2.2. Execution Model

This section outlines the execution model of a Vulkan system.

Vulkan exposes one or more devices, each of which exposes one or more queues which may process work asynchronously to one another. The set of queues supported by a device is partitioned into families. Each family supports one or more types of functionality and may contain multiple queues with similar characteristics. Queues within a single family are considered compatible with one another, and work produced for a family of queues can be executed on any queue within that family. This specification defines four types of functionality that queues may support: graphics, compute, transfer, and sparse memory management.


A single device may report multiple similar queue families rather than, or as well as, reporting multiple members of one or more of those families. This indicates that while members of those families have similar capabilities, they are not directly compatible with one another.

Device memory is explicitly managed by the application. Each device may advertise one or more heaps, representing different areas of memory. Memory heaps are either device local or host local, but are always visible to the device. Further detail about memory heaps is exposed via memory types available on that heap. Examples of memory areas that may be available on an implementation include:

On other architectures, there may only be a single heap that can be used for any purpose.

A Vulkan application controls a set of devices through the submission of command buffers which have recorded device commands issued via Vulkan library calls. The content of command buffers is specific to the underlying hardware and is opaque to the application. Once constructed, a command buffer can be submitted once or many times to a queue for execution. Multiple command buffers can be built in parallel by employing multiple threads within the application.

Command buffers submitted to different queues may execute in parallel or even out of order with respect to one another. Command buffers submitted to a single queue respect the submission order, as described further in Queue Operation. Command buffer execution by the device is also asynchronous to host execution. Once a command buffer is submitted to a queue, control may return to the application immediately. Synchronization between the device and host, and between different queues is the responsibility of the application.

2.2.1. Queue Operation

Vulkan queues provide an interface to the execution engines of a device. Commands for these execution engines are recorded into command buffers ahead of execution time. These command buffers are then submitted to queues with a queue submission command for execution in a number of batches. Once submitted to a queue, these commands will begin and complete execution without further application intervention, though the order of this execution is dependent on a number of implicit and explicit ordering constraints.

Work is submitted to queues using queue submission commands that typically take the form vkQueue* (e.g. vkQueueSubmit, vkQueueBindSparse), and optionally take a list of semaphores upon which to wait before work begins and a list of semaphores to signal once work has completed. The work itself, as well as signalling and waiting on the semaphores are all queue operations.

Queue operations on different queues have no implicit ordering constraints, and may execute in any order. Explicit ordering constraints between queues can be expressed with semaphores and fences.

Command buffer submissions to a single queue must always adhere to command order and API order, but otherwise may overlap or execute out of order. Other types of batches and queue submissions against a single queue (e.g. sparse memory binding) have no implicit ordering constraints with any other queue submission or batch. Additional explicit ordering constraints between queue submissions and individual batches can be expressed with semaphores and fences.

Before a fence or semaphore is signaled, it is guaranteed that any previously submitted queue operations have completed execution, and that memory writes from those queue operations are available to future queue operations. Waiting on a signaled semaphore or fence guarantees that previous writes that are available are also visible to subsequent commands.

Command buffer boundaries, both between primary command buffers of the same or different batches or submissions as well as between primary and secondary command buffers, do not introduce any implicit ordering constraints. In other words, submitting the set of command buffers (which can include executing secondary command buffers) between any semaphore or fence operations execute the recorded commands as if they had all been recorded into a single primary command buffer, except that the current state is reset on each boundary. Explicit ordering constraints can be expressed with events and pipeline barriers.

There are a few implicit ordering constraints between commands within a command buffer, but only covering a subset of execution. Additional explicit ordering constraints can be expressed with events, pipeline barriers and subpass dependencies.


Implementations have significant freedom to overlap execution of work submitted to a queue, and this is common due to deep pipelining and parallelism in Vulkan devices.

Commands recorded in command buffers either perform actions (draw, dispatch, clear, copy, query/timestamp operations, begin/end subpass operations), set state (bind pipelines, descriptor sets, and buffers, set dynamic state, push constants, set render pass/subpass state), or perform synchronization (set/wait events, pipeline barrier, render pass/subpass dependencies). Some commands perform more than one of these tasks. State setting commands update the current state of the command buffer. Some commands that perform actions (e.g. draw/dispatch) do so based on the current state set cumulatively since the start of the command buffer. The work involved in performing action commands is often allowed to overlap or to be reordered, but doing so must not alter the state to be used by each action command. In general, action commands are those commands that alter framebuffer attachments, read/write buffer or image memory, or write to query pools.

Synchronization commands introduce explicit execution and memory dependencies between two sets of action commands, where the second set of commands depends on the first set of commands. These dependencies enforce that both the execution of certain pipeline stages in the later set occur after the execution of certain stages in the source set, and that the effects of memory accesses performed by certain pipeline stages occur in order and are visible to each other. When not enforced by an explicit dependency or otherwise forbidden by the specification, action commands may overlap execution or execute out of order, and may not see the side effects of each other’s memory accesses.

The execution order of an action command with respect to any synchronization commands that affect that action command must match the recording and submission order, within submissions to a single queue.

Within a subpass of a render pass instance, for a given (x,y,layer,sample) sample location, the following stages are guaranteed to execute in API order for each separate primitive that includes that sample location:

  • depth bounds test
  • stencil test, stencil op and stencil write
  • depth test and depth write
  • occlusion queries
  • blending, logic op and color write

where the API order sorts primitives:

  • First, by the action command that generates them.
  • Second, by the order they are processed by primitive assembly.

Within this order, implementations also sort primitives:

  • Third, by an implementation-dependent ordering of new primitives generated by tessellation, if a tessellation shader is active.
  • Fourth, by the order new primitives are generated by geometry shading, if geometry shading is active.
  • Fifth, by an implementation-dependent ordering of primitives generated due to the polygon mode.

The device executes queue operations asynchronously with respect to the host. Control is returned to an application immediately following command buffer submission to a queue. The application must synchronize work between the host and device as needed.