PDA

View Full Version : Suggestions for next release



  1. build sampler direcly on buffer object
  2. Two small problems in Appendix D.3
  3. Semantics system
  4. Not having functions return values via pointer arguments
  5. Non-power of 2 vector types
  6. Whises for OpenCL 1.1 and more!
  7. Asynchronous Events
  8. querying kernel execution status
  9. malloc/free and executing sub kernels
  10. support for none case sensitive languages
  11. Provide typedefs for OpenCL functions
  12. Automatic conversion on big-endian hardware
  13. OpenCL 1.0.48 typo
  14. Private Memory
  15. "clDevicePointer" function needed!
  16. Specification 1.0 to 1.1 diff
  17. OpenCL overview, vec3 missing
  18. Open Intermediate Language
  19. Function to get human readable error messages
  20. Error in online manpages
  21. Possibility to access every cl_mem object as buffer NEEDED!!
  22. Share OpenCL buffers between processes
  23. Clarify CL_MEM_USE_HOST_PTR
  24. Recursion
  25. OpenCL C++ Bindings and premature program deallocation
  26. Describing kernel parameters
  27. Bring this DirectCompute features to OpenCL!
  28. Add support for #warning directive
  29. Make it easier to get the CL version number
  30. Ability to change border color with image objects
  31. Kernel's source code protection
  32. Support for cubemaps
  33. Support for bit-fields
  34. Make 3D image writes part of the core specification
  35. Support for fenv.h as in C99
  36. 3rd-party SDKs : a chaos
  37. Arrays of images
  38. Mipmap with Trilinear interpolation
  39. Reduce, Map
  40. ULP accuracy profiles
  41. work-group to work-group direct data transfer (DMA)
  42. Built-in random function
  43. Built-in Functions: Work-Item Functions
  44. Linking
  45. Complex Numbers
  46. Specification error
  47. Way to know if OpenCL is installed
  48. Math functions take combined vector and scalar inputs
  49. read/write support for image2d and image3d
  50. MPI like support
  51. MSAA images
  52. Two suggestions for supporting a multi-process environment
  53. Needed: information about kernel parameters
  54. Carry in/out add and borrow in/out sub
  55. clCreateBuffer2D and clCreateBuffer3D
  56. char vs cl_char in API Calls
  57. Some suggestions!
  58. Replace get_num_groups with get_group_size
  59. clCreateSubBufferRect
  60. Open-source libOpenCL.so
  61. Sharing host memory with clSetKernelArg!
  62. Calling CPU kernels with lower overhead!
  63. Ability to READ multisample image buffers
  64. Need a way to calculate theoretical FLOPS of a device
  65. Templates support
  66. New feature request : kernel parameters as 'static' !
  67. Add to OpenCL 1.2 some CUDA 4.0 and DX11.1 CS features!
  68. Robustness
  69. Predefined Macros: device type
  70. setKernelArg as size in bytes and NULL for __global pointers
  71. Support of Multi-dimensional arrays
  72. CL_MAP_WRITE: avoid copying data from device to host
  73. Tails automatization
  74. Dynamic workitems insertion while kernel is running
  75. letting runtime decide where to execute stuff
  76. 16 bit integer Atomics
  77. OpenGL-Interop for 16bit Unsigned Integer (GL_LUMINANCE16UI)
  78. Wishlist
  79. Add gentype1 in addition to gentype to complete gentypen
  80. Image3D with different size for each slice
  81. Add the option to chose the OpenCL version to instance
  82. get_compute_unit_id and get_processing_element_id functions
  83. Device Fusion
  84. Ability to determine main host's CPU
  85. Device info: maximum number of workitems per compute unit
  86. Reverse (.rev) Vector Components
  87. get_group_offset
  88. Kernel variables access in any level
  89. clCreateImage : 'host_ptr_max_size' argument
  90. Debugger and Profiler API
  91. Timeouts for GPU kernels
  92. Well defined ways of detecting product ID
  93. Public extensions and add-ons on github
  94. Atomic Function
  95. Array of images as kernel parameter
  96. SPIR kernel precompilation tool
  97. Support for complex math
  98. My suggestions for OpenCL 2.0..
  99. Add support for multiple binary formats and query mecanism
  100. rgba vector accessors
  101. goto statement
  102. cl_khr_gl_sharing and creating shared vbo(s) after cl context creation.
  103. quad,quadn
  104. OpenCL 2.0 considerations
  105. Suggestions for improving the OpenCL C++ API?
  106. detecting if a gpu device is connected to display device driver
  107. Spec confusion regarding convert_ functions
  108. Accept (int) comparison for select() for all scalar types.