Originally posted by: MicahVillmow
Local arrays are supported in CAL only. |
Oh, I thought that was on the list for being added to 1.2. Is this going to be supported in the future at all? This is really limiting. For a lot of real applications, CAL is too time consuming and complex. For example, for one application that has 5+ large kernels you can imagine the difficulty in writing in CAL when I can port the same code in 1 DAY in CUDA to a Nvidia card with little to no problems but can't use Brook+ due to this and other limitations.
What about my first question? Is the multi-kernel output scatter bug fixed?