5 Replies Latest reply on May 2, 2011 8:38 AM by himanshu.gautam

    Questions about OpenCL programming guide

      Exact limits from section 4

      In a number of places in the April 2011 version of the OpenCL programming guide there are references to "limited" memory, but I can't find anything saying exactly what those limits are.

      The runtime allocates a limited amount of pinned host memory that is accessible by the GPU without using the CPU cache coherency protocol.

      A limited portion of discrete GPU device memory is configured to be directly accessible by the CPU.

      There is a limit on the maximum size per [zero-copy] buffer, as well as on the total size of all buffers. This is platform-dependent, limited in size for each buffer, and also for the total size of all buffers of that type (a good working assumption is 64 MB for the per-buffer limit, and 128 MB for the total).

      I don't see anything about what those limits would depend on, or how to check what my system's limits are. 3 lists a "good working assumption" but it doesn't say why the limits would be any more or less than that.

      Also, I haven't run into any actual limits on allocation testing out quotes 1 and 2. That leads me to believe that either the limits are larger than an allocatable buffer or that the system is silently using a different type of memory when those limits are reached.