Я хочу всё совсем опенсорсное для AMD.

Если я правильно понял, эту страницу
https://wiki.gentoo.org/wiki/OpenCL

то раньше был пакет pocl, который мог бы заменить AMD SDK.

Является ли AMD SDK совсем опенсорсным?
Куда делся dev-libs/pocl?

Какие пакеты вообще должны быть установлены?

Как это потом настраивать. Например сейчас у меня:

# eselect opencl list
Available OpenCL implementations:
[1] mesa *
[2] nvidia
[3] ocl-icd

Это нормально вообще, что mesa выбрана вместо ocl-icd ?

Как проверить версию установленного opencl? какой консольной командой?

Как X-client получает доступ к opencl через XOrg-сервер — через какое расширение X-протокола?

ЭФ>Как проверить версию установленного opencl? какой консольной командой?

пакет dev-util/clinfo

# clinfo
Number of platforms 1
Platform Name Clover
Platform Vendor Mesa
Platform Version OpenCL 1.1 Mesa 17.2.5
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_icd
Platform Extensions function suffix MESA

Platform Name Clover
Number of devices 1
Device Name AMD BONAIRE (DRM 2.50.0 / 4.13.0-sabayon, LLVM 5.0.0)
Device Vendor AMD
Device Vendor ID 0x1002
Device Version OpenCL 1.1 Mesa 17.2.5
Driver Version 17.2.5
Device OpenCL C Version OpenCL C 1.1
Device Type GPU
Device Available Yes
Device Profile FULL_PROFILE
Max compute units 14
Max clock frequency 1050MHz
Max work item dimensions 3
Max work item sizes 256x256x256
Max work group size 256
Compiler Available Yes
Preferred work group size multiple 64
Preferred / native vector sizes
char 16 / 16
short 8 / 8
int 4 / 4
long 2 / 2
half 0 / 0 (n/a)
float 4 / 4
double 2 / 2 (cl_khr_fp64)
Half-precision Floating-point support (n/a)
Single-precision Floating-point support (core)
Denormals No
Infinity and NANs Yes
Round to nearest Yes
Round to zero No
Round to infinity No
IEEE754-2008 fused multiply-add No
Support is emulated in software No
Correctly-rounded divide and sqrt operations No
Double-precision Floating-point support (cl_khr_fp64)
Denormals Yes
Infinity and NANs Yes
Round to nearest Yes
Round to zero Yes
Round to infinity Yes
IEEE754-2008 fused multiply-add Yes
Support is emulated in software No
Address bits 64, Little-Endian
Global memory size 2147483648 (2GiB)
Error Correction support No
Max memory allocation 1503238553 (1.4GiB)
Unified memory for Host and Device No
Minimum alignment for any data type 128 bytes
Alignment of base address 1024 bits (128 bytes)
Global Memory cache type None
Image support No
Local memory type Local
Local memory size 32768 (32KiB)
Max constant buffer size 1503238553 (1.4GiB)
Max number of constant args 16
Max size of kernel argument 1024
Queue properties
Out-of-order execution No
Profiling Yes
Profiling timer resolution 0ns
Execution capabilities
Run OpenCL kernels Yes
Run native kernels No
Device Extensions cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_byte_addressable_store cl_khr_fp64

NULL platform behavior
clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...) No platform
clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...) No platform
clCreateContext(NULL, ...) [default] No platform
clCreateContext(NULL, ...) [other] Success [MESA]
clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT) Success (1)
Platform Name Clover
Device Name AMD BONAIRE (DRM 2.50.0 / 4.13.0-sabayon, LLVM 5.0.0)
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) Success (1)
Platform Name Clover
Device Name AMD BONAIRE (DRM 2.50.0 / 4.13.0-sabayon, LLVM 5.0.0)
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL) Success (1)
Platform Name Clover
Device Name AMD BONAIRE (DRM 2.50.0 / 4.13.0-sabayon, LLVM 5.0.0)

ЭФ>Является ли AMD SDK совсем опенсорсным?

Видимо нет, раз его предлагают скачивать самостоятельно руками:
http://gpo.zugaina.org/dev-util/amdapp-sdk

ЭФ>Какие пакеты вообще должны быть установлены?

Есть ещё два непонятных пакета:

dev-libs/freeocl — OpenCL 1.2 implementation for CPUs
ну тут вроде бы ещё понятно, будет CPU считать (наверное на тот случай, если GPU нет...)
dev-libs/libclc — OpenCL C library (http://libclc.llvm.org/)
а тут уже непонятно, зачем это, почему описания не совпадают
libclc currently supports the AMDGCN, and R600 and NVPTX targets, OpenCL 1.1 Specification

ЭФ>Как X-client получает доступ к opencl через XOrg-сервер — через какое расширение X-протокола?

https://en.wikipedia.org/wiki/X_Window_System_protocols_and_architecture#Extensions

Никакого такого специального экстеншена нет. Может это надо как-то через OpenGL протащить?

Я просто список покажу, может что пригодится...

spc ~ # qsearch opencl   
dev-libs/amdgpu-pro-opencl Proprietary OpenCL implementation for AMD GPUs
dev-python/pyopencl Python wrapper for OpenCL
games-action/openclonk A free multiplayer action game where you control clonks
media-gfx/openclipart Open Clip Art Library (openclipart.org)
net-wireless/cpyrit-opencl A sub-package that adds OpenCL-capability to Pyrit
virtual/opencl Virtual for OpenCL implementations
dev-libs/opencl-clhpp OpenCL Host API C++ bindings (cl.hpp and cl2.hpp)
dev-haskell/opencl Haskell high-level wrapper for OpenCL
dev-haskell/openclwrappers The OpenCL Standard for heterogenous data-parallel computing
spc ~ # qsearch -S opencl
dev-libs/amdgpu-pro-opencl Proprietary OpenCL implementation for AMD GPUs
dev-libs/beignet OpenCL implementation for Intel GPUs
dev-libs/cl OpenCL bindings for Erlang
dev-libs/libclc OpenCL C library
dev-libs/ocl-icd Alternative to vendor specific OpenCL ICD loaders
dev-python/pyopencl Python wrapper for OpenCL
dev-util/clinfo A tool to display info about the system's OpenCL capabilities
dev-util/intel-ocl-sdk Intel's implementation of the OpenCL standard optimized for Intel processors
media-gfx/openclipart Open Clip Art Library (openclipart.org)
net-wireless/cpyrit-opencl A sub-package that adds OpenCL-capability to Pyrit
sci-libs/clblas Library containing BLAS routines for OpenCL
sci-libs/clblast Tuned OpenCL BLAS
virtual/opencl Virtual for OpenCL implementations
dev-libs/boost-compute A header-only C++ Computing Library for OpenCL
dev-libs/fortrancl OpenCL interface for Fortran 90
dev-libs/opencl-clhpp OpenCL Host API C++ bindings (cl.hpp and cl2.hpp)
sci-libs/clblas A software library containing BLAS routines for OpenCL
sci-libs/clfft Library containing FFT functions written in OpenCL
sci-libs/clrng A library for uniform random number generation in OpenCL
sys-devel/gdebugger OpenCL and OpenGL debugger and memory analyzer
dev-haskell/language-c-quote C/CUDA/OpenCL/Objective-C quasiquoting library
dev-haskell/opencl Haskell high-level wrapper for OpenCL
dev-haskell/openclwrappers The OpenCL Standard for heterogenous data-parallel computing
spc ~ # euses opencl
app-crypt/johntheripper:opencl - Enable opencl support for speeding up cracking on capable devices
app-emulation/wine:opencl - Enable OpenCL support
app-emulation/wine-any:opencl - Enable OpenCL support
app-emulation/wine-d3d9:opencl - Enable OpenCL support
app-emulation/wine-staging:opencl - Enable OpenCL support
app-emulation/wine-vanilla:opencl - Enable OpenCL support
app-text/tesseract:opencl - Enable opencl support for speedup using GPU computation.
dev-libs/actor-framework:opencl - Compiles against the provided opencl implementation
dev-libs/libgpuarray:opencl - Enable OpenCL support
dev-libs/starpu:opencl - Enable OpenCL support
dev-python/pygpu:opencl - Enable OpenCL support
dev-util/nvidia-cuda-sdk:opencl - Build OpenCL binaries
games-board/freedoko:openclipartcards - Install the openclipartcards card set
media-gfx/darktable:opencl - Enable opencl support
media-gfx/freewrl:opencl - Enable OpenCL support
media-gfx/imagemagick:opencl - Enable OpenCL support
media-libs/mesa:opencl - Enable the Clover Gallium OpenCL state tracker.
media-libs/opencv:opencl - Add support for OpenCL
media-libs/opensubdiv:opencl - Enable OpenCL support through virtual/opencl.
media-libs/x264:opencl - Add support for OpenCL.
media-video/ffdiaporama:openclipart - Support openclipart database
net-misc/bfgminer:opencl - Enable support for mining with OpenCL-capable devices (GPUs)
net-p2p/go-ethereum:opencl - Add OpenCL GPU mining support
net-p2p/pybitmessage:opencl - Uses dev-python/pyopencl for GPU acceleration.
net-wireless/pyrit:opencl - Enable OpenCL support via net-wireless/cpyrit-opencl
sci-chemistry/gromacs:opencl - Enable opencl non-bonded kernels
sci-geosciences/grass:opencl - Enable OpenCL support
sci-libs/gdal:opencl - Enable OpenCL support
sci-libs/shogun:opencl - Enable support for building against OpenCL