but is missing an extension that my current work requires, so I had to replace it with something newer. My advisor had a spare GTX 480 laying around that I could borrow, so I'm using that for now.

        I'm hoping that the clover project makes some progress, because I'd love to eventually upgrade to a radeon 5xxx/6xxx on Gallium with working OpenCL. Of course this would also depend on a bit more optimisation work in the r600g driver.