public override void CalcToneMap()
{
//toneBeginEvt.Record(mainStream);
DeviceBuffer2D glOutputBuffer = DeviceBuffer2D.GLMapBufferObject(glOutputBufferID, (uint)(4 * xRes));
updateOutputKernel.LaunchAsync(mainStream, glOutputBuffer);
glOutputBuffer.GLUnmapBufferObject(glOutputBufferID);
//toneEndEvt.Record(mainStream);
}
public override void ResetOutput()
{
mainStream.Synchronize();
context.Synchronize();
//resetBeginEvt.Record();
resetIteratorsKernel.Launch();
DeviceBuffer2D glOutputBuffer = DeviceBuffer2D.GLMapBufferObject(glOutputBufferID, (uint)(4 * xRes));
resetOutputKernel.Launch(glOutputBuffer);
glOutputBuffer.GLUnmapBufferObject(glOutputBufferID);
//resetEndEvt.Record();
context.Synchronize();
}
public override void Allocate(int xRes, int yRes)
{
context.Synchronize();
Deallocate();
this.xRes = xRes;
this.yRes = yRes;
xRasterBlockCount = (xRes + RasterBlockSize - 1) / RasterBlockSize;
yRasterBlockCount = (yRes + RasterBlockSize - 1) / RasterBlockSize;
module.WriteConstant("xRes", xRes);
module.WriteConstant("yRes", yRes);
module.WriteConstant("iterBlockCount", IterBlockCount);
accumBuffer = DeviceBuffer2D.Alloc(SubPixelByteSize, xRes * 4, yRes);
module.WriteConstant("accumBuffer", accumBuffer);
GL.GenBuffers(1, out glOutputBufferID);
GL.BindBuffer(BufferTarget.PixelUnpackBuffer, glOutputBufferID);
GL.BufferData(BufferTarget.PixelUnpackBuffer, (IntPtr)(4 * xRes * yRes), IntPtr.Zero, BufferUsageHint.StreamCopy);
GL.BindBuffer(BufferTarget.PixelUnpackBuffer, 0);
context.GLRegisterBufferObject(glOutputBufferID);
glOutputTexID = GL.GenTexture();
GL.BindTexture(TextureTarget.Texture2D, glOutputTexID);
GL.TexImage2D(TextureTarget.Texture2D, 0, PixelInternalFormat.Four,
xRes, yRes, 0, PixelFormat.Rgba, PixelType.UnsignedByte, IntPtr.Zero);
GL.TexParameter(TextureTarget.Texture2D, TextureParameterName.GenerateMipmap, 1);
GL.TexParameter(TextureTarget.Texture2D, TextureParameterName.TextureMinFilter, (int)TextureMinFilter.LinearMipmapLinear);
GL.TexParameter(TextureTarget.Texture2D, TextureParameterName.TextureMagFilter, (int)TextureMagFilter.Linear);
GL.TexParameter(TextureTarget.Texture2D, TextureParameterName.TextureWrapS, (int)TextureWrapMode.Clamp);
GL.TexParameter(TextureTarget.Texture2D, TextureParameterName.TextureWrapT, (int)TextureWrapMode.Clamp);
GL.BindTexture(TextureTarget.Texture2D, 0);
resetOutputKernel.SetBlockShape(RasterBlockSize, RasterBlockSize, 1);
resetOutputKernel.SetGridDim(xRasterBlockCount, yRasterBlockCount);
resetOutputKernel.SetSharedSize(0);
updateOutputKernel.SetBlockShape(RasterBlockSize, RasterBlockSize, 1);
updateOutputKernel.SetGridDim(xRasterBlockCount, yRasterBlockCount);
updateOutputKernel.SetSharedSize(0);
}
public CudaFractalEngine()
{
device = Device.Devices[0];
context = device.CreateContext();
iterBlockCount = Util.Clamp(device.MultiprocessorCount * 2, 2, 64);
//System.Reflection.Assembly loadedAssembly = typeof(CudaFractalEngine).Assembly;
//System.IO.Stream stream = loadedAssembly.GetManifestResourceStream(typeof(CudaFractalEngine), "kernels.ptx");
System.IO.MemoryStream stream = new System.IO.MemoryStream(CudaResources.kernels_ptx);
module = context.LoadModule(stream);
initIteratorsKernel = module.GetKernel("init_iterators_kernel");
resetIteratorsKernel = module.GetKernel("reset_iterators_kernel");
iterateKernel = module.GetKernel("iterate_kernel");
updateStatsKernel = module.GetKernel("update_stats_kernel");
resetOutputKernel = module.GetKernel("reset_output_kernel");
updateOutputKernel = module.GetKernel("update_output_kernel");
glOutputBufferID = 0;
mainStream = new Cuda.Stream();
iterPosStateBuffer = DeviceBuffer2D.Alloc(8, IterBlockSize, IterBlockCount);
module.WriteConstant("iterPosStateBuffer", iterPosStateBuffer);
iterColorStateBuffer = DeviceBuffer2D.Alloc(16, IterBlockSize, IterBlockCount);
module.WriteConstant("iterColorStateBuffer", iterColorStateBuffer);
entropyXBuffer = DeviceBuffer2D.Alloc(16, IterBlockSize, IterBlockCount);
module.WriteConstant("entropyXBuffer", entropyXBuffer);
entropyCBuffer = DeviceBuffer2D.Alloc(4, IterBlockSize, IterBlockCount);
module.WriteConstant("entropyCBuffer", entropyCBuffer);
entropySeedBuffer = DeviceBuffer2D.Alloc(4, IterBlockSize, IterBlockCount);
module.WriteConstant("entropySeedBuffer", entropySeedBuffer);
HostBuffer2D <uint> hostEntropySeedBuffer = HostBuffer2D <uint> .Alloc(IterBlockSize, IterBlockCount);
uint rnd;
for (int y = 0; y < IterBlockCount; y++)
{
for (int x = 0; x < IterBlockSize; x++)
{
rnd = (uint)rand.Next(65536);
hostEntropySeedBuffer[y, x] = rnd;
}
}
CudaMem.Copy(hostEntropySeedBuffer, entropySeedBuffer);
hostEntropySeedBuffer.Free();
dotCountBuffer = DeviceBuffer2D.Alloc(8, IterBlockSize, IterBlockCount);
module.WriteConstant("dotCountBuffer", dotCountBuffer);
peakDensityBuffer = DeviceBuffer2D.Alloc(4, IterBlockSize, IterBlockCount);
module.WriteConstant("peakDensityBuffer", peakDensityBuffer);
totalIterCountMem = DevicePtr.AllocRaw(8);
module.WriteConstant("totalIterCountMem", totalIterCountMem);
totalDotCountMem = DevicePtr.AllocRaw(8);
module.WriteConstant("totalDotCountMem", totalDotCountMem);
densityMem = DevicePtr.AllocRaw(4);
module.WriteConstant("densityMem", densityMem);
peakDensityMem = DevicePtr.AllocRaw(4);
module.WriteConstant("peakDensityMem", peakDensityMem);
scaleConstantMem = DevicePtr.AllocRaw(4);
module.WriteConstant("scaleConstantMem", scaleConstantMem);
paletteImage = CudaArray.Null;
paletteTex = module.GetTexRef("paletteTex");
resetBeginEvt = new Event();
resetEndEvt = new Event();
cycleIterEvt = new Event();
cycleStatEvt = new Event();
cycleEndEvt = new Event();
toneBeginEvt = new Event();
toneEndEvt = new Event();
initIteratorsKernel.SetBlockShape(IterBlockSize, 1, 1);
initIteratorsKernel.SetGridDim(IterBlockCount, 1);
initIteratorsKernel.SetSharedSize(0);
resetIteratorsKernel.SetBlockShape(IterBlockSize, 1, 1);
resetIteratorsKernel.SetGridDim(IterBlockCount, 1);
resetIteratorsKernel.SetSharedSize(0);
iterateKernel.SetBlockShape(IterBlockSize, 1, 1);
iterateKernel.SetGridDim(IterBlockCount, 1);
iterateKernel.SetSharedSize(0);
updateStatsKernel.SetBlockShape(1, 1, 1);
updateStatsKernel.SetGridDim(1, 1);
updateStatsKernel.SetSharedSize(0);
initIteratorsKernel.Launch();
context.Synchronize();
}