public MgTextureInfo Load(byte[] imageData, MgImageSource source, IMgAllocationCallbacks allocator, IMgFence fence)
{
// Prefer using optimal tiling, as linear tiling
// may support only a small set of features
// depending on implementation (e.g. no mip maps, only one layer, etc.)
IMgImage mappableImage;
IMgDeviceMemory mappableMemory;
uint mipLevels = (uint)source.Mipmaps.Length;
// Load mip map level 0 to linear tiling image
var imageCreateInfo = new MgImageCreateInfo
{
ImageType = MgImageType.TYPE_2D,
Format = source.Format,
MipLevels = mipLevels,
ArrayLayers = 1,
Samples = MgSampleCountFlagBits.COUNT_1_BIT,
Tiling = MgImageTiling.LINEAR,
Usage = MgImageUsageFlagBits.SAMPLED_BIT,
SharingMode = MgSharingMode.EXCLUSIVE,
InitialLayout = MgImageLayout.PREINITIALIZED,
Extent = new MgExtent3D {
Width = source.Width,
Height = source.Height,
Depth = 1
},
};
var device = mGraphicsConfiguration.Device;
var err = device.CreateImage(imageCreateInfo, allocator, out mappableImage);
Debug.Assert(err == Result.SUCCESS, err + " != Result.SUCCESS");
// Get memory requirements for this image
// like size and alignment
MgMemoryRequirements memReqs;
device.GetImageMemoryRequirements(mappableImage, out memReqs);
// Set memory allocation size to required memory size
var memAllocInfo = new MgMemoryAllocateInfo
{
AllocationSize = memReqs.Size,
};
Debug.Assert(mGraphicsConfiguration.Partition != null);
// Get memory type that can be mapped to host memory
uint memoryTypeIndex;
bool isValid = mGraphicsConfiguration.Partition.GetMemoryType(memReqs.MemoryTypeBits, MgMemoryPropertyFlagBits.HOST_VISIBLE_BIT, out memoryTypeIndex);
Debug.Assert(isValid);
memAllocInfo.MemoryTypeIndex = memoryTypeIndex;
// Allocate host memory
err = device.AllocateMemory(memAllocInfo, allocator, out mappableMemory);
Debug.Assert(err == Result.SUCCESS, err + " != Result.SUCCESS");
// Bind allocated image for use
err = mappableImage.BindImageMemory(device, mappableMemory, 0);
Debug.Assert(err == Result.SUCCESS, err + " != Result.SUCCESS");
// Get sub resource layout
// Mip map count, array layer, etc.
var subRes = new MgImageSubresource
{
AspectMask = MgImageAspectFlagBits.COLOR_BIT,
};
MgSubresourceLayout subResLayout;
IntPtr data;
// Get sub resources layout
// Includes row pitch, size offsets, etc.
device.GetImageSubresourceLayout(mappableImage, subRes, out subResLayout);
// Map image memory
err = mappableMemory.MapMemory(device, 0, memReqs.Size, 0, out data);
Debug.Assert(err == Result.SUCCESS, err + " != Result.SUCCESS");
// Copy image data into memory
//memcpy(data, tex2D[subRes.mipLevel].data(), tex2D[subRes.mipLevel].size());
Marshal.Copy(imageData, 0, data, (int)source.Size);
mappableMemory.UnmapMemory(device);
// Linear tiled images don't need to be staged
// and can be directly used as textures
var texture = new MgTextureInfo {
Image = mappableImage,
DeviceMemory = mappableMemory,
ImageLayout = MgImageLayout.SHADER_READ_ONLY_OPTIMAL,
};
var cmdPool = mGraphicsConfiguration.Partition.CommandPool;
var cmdBufAllocateInfo = new MgCommandBufferAllocateInfo
{
CommandPool = cmdPool,
Level = MgCommandBufferLevel.PRIMARY,
CommandBufferCount = 1,
};
var commands = new IMgCommandBuffer[1];
err = device.AllocateCommandBuffers(cmdBufAllocateInfo, commands);
Debug.Assert(err == Result.SUCCESS, err + " != Result.SUCCESS");
var cmdBufInfo = new MgCommandBufferBeginInfo
{
Flags = 0,
};
texture.Command = commands [0];
err = commands[0].BeginCommandBuffer(cmdBufInfo);
Debug.Assert(err == Result.SUCCESS, err + " != Result.SUCCESS");
// Setup image memory barrier transfer image to shader read layout
mImageTools.SetImageLayout(
texture.Command,
texture.Image,
MgImageAspectFlagBits.COLOR_BIT,
MgImageLayout.PREINITIALIZED,
texture.ImageLayout,
0,
mipLevels);
err = texture.Command.EndCommandBuffer();
Debug.Assert(err == Result.SUCCESS, err + " != Result.SUCCESS");
var submitInfo = new MgSubmitInfo {
CommandBuffers = commands,
};
var queue = mGraphicsConfiguration.Queue;
queue.QueueSubmit(new[] { submitInfo }, fence);
// // NOT SURE IF
// err = queue.QueueWaitIdle();
// Debug.Assert (err == Result.SUCCESS, err + " != Result.SUCCESS");
//
// device.FreeCommandBuffers(cmdPool, commands);
// texture.Command = copyCmd;
return(texture);
}
// FROM texture.cpp (2016) Sascha Williams
public MgTextureInfo Load(byte[] imageData, MgImageSource source, IMgAllocationCallbacks allocator, IMgFence fence)
{
var device = mGraphicsConfiguration.Device;
var queue = mGraphicsConfiguration.Queue;
var cmdPool = mGraphicsConfiguration.Partition.CommandPool;
uint mipLevels = (uint)source.Mipmaps.Length;
// Create a host-visible staging buffer that contains the raw image data
IMgBuffer stagingBuffer;
IMgDeviceMemory stagingMemory;
var bufferCreateInfo = new MgBufferCreateInfo {
Size = source.Size,
Usage = MgBufferUsageFlagBits.TRANSFER_SRC_BIT,
SharingMode = MgSharingMode.EXCLUSIVE,
};
// This buffer is used as a transfer source for the buffer copy
var result = device.CreateBuffer(bufferCreateInfo, allocator, out stagingBuffer);
Debug.Assert(result == Result.SUCCESS);
MgMemoryRequirements memReqs;
// Get memory requirements for the staging buffer (alignment, memory type bits)
mGraphicsConfiguration.Device.GetBufferMemoryRequirements(stagingBuffer, out memReqs);
var memAllocInfo = new MgMemoryAllocateInfo {
AllocationSize = memReqs.Size,
};
// Get memory type index for a host visible buffer
uint memoryTypeIndex;
bool isTypeValid = mGraphicsConfiguration.Partition.GetMemoryType(memReqs.MemoryTypeBits, MgMemoryPropertyFlagBits.HOST_VISIBLE_BIT, out memoryTypeIndex);
Debug.Assert(isTypeValid);
memAllocInfo.MemoryTypeIndex = memoryTypeIndex;
result = device.AllocateMemory(memAllocInfo, allocator, out stagingMemory);
Debug.Assert(result == Result.SUCCESS);
result = stagingBuffer.BindBufferMemory(device, stagingMemory, 0);
Debug.Assert(result == Result.SUCCESS);
// Copy texture data into staging buffer
IntPtr data;
result = stagingMemory.MapMemory(device, 0, memReqs.Size, 0, out data);
Debug.Assert(result == Result.SUCCESS);
// TODO : Copy here
Marshal.Copy(imageData, 0, data, (int)source.Size);
stagingMemory.UnmapMemory(device);
// Setup buffer copy regions for each mip level
var bufferCopyRegions = new MgBufferImageCopy[source.Mipmaps.Length];
for (uint i = 0; i < bufferCopyRegions.Length; ++i)
{
bufferCopyRegions [i] = new MgBufferImageCopy {
ImageSubresource = new MgImageSubresourceLayers {
AspectMask = MgImageAspectFlagBits.COLOR_BIT,
MipLevel = i,
BaseArrayLayer = 0,
LayerCount = 1,
},
ImageExtent = new MgExtent3D {
Width = source.Mipmaps[i].Width,
Height = source.Mipmaps[i].Height,
Depth = 1,
},
BufferOffset = source.Mipmaps[i].Offset,
};
}
// Create optimal tiled target image
var imageCreateInfo = new MgImageCreateInfo
{
ImageType = MgImageType.TYPE_2D,
Format = source.Format,
MipLevels = mipLevels,
ArrayLayers = 1,
Samples = MgSampleCountFlagBits.COUNT_1_BIT,
Tiling = MgImageTiling.OPTIMAL,
SharingMode = MgSharingMode.EXCLUSIVE,
InitialLayout = MgImageLayout.PREINITIALIZED,
Extent = new MgExtent3D
{
Width = source.Width,
Height = source.Height,
Depth = 1
},
Usage = MgImageUsageFlagBits.TRANSFER_DST_BIT | MgImageUsageFlagBits.SAMPLED_BIT,
};
var texture = new MgTextureInfo();
IMgImage image;
result = device.CreateImage(imageCreateInfo, allocator, out image);
Debug.Assert(result == Result.SUCCESS);
texture.Image = image;
device.GetImageMemoryRequirements(texture.Image, out memReqs);
memAllocInfo.AllocationSize = memReqs.Size;
isTypeValid = mGraphicsConfiguration.Partition.GetMemoryType(memReqs.MemoryTypeBits, MgMemoryPropertyFlagBits.DEVICE_LOCAL_BIT, out memoryTypeIndex);
Debug.Assert(isTypeValid);
memAllocInfo.MemoryTypeIndex = memoryTypeIndex;
IMgDeviceMemory deviceMem;
result = device.AllocateMemory(memAllocInfo, allocator, out deviceMem);
Debug.Assert(result == Result.SUCCESS);
texture.DeviceMemory = deviceMem;
result = texture.Image.BindImageMemory(device, texture.DeviceMemory, 0);
Debug.Assert(result == Result.SUCCESS);
var cmdBufAllocateInfo = new MgCommandBufferAllocateInfo
{
CommandPool = cmdPool,
Level = MgCommandBufferLevel.PRIMARY,
CommandBufferCount = 1,
};
var commands = new IMgCommandBuffer[1];
result = device.AllocateCommandBuffers(cmdBufAllocateInfo, commands);
Debug.Assert(result == Result.SUCCESS);
var cmdBufInfo = new MgCommandBufferBeginInfo
{
Flags = 0,
};
texture.Command = commands [0];
result = commands[0].BeginCommandBuffer(cmdBufInfo);
Debug.Assert(result == Result.SUCCESS);
// Image barrier for optimal image (target)
// Optimal image will be used as destination for the copy
mImageTools.SetImageLayout(
texture.Command,
texture.Image,
MgImageAspectFlagBits.COLOR_BIT,
MgImageLayout.PREINITIALIZED,
MgImageLayout.TRANSFER_DST_OPTIMAL,
0,
mipLevels);
// Copy mip levels from staging buffer
texture.Command.CmdCopyBufferToImage(
stagingBuffer,
texture.Image,
MgImageLayout.TRANSFER_DST_OPTIMAL,
bufferCopyRegions
);
// Change texture image layout to shader read after all mip levels have been copied
texture.ImageLayout = MgImageLayout.SHADER_READ_ONLY_OPTIMAL;
mImageTools.SetImageLayout(
texture.Command,
texture.Image,
MgImageAspectFlagBits.COLOR_BIT,
MgImageLayout.TRANSFER_DST_OPTIMAL,
texture.ImageLayout,
0,
mipLevels);
result = texture.Command.EndCommandBuffer();
Debug.Assert(result == Result.SUCCESS);
var submitInfo = new MgSubmitInfo {
CommandBuffers = commands,
};
queue.QueueSubmit(new[] { submitInfo }, fence);
// result = queue.QueueWaitIdle();
// Debug.Assert (result == Result.SUCCESS);
//
// device.FreeCommandBuffers(cmdPool, commands);
// texture.Command = copyCmd;
// Clean up staging resources
stagingMemory.FreeMemory(device, allocator);
stagingBuffer.DestroyBuffer(device, allocator);
return(texture);
}