The RenderDevice encapsulates an SDL_GPUDevice and provides methods for allocating GPU resources, managing compute, copy and render passes, and . More...

#include <RenderDevice.h>

Inheritance diagram for RenderDevice:

Public Member Functions
CommandBuffer *	acquireCommandBuffer (const RenderDevice *self)
	Acquires a new CommandBuffer for recording GPU commands this frame.

CommandBuffer *	beginFrame (RenderDevice *self)
	Begins a frame: acquires a command buffer and the swapchain, and prepares the framebuffer.

Fence *	endFrameAndFence (RenderDevice *self)
	Ends the frame begun by `beginFrame`, as `endFrame`, but returns a Fence.

void	endFrame (RenderDevice *self)
	Ends the frame begun by `beginFrame`: presents the framebuffer and submits.

Buffer *	createBuffer (RenderDevice self, const SDL_GPUBufferCreateInfo info)
	Creates a Buffer (vertex, index, indirect, storage, etc.).

Buffer *	createBufferWithConstMem (RenderDevice self, SDL_GPUBufferUsageFlags usage, const void mem, Uint32 size)
	Creates a Buffer and uploads `mem` in a single call.

Buffer *	createBufferWithData (RenderDevice self, SDL_GPUBufferUsageFlags usage, const Data data)
	Creates a Buffer and uploads the contents of an Objectively `Data` object.

ComputePipeline *	createComputePipeline (RenderDevice self, const SDL_GPUComputePipelineCreateInfo info)
	Creates a ComputePipeline from a compiled compute shader.

Framebuffer *	createFramebuffer (RenderDevice self, const GPU_FramebufferCreateInfo info)
	Creates a Framebuffer aggregating color/depth attachments (and an MSAA resolve target).

GraphicsPipeline *	createGraphicsPipeline (RenderDevice self, const SDL_GPUGraphicsPipelineCreateInfo info)
	Creates a GraphicsPipeline from vertex and fragment shaders plus render state.

QueryPool *	createQueryPool (RenderDevice self, const SDL_GPUQueryPoolCreateInfo info)
	Creates a QueryPool for occlusion or timestamp queries.

Sampler *	createSampler (RenderDevice self, const SDL_GPUSamplerCreateInfo info)
	Creates a Sampler describing filter and address modes.

Sampler *	createSamplerLinearRepeat (RenderDevice *self)
	Creates a trilinear, wrapping Sampler – the common preset for tiled surface textures.

Sampler *	createSamplerLinearClamp (RenderDevice *self)
	Creates a trilinear, edge-clamped Sampler – the common preset for continuously-sampled volumes and non-tiling images (ambient/voxel volumes, cubemaps, post-process targets, UI images).

Sampler *	createSamplerNearestClamp (RenderDevice *self)
	Creates a point-filtered, edge-clamped Sampler – the common preset for data textures fetched by integer coordinate (voxel light data, depth-attachment copies) rather than sampled for appearance.

Sampler *	createSamplerShadowCompare (RenderDevice *self)
	Creates a hardware-PCF comparison Sampler for shadow map sampling.

Shader *	createShader (RenderDevice self, const SDL_GPUShaderCreateInfo info)
	Creates a Shader from a fully-filled `SDL_GPUShaderCreateInfo`.

Texture *	createTexture (RenderDevice self, const SDL_GPUTextureCreateInfo info, const void *pixels)
	Creates a Texture, optionally uploading initial pixel data.

TransferBuffer *	createTransferBuffer (RenderDevice self, const SDL_GPUTransferBufferCreateInfo info)
	Creates a CPU-accessible transfer buffer for staging data to or from the GPU.

Texture *	createTextureFromSurface (RenderDevice self, SDL_Surface surface, SDL_GPUTextureUsageFlags usage, bool generateMipmaps)
	Creates a Texture from an `SDL_Surface`, uploading pixel data immediately.

Texture *	createSolidColorTexture (RenderDevice *self, SDL_GPUTextureType type, Uint32 layerCount, Uint32 rgba)
	Creates a small, solid-color Texture – the common "placeholder" pattern for a sampler slot a shader declares but a given draw never actually reads.

SDL_GPUTextureFormat	getSwapchainTextureFormat (const RenderDevice *self)
	Returns the pixel format of this device's swapchain.

RenderDevice *	init (RenderDevice *self)
	Initialises a RenderDevice, creating an `SDL_GPUDevice` with no window claimed.

RenderDevice *	initWithWindow (RenderDevice self, SDL_Window window)
	Initialises a RenderDevice and immediately claims `window` for rendering.

Shader *	loadShader (RenderDevice self, const char name, const SDL_GPUShaderCreateInfo *info)
	Loads a shader blob via the Resource system and creates a Shader.

ComputePipeline *	loadComputePipeline (RenderDevice self, const char name, const SDL_GPUComputePipelineCreateInfo *info)
	Loads a compiled compute shader blob via the Resource system and creates a ComputePipeline.

GraphicsPipeline *	loadGraphicsPipeline (RenderDevice self, const char vertexShaderName, const SDL_GPUShaderCreateInfo vertexShaderInfo, const char fragmentShaderName, const SDL_GPUShaderCreateInfo fragmentShaderInfo, SDL_GPUGraphicsPipelineCreateInfo info)
	Loads a vertex/fragment shader pair and creates a GraphicsPipeline from them.

bool	setAllowedFramesInFlight (const RenderDevice *self, Uint32 allowed)
	Sets the maximum number of GPU frames that may be in flight concurrently.

void	setFramebuffer (RenderDevice self, Framebuffer framebuffer)
	Sets the present-target Framebuffer driven by `beginFrame`/`endFrame`.

bool	setSwapchainParameters (const RenderDevice self, SDL_Window window, SDL_GPUSwapchainComposition composition, SDL_GPUPresentMode mode)

void	setWindow (RenderDevice self, SDL_Window window)
	Claims `window` for this device, releasing any previously claimed window.

bool	supportsPresentMode (const RenderDevice *self, SDL_GPUPresentMode mode)
	Queries whether this device's swapchain supports the given present mode.

bool	supportsSwapchainComposition (const RenderDevice *self, SDL_GPUSwapchainComposition composition)
	Queries whether this device's swapchain supports the given composition (colour space / HDR mode).

bool	textureSupportsFormat (const RenderDevice *self, SDL_GPUTextureFormat format, SDL_GPUTextureType type, SDL_GPUTextureUsageFlags usage)
	Queries whether a texture format is supported for a given type and usage.

bool	textureSupportsSampleCount (const RenderDevice *self, SDL_GPUTextureFormat format, SDL_GPUSampleCount sample_count)
	Queries whether a texture format supports a given MSAA sample count.

bool	waitForIdle (const RenderDevice *self)
	Blocks until the GPU has finished all submitted work.

bool	waitForSwapchain (const RenderDevice *self)
	Blocks until this device's swapchain is available for the next frame.

Class *	_RenderDevice (void)
	The RenderDevice archetype.

bool	setSwapchainParameters (const RenderDevice *self, SDL_GPUSwapchainComposition composition, SDL_GPUPresentMode mode)
	Configures swapchain composition and present mode for this device's window.

Data Fields
Object	object
	The superclass.

SDL_GPUDevice *	device
	The `SDL_GPUDevice`.

SDL_Window *	window
	The `SDL_Window` associated with `device`.

SDL_GPUShaderFormat	shaderFormats

float	maxAnisotropy
	The anisotropic filtering level applied by `createSamplerLinearRepeat` and `createSamplerLinearClamp`. Defaults to `0` (anisotropic filtering disabled). Assign directly; only affects Samplers created afterward, since SDL_gpu Samplers are immutable once created.

Framebuffer *	framebuffer
	The present-target Framebuffer driven by `beginFrame`/`endFrame`, or `NULL`.

ident	data
	User data.

Protected Attributes
RenderDeviceInterface *	interface
	The interface.

Detailed Description

The RenderDevice encapsulates an SDL_GPUDevice and provides methods for allocating GPU resources, managing compute, copy and render passes, and .

Member Function Documentation

◆ _RenderDevice()

Class * _RenderDevice ( void )

The RenderDevice archetype.

Returns: The RenderDevice Class.

◆ acquireCommandBuffer()

CommandBuffer * acquireCommandBuffer ( const RenderDevice * self )

Acquires a new CommandBuffer for recording GPU commands this frame.

Wraps SDL_AcquireGPUCommandBuffer. The returned CommandBuffer is owned by the caller and must be submitted via submit or submitAndFence and then released with release.

Parameters

self	The RenderDevice.

Returns: A new, retained CommandBuffer. GPU_Asserts on failure.

◆ beginFrame()

CommandBuffer * beginFrame ( RenderDevice * self )

Begins a frame: acquires a command buffer and the swapchain, and prepares the framebuffer.

Convenience over the manual acquire→wait-swapchain→resize boilerplate. Acquires a CommandBuffer, blocks for the swapchain texture, resizes the framebuffer set via setFramebuffer to the swapchain dimensions, and returns the command buffer so the caller can record passes into framebuffer. Returns NULL when the swapchain is unavailable (e.g. the window is minimised); the command buffer is cancelled and the frame should be skipped. Pair every non-NULL return with endFrame. The returned CommandBuffer is owned by the device; do not release it. Applications that render directly to the swapchain can ignore beginFrame/endFrame and drive the command buffer themselves.

Parameters

self	The RenderDevice.

Returns: The frame's CommandBuffer (borrowed), or NULL to skip the frame.

◆ createBuffer()

Buffer * createBuffer	(	RenderDevice *	self,
		const SDL_GPUBufferCreateInfo *	info
	)

Creates a Buffer (vertex, index, indirect, storage, etc.).

Convenience factory for Buffer::initWithDevice. The buffer's contents are uninitialised; use Buffer::upload to populate it.

Parameters

self	The RenderDevice.
info	Buffer creation parameters (usage flags, size).

Returns: A new, retained Buffer. GPU_Asserts on failure. Free with release.

◆ createBufferWithConstMem()

Buffer * createBufferWithConstMem	(	RenderDevice *	self,
		SDL_GPUBufferUsageFlags	usage,
		const void *	mem,
		Uint32	size
	)

Creates a Buffer and uploads mem in a single call.

Convenience factory for Buffer::initWithConstMem. The returned buffer is immediately GPU-resident.

Parameters

self	The RenderDevice.
usage	Buffer usage flags (e.g. `SDL_GPU_BUFFERUSAGE_VERTEX`).
mem	CPU pointer to the source data. Must not be NULL.
size	Number of bytes to allocate and upload.

Returns: A new, retained Buffer populated with mem. GPU_Asserts on failure. Free with release.

◆ createBufferWithData()

Buffer * createBufferWithData	(	RenderDevice *	self,
		SDL_GPUBufferUsageFlags	usage,
		const Data *	data
	)

Creates a Buffer and uploads the contents of an Objectively Data object.

Convenience factory for Buffer::initWithData, for data loaded via the Objectively Resource system.

Parameters

self	The RenderDevice.
usage	Buffer usage flags (e.g. `SDL_GPU_BUFFERUSAGE_VERTEX`).
data	Source data object. Must not be NULL.

Returns: A new, retained Buffer populated with data. GPU_Asserts on failure. Free with release.

◆ createComputePipeline()

ComputePipeline * createComputePipeline	(	RenderDevice *	self,
		const SDL_GPUComputePipelineCreateInfo *	info
	)

Creates a ComputePipeline from a compiled compute shader.

Convenience factory for ComputePipeline::initWithDevice. Prefer loadComputePipeline to load a compiled blob from the Resource system.

Parameters

self	The RenderDevice.
info	Compute pipeline creation parameters, including the shader.

Returns: A new, retained ComputePipeline. GPU_Asserts on failure. Free with release.

◆ createFramebuffer()

Framebuffer * createFramebuffer	(	RenderDevice *	self,
		const GPU_FramebufferCreateInfo *	info
	)

Creates a Framebuffer aggregating color/depth attachments (and an MSAA resolve target).

Convenience factory for Framebuffer::initWithDevice. Pair with setFramebuffer to drive it via beginFrame/endFrame.

Parameters

self	The RenderDevice.
info	Framebuffer creation parameters (size, formats, sample count).

Returns: A new, retained Framebuffer. GPU_Asserts on failure. Free with release.

◆ createGraphicsPipeline()

GraphicsPipeline * createGraphicsPipeline	(	RenderDevice *	self,
		const SDL_GPUGraphicsPipelineCreateInfo *	info
	)

Creates a GraphicsPipeline from vertex and fragment shaders plus render state.

Convenience factory for GraphicsPipeline::initWithDevice.

Parameters

self	The RenderDevice.
info	Graphics pipeline creation parameters (shaders, vertex layout, blend state, rasteriser state, depth/stencil state, colour target formats).

Returns: A new, retained GraphicsPipeline. GPU_Asserts on failure. Free with release.

◆ createQueryPool()

QueryPool * createQueryPool	(	RenderDevice *	self,
		const SDL_GPUQueryPoolCreateInfo *	info
	)

Creates a QueryPool for occlusion or timestamp queries.

Convenience factory for QueryPool::initWithDevice. Always succeeds, even when occlusion queries are unsupported by the linked SDL3 – see QueryPool.

Parameters

self	The RenderDevice.
info	Query pool creation parameters (query type, query count).

Returns: A new, retained QueryPool. GPU_Asserts on failure. Free with release.

◆ createSampler()

Sampler * createSampler	(	RenderDevice *	self,
		const SDL_GPUSamplerCreateInfo *	info
	)

Creates a Sampler describing filter and address modes.

Convenience factory for Sampler::initWithDevice. For the common filtering presets below, prefer createSamplerLinearRepeat et al.

Parameters

self	The RenderDevice.
info	Sampler creation parameters (min/mag filter, mip mode, address modes, anisotropy, etc.).

Returns: A new, retained Sampler. GPU_Asserts on failure. Free with release.

◆ createSamplerLinearClamp()

Sampler * createSamplerLinearClamp ( RenderDevice * self )

Creates a trilinear, edge-clamped Sampler – the common preset for continuously-sampled volumes and non-tiling images (ambient/voxel volumes, cubemaps, post-process targets, UI images).

Linear min/mag filter, linear mipmap mode, CLAMP_TO_EDGE addressing on all three axes. Anisotropic filtering is enabled when self->maxAnisotropy is greater than 0.

Parameters

self	The RenderDevice.

Returns: A new, retained Sampler. GPU_Asserts on failure. Free with release.

◆ createSamplerLinearRepeat()

Sampler * createSamplerLinearRepeat ( RenderDevice * self )

Creates a trilinear, wrapping Sampler – the common preset for tiled surface textures.

Linear min/mag filter, linear mipmap mode, REPEAT addressing on all three axes. Anisotropic filtering is enabled when self->maxAnisotropy is greater than 0.

Parameters

self	The RenderDevice.

Returns: A new, retained Sampler. GPU_Asserts on failure. Free with release.

◆ createSamplerNearestClamp()

Sampler * createSamplerNearestClamp ( RenderDevice * self )

Creates a point-filtered, edge-clamped Sampler – the common preset for data textures fetched by integer coordinate (voxel light data, depth-attachment copies) rather than sampled for appearance.

Nearest min/mag filter, nearest mipmap mode, CLAMP_TO_EDGE addressing on all three axes.

Parameters

self	The RenderDevice.

Returns: A new, retained Sampler. GPU_Asserts on failure. Free with release.

◆ createSamplerShadowCompare()

Sampler * createSamplerShadowCompare ( RenderDevice * self )

Creates a hardware-PCF comparison Sampler for shadow map sampling.

Linear min/mag filter, nearest mipmap mode, CLAMP_TO_EDGE addressing, depth comparison enabled with SDL_GPU_COMPAREOP_LESS_OR_EQUAL.

Parameters

self	The RenderDevice.

Returns: A new, retained Sampler. GPU_Asserts on failure. Free with release.

◆ createShader()

Shader * createShader	(	RenderDevice *	self,
		const SDL_GPUShaderCreateInfo *	info
	)

Creates a Shader from a fully-filled SDL_GPUShaderCreateInfo.

Convenience factory for Shader::initWithDevice. All fields of info, including code, code_size, and format, must be set by the caller. Prefer loadShader to load compiled blobs from the Resource system with automatic format selection.

Parameters

self	The RenderDevice.
info	Shader creation parameters with all fields populated.

Returns: A new, retained Shader. GPU_Asserts on failure. Free with release.

◆ createSolidColorTexture()

Texture * createSolidColorTexture	(	RenderDevice *	self,
		SDL_GPUTextureType	type,
		Uint32	layerCount,
		Uint32	rgba
	)

Creates a small, solid-color Texture – the common "placeholder" pattern for a sampler slot a shader declares but a given draw never actually reads.

type may be SDL_GPU_TEXTURETYPE_2D (layerCount ignored, treated as 1) or SDL_GPU_TEXTURETYPE_CUBE (layerCount must be 6); every texel of every layer is filled with rgba. Always 1x1 per layer/face, SDL_GPU_TEXTUREFORMAT_R8G8B8A8_UNORM, one mip level, SAMPLER usage only.

Parameters

self	The RenderDevice.
type	`SDL_GPU_TEXTURETYPE_2D` or `SDL_GPU_TEXTURETYPE_CUBE`.
layerCount	The layer/face count (6 for `_CUBE`, otherwise ignored).
rgba	The fill color, packed as `0xAABBGGRR` (i.e. red in the lowest byte).

Returns: A new, retained Texture. GPU_Asserts on failure. Free with release.

◆ createTexture()

Texture * createTexture	(	RenderDevice *	self,
		const SDL_GPUTextureCreateInfo *	info,
		const void *	pixels
	)

Creates a Texture, optionally uploading initial pixel data.

Convenience factory for Texture::initWithDevice. When pixels is non-NULL, the pixel data is uploaded immediately so the texture is GPU-resident before returning.

Parameters

self	The RenderDevice.
info	Texture creation parameters (format, type, dimensions, usage, etc.).
pixels	Initial pixel data to upload, or `NULL` to leave the texture uninitialised.

Returns: A new, retained Texture. GPU_Asserts on failure. Free with release.

◆ createTextureFromSurface()

Texture * createTextureFromSurface	(	RenderDevice *	self,
		SDL_Surface *	surface,
		SDL_GPUTextureUsageFlags	usage,
		bool	generateMipmaps
	)

Creates a Texture from an SDL_Surface, uploading pixel data immediately.

Convenience factory for Texture::initWithSurface. Converts surface to SDL_PIXELFORMAT_RGBA32 if needed and uploads it as SDL_GPU_TEXTUREFORMAT_R8G8B8A8_UNORM. The surface is not modified or freed.

Parameters

self	The RenderDevice.
surface	The source surface. Must not be NULL.
usage	Texture usage flags (e.g. `SDL_GPU_TEXTUREUSAGE_SAMPLER`).
mipmaps	If true, a full mip chain is allocated and generated; see `Texture::initWithSurface`.

Returns: A new, retained Texture. GPU_Asserts on failure. Free with release.

◆ createTransferBuffer()

TransferBuffer * createTransferBuffer	(	RenderDevice *	self,
		const SDL_GPUTransferBufferCreateInfo *	info
	)

Creates a CPU-accessible transfer buffer for staging data to or from the GPU.

Parameters

self	The RenderDevice.
info	Transfer buffer creation parameters (usage: upload or download, size).

Returns: A new, retained TransferBuffer. GPU_Asserts on failure. Free with release.

◆ endFrame()

void endFrame ( RenderDevice * self )

Ends the frame begun by beginFrame: presents the framebuffer and submits.

Blits the framebuffer's resolved color (Framebuffer::resolveColorTexture, the resolve target when multisampled) into the acquired swapchain texture, submits the frame's command buffer, and releases it. Must be paired with a non-NULL beginFrame return.

Parameters

self	The RenderDevice.

◆ endFrameAndFence()

Fence * endFrameAndFence ( RenderDevice * self )

Ends the frame begun by beginFrame, as endFrame, but returns a Fence.

Identical to endFrame, except the frame's command buffer is submitted via CommandBuffer::submitAndFence, so callers may track when the frame's GPU work has completed, e.g. to safely read back data written during the frame. The returned Fence is retained; release it when no longer needed.

Parameters

self	The RenderDevice.

Returns: A new, retained Fence, or NULL on error.

◆ getSwapchainTextureFormat()

SDL_GPUTextureFormat getSwapchainTextureFormat ( const RenderDevice * self )

Returns the pixel format of this device's swapchain.

Useful for configuring render-pass colour target formats or creating pipelines that write to the swapchain. The device must have a window claimed.

Parameters

self	The RenderDevice.

Returns: The SDL_GPUTextureFormat of the window's swapchain.

◆ init()

RenderDevice * init ( RenderDevice * self )

Initialises a RenderDevice, creating an SDL_GPUDevice with no window claimed.

Creates the device supporting MSL, SPIR-V, and DXIL shader formats. Use setWindow or initWithWindow to associate a window after init.

Parameters

self	The RenderDevice.

Returns: The initialised RenderDevice, or NULL on failure.

◆ initWithWindow()

RenderDevice * initWithWindow	(	RenderDevice *	self,
		SDL_Window *	window
	)

Initialises a RenderDevice and immediately claims window for rendering.

Parameters

self	The RenderDevice.
window	The `SDL_Window` to claim for this device.

Returns: The initialised RenderDevice, or NULL on failure.

◆ loadComputePipeline()

ComputePipeline * loadComputePipeline	(	RenderDevice *	self,
		const char *	name,
		const SDL_GPUComputePipelineCreateInfo *	info
	)

Loads a compiled compute shader blob via the Resource system and creates a ComputePipeline.

Parallel to loadShader for compute stages. Appends each supported format's extension to name and resolves it via Objectively's ResourceProvider chain:

Metal (macOS/iOS): .metallib, then .metal
Vulkan (Linux/Android): .spv

D3D12 (Windows): .dxil The caller fills in entrypoint, thread counts, and binding counts in info. code, code_size, format, and a default entrypoint are filled in for you, then the completed info is passed to ComputePipeline::initWithDevice. Shader blobs are produced offline by sdl-shadercross.

Parameters

self	The RenderDevice.
name	Shader base name without extension, e.g. `"HelloCompute.comp"`.
info	Compute pipeline creation parameters. `code`, `code_size`, and `format` are ignored; all other fields must be set by the caller.

Returns: A new, retained ComputePipeline. GPU_Asserts on failure. Free with release.

◆ loadGraphicsPipeline()

GraphicsPipeline * loadGraphicsPipeline	(	RenderDevice *	self,
		const char *	vertexShaderName,
		const SDL_GPUShaderCreateInfo *	vertexShaderInfo,
		const char *	fragmentShaderName,
		const SDL_GPUShaderCreateInfo *	fragmentShaderInfo,
		SDL_GPUGraphicsPipelineCreateInfo *	info
	)

Loads a vertex/fragment shader pair and creates a GraphicsPipeline from them.

Convenience over the load-both/assign/create/release-both boilerplate every graphics pipeline needs: calls loadShader for vertexShaderName and fragmentShaderName, assigns the results to info's vertex_shader/ fragment_shader fields, calls createGraphicsPipeline, then releases both Shaders (they're retained by the pipeline internally; the caller never sees or owns them). info must otherwise be fully populated (vertex layout, blend/rasterizer/depth-stencil state, colour target formats) by the caller.

Parameters

self	The RenderDevice.
vertexShaderName	Vertex shader base name, e.g. `"shaders/bsp_vs"`.
vertexShaderInfo	Vertex shader creation parameters (see `loadShader`).
fragmentShaderName	Fragment shader base name, e.g. `"shaders/bsp_fs"`.
fragmentShaderInfo	Fragment shader creation parameters (see `loadShader`).
info	Graphics pipeline creation parameters; `vertex_shader`/`fragment_shader` are overwritten, all other fields must already be set.

Returns: A new, retained GraphicsPipeline. GPU_Asserts on failure. Free with release.

◆ loadShader()

Shader * loadShader	(	RenderDevice *	self,
		const char *	name,
		const SDL_GPUShaderCreateInfo *	info
	)

Loads a shader blob via the Resource system and creates a Shader.

Appends each supported format's extension to name and resolves it via Objectively's ResourceProvider chain:

Metal (macOS/iOS): .metallib, then .metal
Vulkan (Linux/Android): .spv

D3D12 (Windows): .dxil The caller fills in stage, entrypoint, and binding counts in info. code, code_size, format, and a default entrypoint are filled in for you, then the completed info is passed to Shader::initWithDevice. Shader blobs are produced offline by sdl-shadercross.

Parameters

self	The RenderDevice.
name	Shader base name without extension, e.g. `"shaders/Renderer.vert"`.
info	Shader creation parameters. `code`, `code_size`, and `format` are ignored; all other fields must be set by the caller.

Returns: A new, retained Shader. GPU_Asserts on failure. Free with release.

◆ setAllowedFramesInFlight()

bool setAllowedFramesInFlight	(	const RenderDevice *	self,
		Uint32	allowed
	)

Sets the maximum number of GPU frames that may be in flight concurrently.

Lower values reduce latency; higher values improve throughput. The default is determined by SDL. Takes effect on the next frame.

Parameters

self	The RenderDevice.
allowed	Maximum concurrent in-flight frames (1–3 are typical values).

Returns: true on success, false if the value is unsupported.

◆ setFramebuffer()

void setFramebuffer	(	RenderDevice *	self,
		Framebuffer *	framebuffer
	)

Sets the present-target Framebuffer driven by beginFrame/endFrame.

The device retains framebuffer and releases any previously set one. Construct the Framebuffer with the desired color/depth formats and MSAA sample count; beginFrame resizes it to the swapchain each frame. Pass NULL to clear.

Parameters

self	The RenderDevice.
framebuffer	The Framebuffer to present each frame, or `NULL`.

◆ setSwapchainParameters() [1/2]

bool setSwapchainParameters	(	const RenderDevice *	self,
		SDL_GPUSwapchainComposition	composition,
		SDL_GPUPresentMode	mode
	)

Configures swapchain composition and present mode for this device's window.

Use supportsSwapchainComposition and supportsPresentMode to guard against unsupported combinations before calling this.

Parameters

self	The RenderDevice.
composition	Colour space / HDR composition mode.
mode	Presentation mode (vsync, mailbox, immediate, etc.).

Returns: true on success, false if the combination is unsupported.

◆ setSwapchainParameters() [2/2]

bool setSwapchainParameters	(	const RenderDevice *	self,
		SDL_Window *	window,
		SDL_GPUSwapchainComposition	composition,
		SDL_GPUPresentMode	mode
	)

◆ setWindow()

void setWindow	(	RenderDevice *	self,
		SDL_Window *	window
	)

Claims window for this device, releasing any previously claimed window.

If a different window is already claimed it is released via SDL_ReleaseWindowFromGPUDevice first. Passing NULL releases the current window without claiming a new one.

Parameters

self	The RenderDevice.
window	The `SDL_Window` to claim, or `NULL` to release the current window.

◆ supportsPresentMode()

bool supportsPresentMode	(	const RenderDevice *	self,
		SDL_GPUPresentMode	mode
	)

Queries whether this device's swapchain supports the given present mode.

Parameters

self	The RenderDevice.
mode	The present mode to test.

Returns: true if the present mode is supported.

◆ supportsSwapchainComposition()

bool supportsSwapchainComposition	(	const RenderDevice *	self,
		SDL_GPUSwapchainComposition	composition
	)

Queries whether this device's swapchain supports the given composition (colour space / HDR mode).

Parameters

self	The RenderDevice.
composition	The swapchain composition to test.

Returns: true if the composition is supported.

◆ textureSupportsFormat()

bool textureSupportsFormat	(	const RenderDevice *	self,
		SDL_GPUTextureFormat	format,
		SDL_GPUTextureType	type,
		SDL_GPUTextureUsageFlags	usage
	)

Queries whether a texture format is supported for a given type and usage.

Parameters

self	The RenderDevice.
format	The texture format to query.
type	The texture dimensionality (2D, cube, 3D, etc.).
usage	Bitmask of intended usages (sampler, colour target, storage, etc.).

Returns: true if the format/type/usage combination is supported.

◆ textureSupportsSampleCount()

bool textureSupportsSampleCount	(	const RenderDevice *	self,
		SDL_GPUTextureFormat	format,
		SDL_GPUSampleCount	sample_count
	)

Queries whether a texture format supports a given MSAA sample count.

Parameters

self	The RenderDevice.
format	The texture format to query.
sample_count	The MSAA sample count to test.

Returns: true if the format supports the requested sample count.

◆ waitForIdle()

bool waitForIdle ( const RenderDevice * self )

Blocks until the GPU has finished all submitted work.

Use sparingly — typically only at shutdown or before releasing GPU resources that may still be in use.

Parameters

self	The RenderDevice.

Returns: true on success, false on error.

◆ waitForSwapchain()

bool waitForSwapchain ( const RenderDevice * self )

Blocks until this device's swapchain is available for the next frame.

Parameters

self	The RenderDevice.

Returns: true on success, false on error.

Field Documentation

◆ data

ident RenderDevice::data

User data.

◆ device

SDL_GPUDevice* RenderDevice::device

The SDL_GPUDevice.

◆ framebuffer

Framebuffer* RenderDevice::framebuffer

The present-target Framebuffer driven by beginFrame/endFrame, or NULL.

Set via setFramebuffer (retained). beginFrame resizes it to the swapchain each frame and endFrame blits its resolved color to the swapchain.

◆ interface

RenderDeviceInterface* RenderDevice::interface

protected

The interface.

◆ maxAnisotropy

float RenderDevice::maxAnisotropy

The anisotropic filtering level applied by createSamplerLinearRepeat and createSamplerLinearClamp. Defaults to 0 (anisotropic filtering disabled). Assign directly; only affects Samplers created afterward, since SDL_gpu Samplers are immutable once created.

◆ object

Object RenderDevice::object

The superclass.

◆ shaderFormats

SDL_GPUShaderFormat RenderDevice::shaderFormats

The bitmask of supported shader formats.

◆ window

SDL_Window* RenderDevice::window

The SDL_Window associated with device.

The documentation for this struct was generated from the following files:

Sources/ObjectivelyGPU/RenderDevice.h
Sources/ObjectivelyGPU/RenderDevice.c

Public Member Functions

Data Fields

Protected Attributes

Detailed Description

Member Function Documentation

◆ _RenderDevice()

◆ acquireCommandBuffer()

◆ beginFrame()

◆ createBuffer()

◆ createBufferWithConstMem()

◆ createBufferWithData()

◆ createComputePipeline()

◆ createFramebuffer()

◆ createGraphicsPipeline()

◆ createQueryPool()

◆ createSampler()

◆ createSamplerLinearClamp()

◆ createSamplerLinearRepeat()

◆ createSamplerNearestClamp()

◆ createSamplerShadowCompare()

◆ createShader()

◆ createSolidColorTexture()

◆ createTexture()

◆ createTextureFromSurface()

◆ createTransferBuffer()

◆ endFrame()

◆ endFrameAndFence()

◆ getSwapchainTextureFormat()

◆ init()

◆ initWithWindow()

◆ loadComputePipeline()

◆ loadGraphicsPipeline()

◆ loadShader()

◆ setAllowedFramesInFlight()

◆ setFramebuffer()

◆ setSwapchainParameters() [1/2]

◆ setSwapchainParameters() [2/2]

◆ setWindow()

◆ supportsPresentMode()

◆ supportsSwapchainComposition()

◆ textureSupportsFormat()

◆ textureSupportsSampleCount()

◆ waitForIdle()

◆ waitForSwapchain()

Field Documentation

◆ data

◆ device

◆ framebuffer

◆ interface

◆ maxAnisotropy

◆ object

◆ shaderFormats

◆ window