Sparse Resources
Vulkan sparse resources are a way to create VkBuffer
and VkImage
objects which can be bound non-contiguously to one or more VkDeviceMemory
allocations. There are many aspects and features of sparse resources which the spec does a good job explaining. As the implementation guidelines point out, most implementations use sparse resources to expose a linear virtual address range of memory to the application while mapping each sparse block to physical pages when bound.
Binding Sparse Memory
Unlike normal resources that call vkBindBufferMemory()
or vkBindImageMemory()
, sparse memory is bound via a queue operation vkQueueBindSparse()
. The main advantage of this is that an application can rebind memory to a sparse resource throughout its lifetime.
It is important to notice that this requires some extra consideration from the application. Applications must use synchronization primitives to guarantee that other queues do not access ranges of memory concurrently with a binding change. Also, freeing a VkDeviceMemory
object with vkFreeMemory()
will not cause resources (or resource regions) bound to the memory object to become unbound. Applications must not access resources bound to memory that has been freed.
Sparse Buffers
The following example is used to help visually showcase how a sparse VkBuffer
looks in memory. Note, it is not required, but most implementations will use sparse block sizes of 64 KB for VkBuffer
(actual size is returned in VkMemoryRequirements::alignment
).
Imagine a 256 KB VkBuffer
where there are 3 parts that an application wants to update separately.
-
Section A - 64 KB
-
Section B - 128 KB
-
Section C - 64 KB
The following showcases how the application views the VkBuffer
:
Sparse Images
Mip Tail Regions
Sparse images can be used to update mip levels separately which results in a mip tail region. The spec describes the various examples that can occur with diagrams.
Basic Sparse Resources Example
The following examples illustrate basic creation of sparse images and binding them to physical memory.
This basic example creates a normal VkImage
object but uses fine-grained memory allocation to back the resource with multiple memory ranges.
VkDevice device;
VkQueue queue;
VkImage sparseImage;
VkAllocationCallbacks* pAllocator = NULL;
VkMemoryRequirements memoryRequirements = {};
VkDeviceSize offset = 0;
VkSparseMemoryBind binds[MAX_CHUNKS] = {}; // MAX_CHUNKS is NOT part of Vulkan
uint32_t bindCount = 0;
// ...
// Allocate image object
const VkImageCreateInfo sparseImageInfo =
{
VK_STRUCTURE_TYPE_IMAGE_CREATE_INFO, // sType
NULL, // pNext
VK_IMAGE_CREATE_SPARSE_BINDING_BIT | ..., // flags
...
};
vkCreateImage(device, &sparseImageInfo, pAllocator, &sparseImage);
// Get memory requirements
vkGetImageMemoryRequirements(
device,
sparseImage,
&memoryRequirements);
// Bind memory in fine-grained fashion, find available memory ranges
// from potentially multiple VkDeviceMemory pools.
// (Illustration purposes only, can be optimized for perf)
while (memoryRequirements.size && bindCount < MAX_CHUNKS)
{
VkSparseMemoryBind* pBind = &binds[bindCount];
pBind->resourceOffset = offset;
AllocateOrGetMemoryRange(
device,
&memoryRequirements,
&pBind->memory,
&pBind->memoryOffset,
&pBind->size);
// memory ranges must be sized as multiples of the alignment
assert(IsMultiple(pBind->size, memoryRequirements.alignment));
assert(IsMultiple(pBind->memoryOffset, memoryRequirements.alignment));
memoryRequirements.size -= pBind->size;
offset += pBind->size;
bindCount++;
}
// Ensure entire image has backing
if (memoryRequirements.size)
{
// Error condition - too many chunks
}
const VkSparseImageOpaqueMemoryBindInfo opaqueBindInfo =
{
sparseImage, // image
bindCount, // bindCount
binds // pBinds
};
const VkBindSparseInfo bindSparseInfo =
{
VK_STRUCTURE_TYPE_BIND_SPARSE_INFO, // sType
NULL, // pNext
...
1, // imageOpaqueBindCount
&opaqueBindInfo, // pImageOpaqueBinds
...
};
// vkQueueBindSparse is externally synchronized per queue object.
AcquireQueueOwnership(queue);
// Actually bind memory
vkQueueBindSparse(queue, 1, &bindSparseInfo, VK_NULL_HANDLE);
ReleaseQueueOwnership(queue);
Advanced Sparse Resources
This more advanced example creates an arrayed color attachment / texture image and binds only LOD zero and the required metadata to physical memory.
VkDevice device;
VkQueue queue;
VkImage sparseImage;
VkAllocationCallbacks* pAllocator = NULL;
VkMemoryRequirements memoryRequirements = {};
uint32_t sparseRequirementsCount = 0;
VkSparseImageMemoryRequirements* pSparseReqs = NULL;
VkSparseMemoryBind binds[MY_IMAGE_ARRAY_SIZE] = {};
VkSparseImageMemoryBind imageBinds[MY_IMAGE_ARRAY_SIZE] = {};
uint32_t bindCount = 0;
// Allocate image object (both renderable and sampleable)
const VkImageCreateInfo sparseImageInfo =
{
VK_STRUCTURE_TYPE_IMAGE_CREATE_INFO, // sType
NULL, // pNext
VK_IMAGE_CREATE_SPARSE_RESIDENCY_BIT | ..., // flags
...
VK_FORMAT_R8G8B8A8_UNORM, // format
...
MY_IMAGE_ARRAY_SIZE, // arrayLayers
...
VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT |
VK_IMAGE_USAGE_SAMPLED_BIT, // usage
...
};
vkCreateImage(device, &sparseImageInfo, pAllocator, &sparseImage);
// Get memory requirements
vkGetImageMemoryRequirements(
device,
sparseImage,
&memoryRequirements);
// Get sparse image aspect properties
vkGetImageSparseMemoryRequirements(
device,
sparseImage,
&sparseRequirementsCount,
NULL);
pSparseReqs = (VkSparseImageMemoryRequirements*)
malloc(sparseRequirementsCount * sizeof(VkSparseImageMemoryRequirements));
vkGetImageSparseMemoryRequirements(
device,
sparseImage,
&sparseRequirementsCount,
pSparseReqs);
// Bind LOD level 0 and any required metadata to memory
for (uint32_t i = 0; i < sparseRequirementsCount; ++i)
{
if (pSparseReqs[i].formatProperties.aspectMask &
VK_IMAGE_ASPECT_METADATA_BIT)
{
// Metadata must not be combined with other aspects
assert(pSparseReqs[i].formatProperties.aspectMask ==
VK_IMAGE_ASPECT_METADATA_BIT);
if (pSparseReqs[i].formatProperties.flags &
VK_SPARSE_IMAGE_FORMAT_SINGLE_MIPTAIL_BIT)
{
VkSparseMemoryBind* pBind = &binds[bindCount];
pBind->memorySize = pSparseReqs[i].imageMipTailSize;
bindCount++;
// ... Allocate memory range
pBind->resourceOffset = pSparseReqs[i].imageMipTailOffset;
pBind->memoryOffset = /* allocated memoryOffset */;
pBind->memory = /* allocated memory */;
pBind->flags = VK_SPARSE_MEMORY_BIND_METADATA_BIT;
}
else
{
// Need a mip tail region per array layer.
for (uint32_t a = 0; a < sparseImageInfo.arrayLayers; ++a)
{
VkSparseMemoryBind* pBind = &binds[bindCount];
pBind->memorySize = pSparseReqs[i].imageMipTailSize;
bindCount++;
// ... Allocate memory range
pBind->resourceOffset = pSparseReqs[i].imageMipTailOffset +
(a * pSparseReqs[i].imageMipTailStride);
pBind->memoryOffset = /* allocated memoryOffset */;
pBind->memory = /* allocated memory */
pBind->flags = VK_SPARSE_MEMORY_BIND_METADATA_BIT;
}
}
}
else
{
// resource data
VkExtent3D lod0BlockSize =
{
AlignedDivide(
sparseImageInfo.extent.width,
pSparseReqs[i].formatProperties.imageGranularity.width);
AlignedDivide(
sparseImageInfo.extent.height,
pSparseReqs[i].formatProperties.imageGranularity.height);
AlignedDivide(
sparseImageInfo.extent.depth,
pSparseReqs[i].formatProperties.imageGranularity.depth);
}
size_t totalBlocks =
lod0BlockSize.width *
lod0BlockSize.height *
lod0BlockSize.depth;
// Each block is the same size as the alignment requirement,
// calculate total memory size for level 0
VkDeviceSize lod0MemSize = totalBlocks * memoryRequirements.alignment;
// Allocate memory for each array layer
for (uint32_t a = 0; a < sparseImageInfo.arrayLayers; ++a)
{
// ... Allocate memory range
VkSparseImageMemoryBind* pBind = &imageBinds[a];
pBind->subresource.aspectMask = pSparseReqs[i].formatProperties.aspectMask;
pBind->subresource.mipLevel = 0;
pBind->subresource.arrayLayer = a;
pBind->offset = (VkOffset3D){0, 0, 0};
pBind->extent = sparseImageInfo.extent;
pBind->memoryOffset = /* allocated memoryOffset */;
pBind->memory = /* allocated memory */;
pBind->flags = 0;
}
}
free(pSparseReqs);
}
const VkSparseImageOpaqueMemoryBindInfo opaqueBindInfo =
{
sparseImage, // image
bindCount, // bindCount
binds // pBinds
};
const VkSparseImageMemoryBindInfo imageBindInfo =
{
sparseImage, // image
sparseImageInfo.arrayLayers, // bindCount
imageBinds // pBinds
};
const VkBindSparseInfo bindSparseInfo =
{
VK_STRUCTURE_TYPE_BIND_SPARSE_INFO, // sType
NULL, // pNext
...
1, // imageOpaqueBindCount
&opaqueBindInfo, // pImageOpaqueBinds
1, // imageBindCount
&imageBindInfo, // pImageBinds
...
};
// vkQueueBindSparse is externally synchronized per queue object.
AcquireQueueOwnership(queue);
// Actually bind memory
vkQueueBindSparse(queue, 1, &bindSparseInfo, VK_NULL_HANDLE);
ReleaseQueueOwnership(queue);