Runtime Concerns

Get the Horse64 runtime as part of the SDK. This document talks about technical detail concerns regarding the runtime and underlying implementations.

How code reaches the runtime

Check the compilation section.

Garbage collection

To make Horse64 powerful to use while still efficient for the programmer, it helps you out with automatic memory management using an approach Garbage Collection. This is implemented in HVM.

Generally, data types that allow referencing to other values that may reference back in cycles need more complex cleanup. Therefore, they cause so-called GC load when created. Items that can't produce complex cyclic references are cheaper to handle (in terms of computation time) and therefore may make your code faster if used instead.

VM worker threads

HVM can run multiple concurrent executions at the same time, as long as any are scheduled in parallel. HVM uses a pool of multiple worker threads using hardware threading to run with true parallelism. The amount of workers depends on the detected amount of CPU cores, but a minimum of two threads is used.

Multimedia, filesystem access, and else

For a lot of its base functionality HVM, and therefore by extension Horse64 programs, use Spew3D.

Platforms

Horse64 should be portable to almost all Unix-like platforms with some effort, at least for headless apps, for all gcc or clang style compilers. For graphical apps, it's mostly bound to SDL2's platform support. Here's the current state on platform support:

Platform names used by Horse64 tooling