tmk::gpu::metal Namespace Reference

Functions
std::size_t	device_count ()
	One stable, registry-ID-ordered snapshot shared by discovery and field dispatch.
id< MTLDevice >	device_at_index (int index)
std::size_t	working_set_headroom (id< MTLDevice > device)
	Process working-set budget, not free VRAM: currentAllocatedSize excludes other processes.
std::mutex &	allocation_mutex ()
	Serialize admission and allocation so callers cannot admit against the same headroom.

Function Documentation

allocation_mutex()

std::mutex & tmk::gpu::metal::allocation_mutex

(

)

Serialize admission and allocation so callers cannot admit against the same headroom.

device_at_index()

id< MTLDevice > tmk::gpu::metal::device_at_index

(

int

index

)

device_count()

std::size_t tmk::gpu::metal::device_count

(

)

One stable, registry-ID-ordered snapshot shared by discovery and field dispatch.

working_set_headroom()

std::size_t tmk::gpu::metal::working_set_headroom

(

id< MTLDevice >

device

)

Process working-set budget, not free VRAM: currentAllocatedSize excludes other processes.