dcompute
dcompute copied to clipboard
libmir
DCompute: Native execution of D on GPUs and other Accelerators
I'm thinking on implementing a GPU support into my 2D game engine (currently relies on SSE2 for graphics), and since using most preexisting libraries would be a bit clunky for...
Due to 18101 and 18472 the OCL getinfo generation fails to compile. This is to remind me to re-enable it when those get fixed.
According to this post, standard C kernels can be used with dcompute: https://forum.dlang.org/post/[email protected] I don't see any mention of that in the project description or on the main wiki page.
Now that we have stuff we can test, we should start testing. Circle appears to have CUDA support for enterprise with nvidia-docker. For OpenCL it is probably easiest to use...
When loading a kernel using `Program.fromFile(...)`, error codes can also be produced that are in the enum `cudaError_enum` that is present in` cuda.h`, not only from cudaError in `driver_types.h`. This...
Can we use library @nothrow @nogc generic code in kernels. For example `Slice!(Contiguous, [1], GlobalPointer!T)` and ndslice topology?
- SharedPointer and constant memory allocations [1] - getGlobal for cuda.program - cuda texture sampling functions dcompute/std/cuda/texture.d - math functions for cuda dcompute.std.cuda.math, extracted from cuda's libdevice.10.bc 1: inline IR...
Resolves #84 The third parameter of clGetDeviceInfo() requires specifying the size of the retrieved data. When retrieving enum type fields, the type size did not match the OpenCL specification. (Enum...
Get Device Type result is incorrect At tests/main.d, I tried printing the device information, ```d auto devices = platform.getDevices(theAllocator); writeln("Devices:"); foreach (i, ref d; devices) { writefln("\t[%d] %s", i, d.name);...