C# CudaDeviceVariable类代码示例

原作者: [db:作者] 来自: [db:来源] 收藏邀请

本文整理汇总了C#中CudaDeviceVariable类的典型用法代码示例。如果您正苦于以下问题：C# CudaDeviceVariable类的具体用法？C# CudaDeviceVariable怎么用？C# CudaDeviceVariable使用的例子？那么恭喜您, 这里精选的类代码示例或许可以为您提供帮助。

CudaDeviceVariable类属于命名空间，在下文中一共展示了CudaDeviceVariable类的20个代码示例，这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞，您的评价将有助于我们的系统推荐出更棒的C#代码示例。

示例1: DCTQuantFwd8x8LS

 /// <summary>
 /// Forward DCT, quantization and level shift part of the JPEG encoding.
 /// Input is expected in 8x8 macro blocks and output is expected to be in 64x1
 /// macro blocks.
 /// </summary>
 /// <param name="src">Source image.</param>
 /// <param name="dst">Destination image</param>
 /// <param name="QuantFwdTable">Forward quantization tables for JPEG encoding created using QuantInvTableInit()</param>
 /// <param name="oSizeRoi">Roi size (in macro blocks?).</param>
 public static void DCTQuantFwd8x8LS(NPPImage_8uC1 src, NPPImage_16sC1 dst, CudaDeviceVariable<ushort> QuantFwdTable, NppiSize oSizeRoi)
 {
     NppStatus status;
     status = NPPNativeMethods.NPPi.ImageCompression.nppiDCTQuantFwd8x8LS_JPEG_8u16s_C1R(src.DevicePointer, src.Pitch, dst.DevicePointer, dst.Pitch, QuantFwdTable.DevicePointer, oSizeRoi);
     Debug.WriteLine(String.Format("{0:G}, {1}: {2}", DateTime.Now, "nppiDCTQuantFwd8x8LS_JPEG_8u16s_C1R", status));
     NPPException.CheckNppStatus(status, null);
 }

开发者ID:furusdara，项目名称:cuda，代码行数:16，代码来源:JPEGCompression.cs

示例2: Init

        protected override void Init()
        {
            var kernelFileName = KernelFile;
            var initKernel = Ctx.LoadKernel(kernelFileName, "generateData");
            B = new CudaDeviceVariable<double>(DimensionsCount);
            M = new CudaDeviceVariable<double>(DimensionsCount * DimensionsCount);
            Xopt = new CudaDeviceVariable<double>(DimensionsCount);

            var d_fopt = new CudaDeviceVariable<double>(1);

            long rseed = FunctionNumber + 10000 * InstanceNumber;

            initKernel.Run(
                DimensionsCount,
                rseed,
                FunctionNumber,
                InstanceNumber,
                M.DevicePointer,
                B.DevicePointer,
                Xopt.DevicePointer,
                d_fopt.DevicePointer);

            double[] fopt_arr = d_fopt;

            Fopt = fopt_arr[0];
        }

开发者ID:trojkac，项目名称:effective_pso，代码行数:26，代码来源:GriewankRosenbrockAlgorithm.cs

示例3: CudaIntersectionDevice

        public CudaIntersectionDevice(RayEngineScene scene, NVContext ctx)
            : base(scene)
        {
            wallclock = new Stopwatch();
            this.todoRayBuffers = new ConcurrentQueue<Tuple<int, RayBuffer>>();
            this.doneRayBuffers = new List<ConcurrentQueue<RayBuffer>>() { { new ConcurrentQueue<RayBuffer>() } };
            this.started = false;
            if (ctx != null)
            {
                this.cudaContext = ctx;
            }
            else
            {
                this.cudaContext = new NVContext() { Context = new CudaContext(CudaContext.GetMaxGflopsDeviceId()) };
            }
            using (var sr = new StreamReader(@"G:\Git\RayDen\CudaMegaRay\x64\Release\kernel.cu.ptx"))
            {
                intersectKernel = cudaContext.Context.LoadKernelPTX(sr.BaseStream, "IntersectLBvh");
            }

            this.rays = new CudaDeviceVariable<RayData>(RayBuffer.RayBufferSize);
            this.hits = new CudaDeviceVariable<RayHit>(RayBuffer.RayBufferSize);
            verts = scene.Vertices.ToArray();
            //scene.Triangles.Select(i => i.GetInfo()).ToArray();

            var ti = scene.Triangles.Select(i => i.GetInfo()).ToArray();
            var da = new BvhDataAdapter(scene);
            var treeData = da.GetMpData();
            bvh = treeData;
            trianglesCount = ti.Length;
            tris = ti; 

            nodesCount = treeData.Length;
            Tracer.TraceLine("BVH Data Size {0:F3} MBytes", (treeData.Length * 32f) / (1024f * 1024f));
        }

开发者ID:HungryBear，项目名称:rayden，代码行数:35，代码来源:CudaIntersectionDevice.cs

示例4: Init

        protected override void Init()
        {
            var kernelFileName = KernelFile;
            var initKernel = Ctx.LoadKernel(kernelFileName, "generateData");
            Xopt = new CudaDeviceVariable<double>(DimensionsCount);

            var d_fopt = new CudaDeviceVariable<double>(1);

            long rseed = FunctionNumber + 10000 * InstanceNumber;

            double[] host_minus_ones = new double[DimensionsCount];

            for (int i = 0; i < DimensionsCount; i++)
            {
                host_minus_ones[i] = -1;
            }

            MinusOnes = host_minus_ones;

            Factor = Math.Max(1.0, Math.Sqrt(DimensionsCount)/8.0);

            initKernel.Run(
                DimensionsCount,
                rseed,
                FunctionNumber,
                InstanceNumber,
                Xopt.DevicePointer,
                d_fopt.DevicePointer);

            double[] fopt_arr = d_fopt;

            Fopt = fopt_arr[0];
        }

开发者ID:trojkac，项目名称:effective_pso，代码行数:33，代码来源:RosenbrockAlgorithm.cs

示例5: Backward

        public void Backward(CudnnPoolingDescriptor pooling, CudnnTensorDescriptor srcTensor, float[] srcData, CudnnTensorDescriptor srcDiffTensor, float[] srcDiffData,
                                                             CudnnTensorDescriptor destTensor, float[] destData, CudnnTensorDescriptor destDiffTensor, float[] destDiffData)
        {
            Contract.Requires(pooling != null);
            Contract.Requires(srcTensor != null);
            Contract.Requires(srcData != null);
            Contract.Requires(destTensor != null);
            Contract.Requires(destData != null);
            Contract.Requires(srcDiffTensor != null);
            Contract.Requires(srcDiffData != null);
            Contract.Requires(destDiffTensor != null);
            Contract.Requires(destDiffData != null);

            ThrowIfNotInitialized();
            CheckIfCompatible(CudnnType.Float, srcTensor, srcDiffTensor, destTensor, destDiffTensor);

            using (var srcDataGpu = new CudaDeviceVariable<float>(srcData.Length))
            using (var srcDiffDataGpu = new CudaDeviceVariable<float>(srcDiffData.Length))
            using (var destDataGpu = new CudaDeviceVariable<float>(destData.Length))
            using (var destDiffDataGpu = new CudaDeviceVariable<float>(destDiffData.Length))
            {
                srcDataGpu.CopyToDevice(srcData);
                srcDiffDataGpu.CopyToDevice(srcDiffData);
                destDataGpu.CopyToDevice(destData);

                Invoke(() => CudnnNativeMethods.cudnnPoolingBackward(handle, pooling.Handle,
                                                                     srcTensor.Handle, srcDataGpu.DevicePointer, srcDiffTensor.Handle, srcDiffDataGpu.DevicePointer,
                                                                     destTensor.Handle, destDataGpu.DevicePointer, destDiffTensor.Handle, destDiffDataGpu.DevicePointer));
                destDiffDataGpu.CopyToHost(destDiffData);
            }
        }

开发者ID:redknightlois，项目名称:cudalearn，代码行数:31，代码来源:CudnnContext.Pooling.cs

示例6: launchConfig

        ////////////////////////////////////////////////////////////////////////////////
        // Occupancy-based launch configurator
        //
        // The launch configurator, cudaOccupancyMaxPotentialBlockSize and
        // cudaOccupancyMaxPotentialBlockSizeVariableSMem, suggests a block
        // size that achieves the best theoretical occupancy. It also returns
        // the minimum number of blocks needed to achieve the occupancy on the
        // whole device.
        //
        // This launch configurator is purely occupancy-based. It doesn't
        // translate directly to performance, but the suggestion should
        // nevertheless be a good starting point for further optimizations.
        //
        // This function configures the launch based on the "automatic"
        // argument, records the runtime, and reports occupancy and runtime.
        ////////////////////////////////////////////////////////////////////////////////
        static int launchConfig(CudaDeviceVariable<int> array, int arrayCount, bool automatic)
        {
            int blockSize = 0;
            int minGridSize = 0;
            int gridSize;
            SizeT dynamicSMemUsage = 0;

            float elapsedTime;

            double potentialOccupancy;

            CudaOccupancy.cudaOccDeviceState state = new CudaOccupancy.cudaOccDeviceState();
            state.cacheConfig = CudaOccupancy.cudaOccCacheConfig.PreferNone;

            if (automatic)
            {
                CudaOccupancy.cudaOccMaxPotentialOccupancyBlockSize(ref minGridSize, ref blockSize, new CudaOccupancy.cudaOccDeviceProp(0), new CudaOccupancy.cudaOccFuncAttributes(kernel), state, dynamicSMemUsage);

                Console.WriteLine("Suggested block size: {0}", blockSize);
                Console.WriteLine("Minimum grid size for maximum occupancy: {0}", minGridSize);
            }
            else
            {
                // This block size is too small. Given limited number of
                // active blocks per multiprocessor, the number of active
                // threads will be limited, and thus unable to achieve maximum
                // occupancy.
                //
                blockSize = manualBlockSize;
            }

            // Round up
            //
            gridSize = (arrayCount + blockSize - 1) / blockSize;

            // Launch and profile
            //
            kernel.GridDimensions = gridSize;
            kernel.BlockDimensions = blockSize;
            elapsedTime = kernel.Run(array.DevicePointer, arrayCount);

            // Calculate occupancy
            //
            potentialOccupancy = reportPotentialOccupancy(blockSize, dynamicSMemUsage);

            Console.WriteLine("Potential occupancy: {0}%", potentialOccupancy * 100);

            // Report elapsed time
            //
            Console.WriteLine("Elapsed time: {0}ms", elapsedTime * 100);

            return 0;
        }

开发者ID:kunzmi，项目名称:managedCuda，代码行数:69，代码来源:Program.cs

示例7: BackwardBias

        public void BackwardBias(CudnnTensorDescriptor srcTensor, CudaDeviceVariable<double> srcData, CudnnTensorDescriptor destTensor, CudaDeviceVariable<double> destData, CudnnAccumulateResult accumulate)
        {
            Contract.Requires(srcTensor != null);
            Contract.Requires(srcData != null);
            Contract.Requires(destTensor != null);
            Contract.Requires(destData != null);

            ThrowIfNotInitialized();
            CheckIfCompatible(CudnnType.Double, srcTensor, destTensor);

            Invoke(() => CudnnNativeMethods.cudnnConvolutionBackwardBias(handle, srcTensor.Handle, srcData.DevicePointer, destTensor.Handle, destData.DevicePointer, accumulate));
        }

开发者ID:redknightlois，项目名称:cudalearn，代码行数:12，代码来源:CudnnContext.Convolution.cs

示例8: BackwardData

        public void BackwardData(CudnnFilterDescriptor filter, CudaDeviceVariable<float> filterData, CudnnTensorDescriptor diffTensor, CudaDeviceVariable<float> diffData, CudnnConvolutionDescriptor convolution, CudnnTensorDescriptor gradient, CudaDeviceVariable<float> gradientData, CudnnAccumulateResult accumulate)
        {
            Contract.Requires(filter != null);
            Contract.Requires(filterData != null);
            Contract.Requires(diffTensor != null);
            Contract.Requires(diffData != null);
            Contract.Requires(convolution != null);
            Contract.Requires(gradient != null);
            Contract.Requires(gradientData != null);

            ThrowIfNotInitialized();
            CheckIfCompatible(CudnnType.Float, filter, diffTensor, gradient);

            Invoke(() => CudnnNativeMethods.cudnnConvolutionBackwardData(handle, filter.Handle, filterData.DevicePointer, diffTensor.Handle, diffData.DevicePointer, convolution.Handle, gradient.Handle, gradientData.DevicePointer, accumulate));
        }

开发者ID:redknightlois，项目名称:cudalearn，代码行数:15，代码来源:CudnnContext.Convolution.cs

示例9: GraphCut4

		/// <summary>
		/// 
		/// </summary>
		/// <param name="size">Graph size</param>
		public GraphCut4(NppiSize size)
		{
			_size = size;
			int bufferSize = 0;
			status = NPPNativeMethods.NPPi.ImageLabeling.nppiGraphcutGetSize(_size, ref bufferSize);
			Debug.WriteLine(String.Format("{0:G}, {1}: {2}", DateTime.Now, "nppiGraphcutGetSize", status));
			NPPException.CheckNppStatus(status, this);

			_buffer = new CudaDeviceVariable<byte>(bufferSize);

			_state = new NppiGraphcutState();
			status = NPPNativeMethods.NPPi.ImageLabeling.nppiGraphcutInitAlloc(_size, ref _state, _buffer.DevicePointer);
			Debug.WriteLine(String.Format("{0:G}, {1}: {2}", DateTime.Now, "nppiGraphcutInitAlloc", status));
			NPPException.CheckNppStatus(status, this);
		}

开发者ID:lvaleriu，项目名称:managedCuda，代码行数:19，代码来源:GraphCut.cs

示例10: Run

        public void Run(MatOperation operation, CudaDeviceVariable<float> A, int ACount, int AColumnHint, CudaDeviceVariable<float> B, int BCount, int BColumnHint, CudaDeviceVariable<float> Result, int ResultCount, int ResultColumnHint, float beta = 1.0f)
        {
            Result.Memset(BitConverter.ToUInt32(BitConverter.GetBytes(0.0f), 0));

            switch (operation)
            {
                case MatOperation.Multiplication:  // vectors/matrices have to be always in the correct dimesions!
                    if (BCount > 1 && ACount > 1 && BColumnHint == 1 && ACount / AColumnHint > 1 && BCount / BColumnHint == AColumnHint) //. A*vecB
                    {
                        MyCublasFactory.Instance.Gemv(Operation.Transpose,  // transpose beacuase it does Ax row wise if x is a row vector :D
                            AColumnHint, ACount / AColumnHint, 1.0f,
                            A, AColumnHint,
                            B, 1,
                            beta, Result, 1);
                    }
                    else if (ACount > 1 && BCount > 1 && ACount / AColumnHint == 1 && BColumnHint > 1 && BCount / BColumnHint == AColumnHint)  // vecA*B
                    {
                        MyCublasFactory.Instance.Gemv(Operation.NonTranspose,  // transpose beacuase it does Ax row wise if x is a row vector :D
                            BColumnHint, BCount / BColumnHint, 1.0f,
                            B, BColumnHint,
                            A, 1,
                            beta, Result, 1);
                    }
                    else if (ACount / AColumnHint == 1 && BColumnHint == 1 && ACount > 1 && BCount > 1) //. trans(vecA) * vecB
                    {
                        Run(MatOperation.DotProd, A, ACount, AColumnHint, B, BCount, BColumnHint, Result, ResultCount, ResultColumnHint, beta);
                    }
                    else if (ACount != 1 || BCount != 1)// A*B   matrix multiplication
                    {
                        MyCublasFactory.Instance.Gemm(Operation.NonTranspose, Operation.NonTranspose,
                            ACount / AColumnHint, BColumnHint, AColumnHint, 1.0f,
                            A, ACount / AColumnHint,
                            B, BCount / BColumnHint,
                            beta, Result, ResultColumnHint);
                    }
                    break;
                case MatOperation.DotProd:
                    MyCublasFactory.Instance.Gemv(Operation.Transpose,  // transpose beacuase it does Ax row wise if x is a row vector :D
                       ACount, 1, 1.0f,
                       A, ACount,
                       B, 1,
                       beta, Result, 1);
                    break;
                default:
                    MyLog.Writer.WriteLine(MyLogLevel.ERROR, "Trying to run cublas for undefined MatOperation");
                    break;
            }
        }

开发者ID:Jlaird，项目名称:BrainSimulator，代码行数:48，代码来源:MyMatrixCublasOps.cs

示例11: GenerateUniformArray

        public CudaArray3D GenerateUniformArray(int width, int height, int depth)
        {
            int count = width * height * depth;

            CudaDeviceVariable<float> randomVariable = new CudaDeviceVariable<float>(count);
            CudaArray3D randomArray = new CudaArray3D(CUArrayFormat.Float, width, height, depth, CudaArray3DNumChannels.One, CUDAArray3DFlags.None);

            randomDevice.SetPseudoRandomGeneratorSeed((ulong)DateTime.Now.Ticks);
            randomDevice.GenerateUniform32(randomVariable.DevicePointer, count);

            randomArray.CopyFromDeviceToThis(randomVariable.DevicePointer, sizeof(float));

            randomVariable.Dispose();

            return randomArray;
        }

开发者ID:barograf，项目名称:VoxelTerrain，代码行数:16，代码来源:CUDANoiseCube.cs

示例12: ActivationBackward

        /* Function to perform backward activation  */
        public void ActivationBackward(cudnnActivationMode mode,
										float alpha,
										TensorDescriptor srcDesc,
										CudaDeviceVariable<float> srcData,
										TensorDescriptor srcDiffDesc,
										CudaDeviceVariable<float> srcDiffData,
										TensorDescriptor destDesc,
										CudaDeviceVariable<float> destData,
										float beta,
										TensorDescriptor destDiffDesc,
										CudaDeviceVariable<float> destDiffData
										)
        {
            res = CudaDNNNativeMethods.cudnnActivationBackward(_handle, mode, ref alpha, srcDesc.Desc, srcData.DevicePointer, srcDiffDesc.Desc, srcDiffData.DevicePointer, destDesc.Desc, destData.DevicePointer, ref beta, destDiffDesc.Desc, destDiffData.DevicePointer);
            Debug.WriteLine(String.Format("{0:G}, {1}: {2}", DateTime.Now, "cudnnActivationForward", res));
            if (res != cudnnStatus.Success) throw new CudaDNNException(res);
        }

开发者ID:MutterOberin，项目名称:managedCuda，代码行数:18，代码来源:CudaDNNContext.cs

示例13: Init

        protected override void Init()
        {
            var kernelFileName = KernelFile;
            var initKernel = Ctx.LoadKernel(kernelFileName, "generateData");
            Rotation2 = new CudaDeviceVariable<double>(DimensionsCount * DimensionsCount);
            Rotation1 = new CudaDeviceVariable<double>(DimensionsCount * DimensionsCount);
            Xopt = new CudaDeviceVariable<double>(DimensionsCount);

            long rseed = FunctionNumber + 10000 * InstanceNumber;

            initKernel.Run(
                DimensionsCount,
                rseed,
                Rotation1.DevicePointer,
                Rotation2.DevicePointer,
                Xopt.DevicePointer);
        }

开发者ID:trojkac，项目名称:effective_pso，代码行数:17，代码来源:LunacekBiRastriginAlgorithm.cs

示例14: DrawStringFromGPUMem

        public static void DrawStringFromGPUMem(CudaDeviceVariable<float> inString, int x, int y, uint bgColor, uint fgColor, CUdeviceptr image, int imageWidth, int imageHeight, int stringOffset, int stringLen)
        {
            MyCudaKernel m_drawDigitKernel = MyKernelFactory.Instance.Kernel(MyKernelFactory.Instance.DevCount - 1, @"Observers\DrawStringKernel");
            CudaDeviceVariable<float> characters = MyMemoryManager.Instance.GetGlobalVariable<float>("CHARACTERS_TEXTURE", MyKernelFactory.Instance.DevCount - 1, LoadDigits);

            //MyKernelFactory.Instance.Synchronize();

            m_drawDigitKernel.SetConstantVariable("D_BG_COLOR", bgColor);
            m_drawDigitKernel.SetConstantVariable("D_FG_COLOR", fgColor);
            m_drawDigitKernel.SetConstantVariable("D_IMAGE_WIDTH", imageWidth);
            m_drawDigitKernel.SetConstantVariable("D_IMAGE_HEIGHT", imageHeight);
            m_drawDigitKernel.SetConstantVariable("D_DIGIT_WIDTH", CharacterWidth);
            m_drawDigitKernel.SetConstantVariable("D_DIGIT_SIZE", CharacterSize);
            m_drawDigitKernel.SetConstantVariable("D_DIGITMAP_NBCHARS", CharacterMapNbChars);

            m_drawDigitKernel.SetupExecution(CharacterSize * stringLen);
            m_drawDigitKernel.Run(image, characters.DevicePointer, x, y, inString.DevicePointer + sizeof(float) * stringOffset, stringLen);
        }

开发者ID:steuerj，项目名称:BrainSimulator，代码行数:18，代码来源:MyTextDrawingOps.cs

示例15: Backward

        public void Backward(CudnnSoftmaxAlgorithm algorithm, CudnnSoftmaxMode mode,
                             CudnnTensorDescriptor srcTensor, CudaDeviceVariable<double> srcData, CudnnTensorDescriptor srcDiffTensor, CudaDeviceVariable<double> srcDiffData,
                             CudnnTensorDescriptor destDiffTensor, CudaDeviceVariable<double> destDiffData)
        {
            Contract.Requires(srcTensor != null);
            Contract.Requires(srcData != null);
            Contract.Requires(srcDiffTensor != null);
            Contract.Requires(srcDiffData != null);
            Contract.Requires(destDiffTensor != null);
            Contract.Requires(destDiffData != null);

            ThrowIfNotInitialized();
            CheckIfCompatible(CudnnType.Double, srcTensor, srcDiffTensor, destDiffTensor);

            Invoke(() => CudnnNativeMethods.cudnnSoftmaxBackward(handle, algorithm, mode,
                                                     srcTensor.Handle, srcData.DevicePointer, srcDiffTensor.Handle, srcDiffData.DevicePointer,
                                                     destDiffTensor.Handle, destDiffData.DevicePointer));
        }

开发者ID:redknightlois，项目名称:cudalearn，代码行数:18，代码来源:CudnnContext.Softmax.cs

示例16: Test

 static void Test(byte[] ptxFile)
 {
     const int size = 16;
     var context = new CudaContext();
     var kernel = context.LoadKernelPTX(ptxFile, "kernel");
     var memory = context.AllocateMemory(4 * size);
     var gpuMemory = new CudaDeviceVariable<int>(memory);
     var cpuMemory = new int[size];
     for (var i = 0; i < size; i++)
         cpuMemory[i] = i - 2;
     gpuMemory.CopyToDevice(cpuMemory);
     kernel.BlockDimensions = 4;
     kernel.GridDimensions = 4;
     kernel.Run(memory);
     gpuMemory.CopyToHost(cpuMemory);
     for (var i = 0; i < size; i++)
         Console.WriteLine("{0} = {1}", i, cpuMemory[i]);
 }

开发者ID:shaunstanislaus，项目名称:CudaSharp，代码行数:18，代码来源:Program.cs

示例17: Backward

        public void Backward(CudnnActivationMode mode, CudnnTensorDescriptor srcTensor, CudaDeviceVariable<float> srcData, CudnnTensorDescriptor srcDiffTensor, CudaDeviceVariable<float> srcDiffData,
                                                       CudnnTensorDescriptor destTensor, CudaDeviceVariable<float> destData, CudnnTensorDescriptor destDiffTensor, CudaDeviceVariable<float> destDiffData)
        {
            Contract.Requires(srcTensor != null);
            Contract.Requires(srcData != null);
            Contract.Requires(destTensor != null);
            Contract.Requires(destData != null);
            Contract.Requires(srcDiffTensor != null);
            Contract.Requires(srcDiffData != null);
            Contract.Requires(destDiffTensor != null);
            Contract.Requires(destDiffData != null);

            ThrowIfNotInitialized();
            CheckIfCompatible(CudnnType.Float, srcTensor, srcDiffTensor, destTensor, destDiffTensor);

            Invoke(() => CudnnNativeMethods.cudnnActivationBackward(handle, mode,
                                srcTensor.Handle, srcData.DevicePointer, srcDiffTensor.Handle, srcDiffData.DevicePointer,
                                destTensor.Handle, destData.DevicePointer, destDiffTensor.Handle, destDiffData.DevicePointer));
        }

开发者ID:redknightlois，项目名称:cudalearn，代码行数:19，代码来源:CudnnContext.Activation.cs

示例18: DadeCudaIntersectionDevice

        public DadeCudaIntersectionDevice(RayEngineScene scene, NVContext ctx)
            : base(scene)
        {
            this.scene = scene;
            wallclock = new Stopwatch();
            this.todoRayBuffers = new InputRayBufferCollection();
            this.doneRayBuffers = new OutputRayBufferCollection();
            this.started = false;
            if (ctx != null)
            {
                this.cudaContext = ctx;
            }
            else
            {
                this.cudaContext = new NVContext() { Context = new CudaContext(CudaContext.GetMaxGflopsDeviceId()) };
            }
            using (var sr = new StreamReader(@"G:\Git\RayDen\CudaMegaRay\x64\Release\Intersection.cu.ptx"))
            {
                intersectKernel = cudaContext.Context.LoadKernelPTX(sr.BaseStream, "Intersect");
            }

            this.rays = new CudaDeviceVariable<RayData>(RayBuffer.RayBufferSize);
            this.hits = new CudaDeviceVariable<RayHit>(RayBuffer.RayBufferSize);
            verts = scene.Vertices.ToArray();
            tris=scene.Triangles.Select(i => i.GetInfo()).ToArray();

            if (GlobalConfiguration.Instance.UseSceneCaching && scene.Cache != null)
            {
                bvh = scene.Cache.BvhData;
                nodesCount = scene.Cache.BvhData.Length;
            }
            else
            {
                var da = new BvhDataAdapter(scene);
                var treeData = da.BuildData();
                bvh = treeData;
                nodesCount = treeData.Length;

            }

            Tracer.TraceLine("BVH Data Size {0:F3} MBytes", (nodesCount * 32f) / (1024f * 1024f));
        }

开发者ID:HungryBear，项目名称:rayden，代码行数:42，代码来源:DadeCudaIntersectionDevice.cs

示例19: Init

        protected override void Init()
        {
            var kernelFileName = KernelFile;
            var initKernel = Ctx.LoadKernel(kernelFileName, "generateData");
            initKernel.BlockDimensions = 1;
            initKernel.GridDimensions = 1;

            Rotation = new CudaDeviceVariable<double>(DimensionsCount * DimensionsCount);
            PeakValues = new CudaDeviceVariable<double>(PeaksCount);
            XLocal = new CudaDeviceVariable<double>(DimensionsCount * PeaksCount);
            ArrScales = new CudaDeviceVariable<double>(DimensionsCount * PeaksCount);

            int rseed = FunctionNumber + 10000 * InstanceNumber;
            initKernel.Run(
                DimensionsCount,
                rseed,
                Rotation.DevicePointer,
                PeaksCount,
                PeakValues.DevicePointer,
                XLocal.DevicePointer,
                ArrScales.DevicePointer);
        }

开发者ID:trojkac，项目名称:effective_pso，代码行数:22，代码来源:GallagherAlgorithm.cs

示例20: Init

        protected override void Init()
        {
            var kernelFileName = KernelFile;
            var initKernel = Ctx.LoadKernel(kernelFileName, "generateData");
            Xopt = new CudaDeviceVariable<double>(DimensionsCount);
            M = new CudaDeviceVariable<double>(DimensionsCount * DimensionsCount);
            B = new CudaDeviceVariable<double>(DimensionsCount);

            var d_fopt = new CudaDeviceVariable<double>(1);

            long rseed = FunctionNumber + 10000 * InstanceNumber;

            double[] h_ak = new double[12];
            double[] h_bk = new double[12];

            for (int i = 0; i < 12; i++)
            {
                h_ak[i] = Math.Pow(0.5, i);
                h_bk[i] = Math.Pow(3.0, i);
                F0 += h_ak[i]*Math.Cos(Math.PI*h_bk[i]);
            }

            Ak = h_ak;
            Bk = h_bk;

            initKernel.Run(
                DimensionsCount,
                rseed,
                FunctionNumber,
                InstanceNumber,
                M.DevicePointer,
                B.DevicePointer,
                Xopt.DevicePointer,
                d_fopt.DevicePointer);

            double[] fopt_arr = d_fopt;

            Fopt = fopt_arr[0];
        }

开发者ID:trojkac，项目名称:effective_pso，代码行数:39，代码来源:WeierstrassAlgorithm.cs

注：本文中的CudaDeviceVariable类示例整理自Github/MSDocs等源码及文档管理平台，相关代码片段筛选自各路编程大神贡献的开源项目，源码版权归原作者所有，传播和使用请参考对应项目的License；未经允许，请勿转载。

鲜花

握手

雷人

路过

鸡蛋

该文章已有0人参与评论

请发表评论

全部评论

专题导读

More+

10-27 六六分期app的软件客服如何联系？(六六分期

11-06 可心卡盟:win10系统火狐flash插件崩溃怎么

11-06 亲亲特价:怎么删除回收站图标

11-06 济南大学虚拟社区:鲁大师节能降温的具体办

11-06 xlueops.exe:无线网络安装向导

11-06 女斗合众国:win7系统cf与主机连接不稳定怎

11-06 0xc000022-[cf烟雾头]cf怎么调烟雾头

11-06 qizideyouhuo:应用程序无法正常启动0xc0000

11-06 ipz-185:win7系统vcf文件怎么打开

11-06 傻哥蹦迪:win10系统s4怎么打开usb调试

11-06 八神浩树gtaste:回收站清空了怎么恢复

11-06 妖尾之黑色守护:win10系统电脑没有1440x900

11-06 校园至尊魔王小说:win7系统浏览网页时字体

11-06 女斗合众国:win10系统访问共享文件夹提示请

11-06 tokyo hot n0654:恢复win7系统默认字体一招

11-06 雨酷仙境:设置win7系统转移临时文件夹腾出

11-06 阿穆纳伊之杖:win7系统开始菜单在右边还原

11-06 tunespotting:win10系统火狐flash插件总是

11-06 甘尔葛分析师：计谋网站seo关键词暴涨有什

11-06 蔡贵霖: 计谋网站seo关键词暴涨有什么秘密

11-06 博益网首页:ao3网页版进入不了解决方法

11-06 漏斗子专栏: 网站数据分析小白易懂精华篇

11-06 见证双虹怎么做:win7系统开启telnet命令的

11-06 颾狐蝶蜋:系统资源不足无法完成请求的服务

11-06 国光中学校歌:提交网站到alexa查询详细步骤

11-06 西安有情天:静态网页和动态网页的区别

11-06 红木雅尚斋:外部链接构造对网站的好处

11-06 前官礼遇：防止域名劫持–增强域安全性的10

11-06 密传二转答案: 中文分词算法有哪些

11-06 金泉家园邮编:百度快照劫持的表现及应对方

C# Culture类代码示例发布时间：2022-05-24

C# CudaArray1D类代码示例发布时间：2022-05-24

剪的笔顺,诠释剪的笔画,认识剪的部首

1 六六分期app的软件客服如何联系？(六六分期

六六分期app的软件客服如何联系？不知道吗？加qq群【895510560】即可！标题：六六分期

阅读：17996|2023-10-27

2 可心卡盟:win10系统火狐flash插件崩溃怎么

今天小编告诉大家如何处理win10系统火狐flash插件总是崩溃的问题，可能很多用户都不知

阅读：9576|2022-11-06

3 亲亲特价:怎么删除回收站图标

今天小编告诉大家如何对win10系统删除桌面回收站图标进行设置，可能很多用户都不知道

阅读：8131|2022-11-06

4 济南大学虚拟社区:鲁大师节能降温的具体办

今天小编告诉大家如何对win10系统电脑设置节能降温的设置方法，想必大家都遇到过需要

阅读：8515|2022-11-06

5 xlueops.exe:无线网络安装向导

我们在使用xp系统的过程中,经常需要对xp系统无线网络安装向导设置进行设置，可能很多

阅读：8418|2022-11-06

6 女斗合众国:win7系统cf与主机连接不稳定怎

今天小编告诉大家如何处理win7系统玩cf老是与主机连接不稳定的问题，可能很多用户都不

阅读：9315|2022-11-06

7 0xc000022-[cf烟雾头]cf怎么调烟雾头

电脑对日常生活的重要性小编就不多说了，可是一旦碰到win7系统设置cf烟雾头的问题，很

阅读：8381|2022-11-06

8 qizideyouhuo:应用程序无法正常启动0xc0000

我们在日常使用电脑的时候，有的小伙伴们可能在打开应用的时候会遇见提示应用程序无法

阅读：7814|2022-11-06

9 ipz-185:win7系统vcf文件怎么打开

今天小编告诉大家如何对win7系统打开vcf文件进行设置，可能很多用户都不知道怎么对win

阅读：8369|2022-11-06

10 傻哥蹦迪:win10系统s4怎么打开usb调试

今天小编告诉大家如何对win10系统s4开启USB调试模式进行设置，可能很多用户都不知道怎

阅读：7365|2022-11-06

客服电话

电子邮件

C# CudaDeviceVariable类代码示例

示例1: DCTQuantFwd8x8LS

示例2: Init

示例3: CudaIntersectionDevice

示例4: Init

示例5: Backward

示例6: launchConfig

示例7: BackwardBias

示例8: BackwardData

示例9: GraphCut4

示例10: Run

示例11: GenerateUniformArray

示例12: ActivationBackward

示例13: Init

示例14: DrawStringFromGPUMem

示例15: Backward

示例16: Test

示例17: Backward

示例18: DadeCudaIntersectionDevice

示例19: Init

示例20: Init

请发表评论

全部评论

上一篇：

下一篇：

organization/KotlinInside: 비공식 디시인

GitbookIO/gitbook:

juleswhite/mobile-cloud-asgn1

kyamagu/matlab-json: Use official API: h

墙壁眼睛膝盖

剪的笔顺,诠释剪的笔画,认识剪的部首

六六分期app的软件客服如何联系？(六六分期

florent37/ViewAnimator: A fluent Android

florent37/Shrine-MaterialDesign2: implem

CVE-2020-36276

SimpleSoftwareIO/simple-sms: Send and re

关于我们

产品与服务

解决方案

139-2527-9053