From bf72a4f2b0c8937102d1b791755c1ae5b0da7cad Mon Sep 17 00:00:00 2001 From: Andy Ayers Date: Thu, 7 Jan 2016 09:13:50 -0800 Subject: [PATCH] Add SciMark benchmark 3rd party attributions retained in file header comments. --- .../Performance/CodeQuality/SciMark/CommandLine.cs | 103 +++++ .../Performance/CodeQuality/SciMark/Constants.cs | 48 +++ .../src/JIT/Performance/CodeQuality/SciMark/FFT.cs | 219 ++++++++++ .../src/JIT/Performance/CodeQuality/SciMark/LU.cs | 268 +++++++++++++ .../Performance/CodeQuality/SciMark/MonteCarlo.cs | 69 ++++ .../JIT/Performance/CodeQuality/SciMark/Random.cs | 302 ++++++++++++++ .../src/JIT/Performance/CodeQuality/SciMark/SOR.cs | 53 +++ .../Performance/CodeQuality/SciMark/SciMark.csproj | 55 +++ .../CodeQuality/SciMark/SparseCompRow.cs | 56 +++ .../Performance/CodeQuality/SciMark/Stopwatch.cs | 109 +++++ .../JIT/Performance/CodeQuality/SciMark/kernel.cs | 444 +++++++++++++++++++++ 11 files changed, 1726 insertions(+) create mode 100644 tests/src/JIT/Performance/CodeQuality/SciMark/CommandLine.cs create mode 100644 tests/src/JIT/Performance/CodeQuality/SciMark/Constants.cs create mode 100644 tests/src/JIT/Performance/CodeQuality/SciMark/FFT.cs create mode 100644 tests/src/JIT/Performance/CodeQuality/SciMark/LU.cs create mode 100644 tests/src/JIT/Performance/CodeQuality/SciMark/MonteCarlo.cs create mode 100644 tests/src/JIT/Performance/CodeQuality/SciMark/Random.cs create mode 100644 tests/src/JIT/Performance/CodeQuality/SciMark/SOR.cs create mode 100644 tests/src/JIT/Performance/CodeQuality/SciMark/SciMark.csproj create mode 100644 tests/src/JIT/Performance/CodeQuality/SciMark/SparseCompRow.cs create mode 100644 tests/src/JIT/Performance/CodeQuality/SciMark/Stopwatch.cs create mode 100644 tests/src/JIT/Performance/CodeQuality/SciMark/kernel.cs diff --git a/tests/src/JIT/Performance/CodeQuality/SciMark/CommandLine.cs b/tests/src/JIT/Performance/CodeQuality/SciMark/CommandLine.cs new file mode 100644 index 0000000..8ee71c7 --- /dev/null +++ b/tests/src/JIT/Performance/CodeQuality/SciMark/CommandLine.cs @@ -0,0 +1,103 @@ +/// +/// This is a port of the SciMark2a Java Benchmark to C# by +/// Chris Re (cmr28@cornell.edu) and Werner Vogels (vogels@cs.cornell.edu) +/// +/// For details on the original authors see http://math.nist.gov/scimark2 +/// +/// This software is likely to burn your processor, bitflip your memory chips +/// anihilate your screen and corrupt all your disks, so you it at your +/// own risk. +/// + + +using System; + +namespace SciMark2 +{ + /// + /// SciMark2: A Java numerical benchmark measuring performance + /// of computational kernels for FFTs, Monte Carlo simulation, + /// sparse matrix computations, Jacobi SOR, and dense LU matrix + /// factorizations. + /// + + public class CommandLine + { + /// + /// Benchmark 5 kernels with individual Mflops. + /// "results[0]" has the average Mflop rate. + /// + public static int Main(System.String[] args) + { +#if DEBUG + double min_time = Constants.RESOLUTION_TINY; +#else + double min_time = Constants.RESOLUTION_DEFAULT; +#endif + + int FFT_size = Constants.FFT_SIZE; + int SOR_size = Constants.SOR_SIZE; + int Sparse_size_M = Constants.SPARSE_SIZE_M; + int Sparse_size_nz = Constants.SPARSE_SIZE_nz; + int LU_size = Constants.LU_SIZE; + + // look for runtime options + if (args.Length > 0) + { + if (args[0].ToUpper().Equals("-h") || + args[0].ToUpper().Equals("-help")) + { + Console.WriteLine("Usage: [-large] [iterations]"); + return -1; + } + + int current_arg = 0; + if (args[current_arg].ToUpper().Equals("-LARGE")) + { + FFT_size = Constants.LG_FFT_SIZE; + SOR_size = Constants.LG_SOR_SIZE; + Sparse_size_M = Constants.LG_SPARSE_SIZE_M; + Sparse_size_nz = Constants.LG_SPARSE_SIZE_nz; + LU_size = Constants.LG_LU_SIZE; + + current_arg++; + } + + if (args.Length > current_arg) + min_time = Double.Parse(args[current_arg]); + } + Console.WriteLine("** **"); + Console.WriteLine("** SciMark2a Numeric Benchmark, see http://math.nist.gov/scimark **"); + Console.WriteLine("** **"); + + // run the benchmark + double[] res = new double[6]; + SciMark2.Random R = new SciMark2.Random(Constants.RANDOM_SEED); + + Console.WriteLine("Mininum running time = {0} seconds", min_time); + + res[1] = kernel.measureFFT(FFT_size, min_time, R); + + res[2] = kernel.measureSOR(SOR_size, min_time, R); + + res[3] = kernel.measureMonteCarlo(min_time, R); + + res[4] = kernel.measureSparseMatmult(Sparse_size_M, Sparse_size_nz, min_time, R); + + res[5] = kernel.measureLU(LU_size, min_time, R); + + res[0] = (res[1] + res[2] + res[3] + res[4] + res[5]) / 5; + + // print out results + Console.WriteLine(); + Console.WriteLine("Composite Score: {0:F2} MFlops", res[0]); + Console.WriteLine("FFT : {0} - ({1})", res[1] == 0.0 ? "ERROR, INVALID NUMERICAL RESULT!" : res[1].ToString("F2"), FFT_size); + Console.WriteLine("SOR : {1:F2} - ({0}x{0})", SOR_size, res[2]); + Console.WriteLine("Monte Carlo : {0:F2}", res[3]); + Console.WriteLine("Sparse MatMult : {2:F2} - (N={0}, nz={1})", Sparse_size_M, Sparse_size_nz, res[4]); + Console.WriteLine("LU : {1} - ({0}x{0})", LU_size, res[1] == 0.0 ? "ERROR, INVALID NUMERICAL RESULT!" : res[5].ToString("F2")); + + return 100; + } + } +} diff --git a/tests/src/JIT/Performance/CodeQuality/SciMark/Constants.cs b/tests/src/JIT/Performance/CodeQuality/SciMark/Constants.cs new file mode 100644 index 0000000..a99bce9 --- /dev/null +++ b/tests/src/JIT/Performance/CodeQuality/SciMark/Constants.cs @@ -0,0 +1,48 @@ +/// +/// This is a port of the SciMark2a Java Benchmark to C# by +/// Chris Re (cmr28@cornell.edu) and Werner Vogels (vogels@cs.cornell.edu) +/// +/// For details on the original authors see http://math.nist.gov/scimark2 +/// +/// This software is likely to burn your processor, bitflip your memory chips +/// anihilate your screen and corrupt all your disks, so you it at your +/// own risk. +/// + + +namespace SciMark2 +{ + public class Constants + { + public const double RESOLUTION_DEFAULT = 2.0; /*secs*/ + public const double RESOLUTION_TINY = 0.02; /*secs*/ + public const int RANDOM_SEED = 101010; + + // default: small (cache-contained) problem sizes + // + public const int FFT_SIZE = 1024; // must be a power of two + public const int SOR_SIZE = 100; // NxN grid + public const int SPARSE_SIZE_M = 1000; + public const int SPARSE_SIZE_nz = 5000; + public const int LU_SIZE = 100; + + // large (out-of-cache) problem sizes + // + public const int LG_FFT_SIZE = 1048576; // must be a power of two + public const int LG_SOR_SIZE = 1000; // NxN grid + public const int LG_SPARSE_SIZE_M = 100000; + public const int LG_SPARSE_SIZE_nz = 1000000; + public const int LG_LU_SIZE = 1000; + + // tiny problem sizes (used to mainly to preload network classes + // for applet, so that network download times + // are factored out of benchmark.) + // + public const int TINY_FFT_SIZE = 16; // must be a power of two + public const int TINY_SOR_SIZE = 10; // NxN grid + public const int TINY_SPARSE_SIZE_M = 10; + public const int TINY_SPARSE_SIZE_N = 10; + public const int TINY_SPARSE_SIZE_nz = 50; + public const int TINY_LU_SIZE = 10; + } +} diff --git a/tests/src/JIT/Performance/CodeQuality/SciMark/FFT.cs b/tests/src/JIT/Performance/CodeQuality/SciMark/FFT.cs new file mode 100644 index 0000000..df32c42 --- /dev/null +++ b/tests/src/JIT/Performance/CodeQuality/SciMark/FFT.cs @@ -0,0 +1,219 @@ +/// +/// This is a port of the SciMark2a Java Benchmark to C# by +/// Chris Re (cmr28@cornell.edu) and Werner Vogels (vogels@cs.cornell.edu) +/// +/// For details on the original authors see http://math.nist.gov/scimark2 +/// +/// This software is likely to burn your processor, bitflip your memory chips +/// anihilate your screen and corrupt all your disks, so you it at your +/// own risk. +/// + + +using System; + +namespace SciMark2 +{ + /// Computes FFT's of complex, double precision data where n is an integer power of 2. + /// This appears to be slower than the Radix2 method, + /// but the code is smaller and simpler, and it requires no extra storage. + ///

+ /// + /// + /// + /// Bruce R. Miller bruce.miller@nist.gov, + /// Derived from GSL (Gnu Scientific Library), + /// GSL's FFT Code by Brian Gough bjg@vvv.lanl.gov + /// + + public class FFT + { + public static double num_flops(int N) + { + double Nd = (double)N; + double logN = (double)log2(N); + + return (5.0 * Nd - 2) * logN + 2 * (Nd + 1); + } + + + /// + /// Compute Fast Fourier Transform of (complex) data, in place. + /// + public static void transform(double[] data) + { + transform_internal(data, -1); + } + + /// + /// Compute Inverse Fast Fourier Transform of (complex) data, in place. + /// + public static void inverse(double[] data) + { + transform_internal(data, +1); + // Normalize + int nd = data.Length; + int n = nd / 2; + double norm = 1 / ((double)n); + for (int i = 0; i < nd; i++) + data[i] *= norm; + } + + /// + /// Accuracy check on FFT of data. Make a copy of data, Compute the FFT, then + /// the inverse and compare to the original. Returns the rms difference. + /// + public static double test(double[] data) + { + int nd = data.Length; + // Make duplicate for comparison + double[] copy = new double[nd]; + Array.Copy(data, 0, copy, 0, nd); + // Transform & invert + transform(data); + inverse(data); + // Compute RMS difference. + double diff = 0.0; + for (int i = 0; i < nd; i++) + { + double d = data[i] - copy[i]; + diff += d * d; + } + return Math.Sqrt(diff / nd); + } + + /// + /// Make a random array of n (complex) elements. + /// + public static double[] makeRandom(int n) + { + int nd = 2 * n; + double[] data = new double[nd]; + System.Random r = new System.Random(); + for (int i = 0; i < nd; i++) + data[i] = r.NextDouble(); + return data; + } + + protected internal static int log2(int n) + { + int log = 0; + for (int k = 1; k < n; k *= 2, log++) + ; + if (n != (1 << log)) + throw new Exception("FFT: Data length is not a power of 2!: " + n); + return log; + } + + protected internal static void transform_internal(double[] data, int direction) + { + if (data.Length == 0) + return; + int n = data.Length / 2; + if (n == 1) + return; + // Identity operation! + int logn = log2(n); + + /* bit reverse the input data for decimation in time algorithm */ + bitreverse(data); + + /* apply fft recursion */ + /* this loop executed log2(N) times */ + for (int bit = 0, dual = 1; bit < logn; bit++, dual *= 2) + { + double w_real = 1.0; + double w_imag = 0.0; + + double theta = 2.0 * direction * Math.PI / (2.0 * (double)dual); + double s = Math.Sin(theta); + double t = Math.Sin(theta / 2.0); + double s2 = 2.0 * t * t; + + /* a = 0 */ + for (int b = 0; b < n; b += 2 * dual) + { + int i = 2 * b; + int j = 2 * (b + dual); + + double wd_real = data[j]; + double wd_imag = data[j + 1]; + + data[j] = data[i] - wd_real; + data[j + 1] = data[i + 1] - wd_imag; + data[i] += wd_real; + data[i + 1] += wd_imag; + } + + /* a = 1 .. (dual-1) */ + for (int a = 1; a < dual; a++) + { + /* trignometric recurrence for w-> exp(i theta) w */ + { + double tmp_real = w_real - s * w_imag - s2 * w_real; + double tmp_imag = w_imag + s * w_real - s2 * w_imag; + w_real = tmp_real; + w_imag = tmp_imag; + } + for (int b = 0; b < n; b += 2 * dual) + { + int i = 2 * (b + a); + int j = 2 * (b + a + dual); + + double z1_real = data[j]; + double z1_imag = data[j + 1]; + + double wd_real = w_real * z1_real - w_imag * z1_imag; + double wd_imag = w_real * z1_imag + w_imag * z1_real; + + data[j] = data[i] - wd_real; + data[j + 1] = data[i + 1] - wd_imag; + data[i] += wd_real; + data[i + 1] += wd_imag; + } + } + } + } + + + protected internal static void bitreverse(double[] data) + { + /* This is the Goldrader bit-reversal algorithm */ + int n = data.Length / 2; + int nm1 = n - 1; + int i = 0; + int j = 0; + for (; i < nm1; i++) + { + //int ii = 2*i; + int ii = i << 1; + + //int jj = 2*j; + int jj = j << 1; + + //int k = n / 2 ; + int k = n >> 1; + + if (i < j) + { + double tmp_real = data[ii]; + double tmp_imag = data[ii + 1]; + data[ii] = data[jj]; + data[ii + 1] = data[jj + 1]; + data[jj] = tmp_real; + data[jj + 1] = tmp_imag; + } + + while (k <= j) + { + //j = j - k ; + j -= k; + + //k = k / 2 ; + k >>= 1; + } + j += k; + } + } + } +} diff --git a/tests/src/JIT/Performance/CodeQuality/SciMark/LU.cs b/tests/src/JIT/Performance/CodeQuality/SciMark/LU.cs new file mode 100644 index 0000000..481922a --- /dev/null +++ b/tests/src/JIT/Performance/CodeQuality/SciMark/LU.cs @@ -0,0 +1,268 @@ +/// +/// This is a port of the SciMark2a Java Benchmark to C# by +/// Chris Re (cmr28@cornell.edu) and Werner Vogels (vogels@cs.cornell.edu) +/// +/// For details on the original authors see http://math.nist.gov/scimark2 +/// +/// This software is likely to burn your processor, bitflip your memory chips +/// anihilate your screen and corrupt all your disks, so you it at your +/// own risk. +/// + + +using System; + +namespace SciMark2 +{ + /// + /// LU matrix factorization. (Based on TNT implementation.) + /// Decomposes a matrix A into a triangular lower triangular + /// factor (L) and an upper triangular factor (U) such that + /// A = L*U. By convnetion, the main diagonal of L consists + /// of 1's so that L and U can be stored compactly in + /// a NxN matrix. + /// + public class LU + { + private double[][] _LU; + private int[] _pivot; + + /// + /// Returns a copy of the compact LU factorization. + /// (useful mainly for debugging.) + /// + /// + /// + /// the compact LU factorization. The U factor + /// is stored in the upper triangular portion, and the L + /// factor is stored in the lower triangular portion. + /// The main diagonal of L consists (by convention) of + /// ones, and is not explicitly stored. + /// + public static double num_flops(int N) + { + // rougly 2/3*N^3 + + double Nd = (double)N; + + return (2.0 * Nd * Nd * Nd / 3.0); + } + + protected internal static double[] new_copy(double[] x) + { + double[] T = new double[x.Length]; + x.CopyTo(T, 0); + return T; + } + + + protected internal static double[][] new_copy(double[][] A) + { + int M = A.Length; + int N = A[0].Length; + + double[][] T = new double[M][]; + for (int i = 0; i < M; i++) + { + T[i] = new double[N]; + } + + for (int i = 0; i < M; i++) + { + A[i].CopyTo(T[i], 0); + } + + return T; + } + + + + public static int[] new_copy(int[] x) + { + int[] T = new int[x.Length]; + x.CopyTo(T, 0); + return T; + } + + protected internal static void insert_copy(double[][] B, double[][] A) + { + for (int i = 0; i < A.Length; i++) + { + A[i].CopyTo(B[i], 0); + } + } + + /// + /// Initalize LU factorization from matrix. + /// + /// + /// (in) the matrix to associate with this factorization. + /// + /// + public LU(double[][] A) + { + _LU = new_copy(A); + _pivot = new int[A.Length]; + + factor(_LU, _pivot); + } + + /// + /// Solve a linear system, with pre-computed factorization. + /// + /// + /// (in) the right-hand side. + /// + /// + /// solution vector. + /// + public virtual double[] solve(double[] b) + { + double[] x = new_copy(b); + + solve(_LU, _pivot, x); + return x; + } + + + /// + /// LU factorization (in place). + /// + /// + /// (in/out) On input, the matrix to be factored. + /// On output, the compact LU factorization. + /// + /// + /// (out) The pivot vector records the + /// reordering of the rows of A during factorization. + /// + /// + /// 0, if OK, nozero value, othewise. + /// + public static int factor(double[][] A, int[] pivot) + { + int N = A.Length; + int M = A[0].Length; + + int minMN = Math.Min(M, N); + + for (int j = 0; j < minMN; j++) + { + // find pivot in column j and test for singularity. + int jp = j; + + double t = Math.Abs(A[j][j]); + for (int i = j + 1; i < M; i++) + { + double ab = Math.Abs(A[i][j]); + if (ab > t) + { + jp = i; + t = ab; + } + } + + pivot[j] = jp; + + // jp now has the index of maximum element + // of column j, below the diagonal + if (A[jp][j] == 0) + return 1; + + // factorization failed because of zero pivot + if (jp != j) + { + // swap rows j and jp + double[] tA = A[j]; + A[j] = A[jp]; + A[jp] = tA; + } + + if (j < M - 1) + { + // compute elements j+1:M of jth column + // note A(j,j), was A(jp,p) previously which was + // guarranteed not to be zero (Label #1) + // + double recp = 1.0 / A[j][j]; + + for (int k = j + 1; k < M; k++) + A[k][j] *= recp; + } + + if (j < minMN - 1) + { + // rank-1 update to trailing submatrix: E = E - x*y; + // + // E is the region A(j+1:M, j+1:N) + // x is the column vector A(j+1:M,j) + // y is row vector A(j,j+1:N) + for (int ii = j + 1; ii < M; ii++) + { + double[] Aii = A[ii]; + double[] Aj = A[j]; + double AiiJ = Aii[j]; + for (int jj = j + 1; jj < N; jj++) + Aii[jj] -= AiiJ * Aj[jj]; + } + } + } + + return 0; + } + + + /// Solve a linear system, using a prefactored matrix + /// in LU form. + /// + /// (in) the factored matrix in LU form. + /// + /// (in) the pivot vector which lists + /// the reordering used during the factorization + /// stage. + /// + /// (in/out) On input, the right-hand side. + /// On output, the solution vector. + /// + /// + public static void solve(double[][] A, int[] pvt, double[] b) + { + int M = A.Length; + int N = A[0].Length; + int ii = 0; + + for (int i = 0; i < M; i++) + { + int ip = pvt[i]; + double sum = b[ip]; + + b[ip] = b[i]; + if (ii == 0) + { + for (int j = ii; j < i; j++) + { + sum -= A[i][j] * b[j]; + } + } + else + { + if (sum == 0.0) + { + ii = i; + } + } + b[i] = sum; + } + + for (int i = N - 1; i >= 0; i--) + { + double sum = b[i]; + for (int j = i + 1; j < N; j++) + { + sum -= A[i][j] * b[j]; + } + b[i] = sum / A[i][i]; + } + } + } +} \ No newline at end of file diff --git a/tests/src/JIT/Performance/CodeQuality/SciMark/MonteCarlo.cs b/tests/src/JIT/Performance/CodeQuality/SciMark/MonteCarlo.cs new file mode 100644 index 0000000..0e5adce --- /dev/null +++ b/tests/src/JIT/Performance/CodeQuality/SciMark/MonteCarlo.cs @@ -0,0 +1,69 @@ +/// +/// This is a port of the SciMark2a Java Benchmark to C# by +/// Chris Re (cmr28@cornell.edu) and Werner Vogels (vogels@cs.cornell.edu) +/// +/// For details on the original authors see http://math.nist.gov/scimark2 +/// +/// This software is likely to burn your processor, bitflip your memory chips +/// anihilate your screen and corrupt all your disks, so you it at your +/// own risk. +/// + + +using System; + +namespace SciMark2 +{ + /// Estimate Pi by approximating the area of a circle. + /// How: generate N random numbers in the unit square, (0,0) to (1,1) + /// and see how are within a radius of 1 or less, i.e. + /// 
  
+    /// sqrt(x^2 + y^2) < r
+    ///
+ /// since the radius is 1.0, we can square both sides + /// and avoid a sqrt() computation: + /// 
+    /// x^2 + y^2 <= 1.0
+    ///
+ /// this area under the curve is (Pi * r^2)/ 4.0, + /// and the area of the unit of square is 1.0, + /// so Pi can be approximated by + /// 
+    /// # points with x^2+y^2 < 1
+    /// Pi =~ 		--------------------------  * 4.0
+    /// total # points
+    ///
+ /// + + public class MonteCarlo + { + internal const int SEED = 113; + + public static double num_flops(int Num_samples) + { + // 3 flops in x^2+y^2 and 1 flop in random routine + + return ((double)Num_samples) * 4.0; + } + + + + public static double integrate(int Num_samples) + { + SciMark2.Random R = new SciMark2.Random(SEED); + + + int under_curve = 0; + for (int count = 0; count < Num_samples; count++) + { + double x = R.nextDouble(); + double y = R.nextDouble(); + + if (x * x + y * y <= 1.0) + under_curve++; + } + + return ((double)under_curve / Num_samples) * 4.0; + } + } +} \ No newline at end of file diff --git a/tests/src/JIT/Performance/CodeQuality/SciMark/Random.cs b/tests/src/JIT/Performance/CodeQuality/SciMark/Random.cs new file mode 100644 index 0000000..b65ec32 --- /dev/null +++ b/tests/src/JIT/Performance/CodeQuality/SciMark/Random.cs @@ -0,0 +1,302 @@ +/// +/// This is a port of the SciMark2a Java Benchmark to C# by +/// Chris Re (cmr28@cornell.edu) and Werner Vogels (vogels@cs.cornell.edu) +/// +/// For details on the original authors see http://math.nist.gov/scimark2 +/// +/// This software is likely to burn your processor, bitflip your memory chips +/// anihilate your screen and corrupt all your disks, so you it at your +/// own risk. +/// + + +using System; +using System.Runtime.CompilerServices; + +namespace SciMark2 +{ + /* Random.java based on Java Numerical Toolkit (JNT) Random.UniformSequence + class. We do not use Java's own java.util.Random so that we can compare + results with equivalent C and Fortran coces.*/ + + public class Random + { + /*------------------------------------------------------------------------------ + CLASS VARIABLES + ------------------------------------------------------------------------------ */ + + internal int seed = 0; + + private int[] _m; + private int _i = 4; + private int _j = 16; + + private const int mdig = 32; + private const int one = 1; + private int _m1; + private int _m2; + + private double _dm1; + + private bool _haveRange = false; + private double _left = 0.0; + private double _right = 1.0; + private double _width = 1.0; + + + /* ------------------------------------------------------------------------------ + CONSTRUCTORS + ------------------------------------------------------------------------------ */ + + /// + /// Initializes a sequence of uniformly distributed quasi random numbers with a + /// seed based on the system clock. + /// + public Random() + { + initialize((int)System.DateTime.Now.Ticks); + } + + /// + /// Initializes a sequence of uniformly distributed quasi random numbers on a + /// given half-open interval [left,right) with a seed based on the system + /// clock. + /// + /// (double)
+ /// The left endpoint of the half-open interval [left,right). + /// + /// (double)
+ /// The right endpoint of the half-open interval [left,right). + /// + public Random(double left, double right) + { + initialize((int)System.DateTime.Now.Ticks); + _left = left; + _right = right; + _width = right - left; + _haveRange = true; + } + + /// + /// Initializes a sequence of uniformly distributed quasi random numbers with a + /// given seed. + /// + /// (int)
+ /// The seed of the random number generator. Two sequences with the same + /// seed will be identical. + /// + public Random(int seed) + { + initialize(seed); + } + + /// Initializes a sequence of uniformly distributed quasi random numbers + /// with a given seed on a given half-open interval [left,right). + /// + /// (int)
+ /// The seed of the random number generator. Two sequences with the same + /// seed will be identical. + /// + /// (double)
+ /// The left endpoint of the half-open interval [left,right). + /// + /// (double)
+ /// The right endpoint of the half-open interval [left,right). + /// + public Random(int seed, double left, double right) + { + initialize(seed); + _left = left; + _right = right; + _width = right - left; + _haveRange = true; + } + + /* ------------------------------------------------------------------------------ + PUBLIC METHODS + ------------------------------------------------------------------------------ */ + + /// + /// Returns the next random number in the sequence. + /// + public double nextDouble() + { + int k; + + k = _m[_i] - _m[_j]; + if (k < 0) + k += _m1; + _m[_j] = k; + + if (_i == 0) + _i = 16; + else + _i--; + + if (_j == 0) + _j = 16; + else + _j--; + + if (_haveRange) + return _left + _dm1 * (double)k * _width; + else + return _dm1 * (double)k; + } + + /// + /// Returns the next N random numbers in the sequence, as + /// a vector. + /// + public void nextDoubles(double[] x) + { + int N = x.Length; + int remainder = N & 3; + + if (_haveRange) + { + for (int count = 0; count < N; count++) + { + int k = _m[_i] - _m[_j]; + + if (_i == 0) + _i = 16; + else + _i--; + + if (k < 0) + k += _m1; + _m[_j] = k; + + if (_j == 0) + _j = 16; + else + _j--; + + x[count] = _left + _dm1 * (double)k * _width; + } + } + else + { + for (int count = 0; count < remainder; count++) + { + int k = _m[_i] - _m[_j]; + + if (_i == 0) + _i = 16; + else + _i--; + + if (k < 0) + k += _m1; + _m[_j] = k; + + if (_j == 0) + _j = 16; + else + _j--; + + + x[count] = _dm1 * (double)k; + } + + for (int count = remainder; count < N; count += 4) + { + int k = _m[_i] - _m[_j]; + if (_i == 0) + _i = 16; + else + _i--; + if (k < 0) + k += _m1; + _m[_j] = k; + if (_j == 0) + _j = 16; + else + _j--; + x[count] = _dm1 * (double)k; + + + k = _m[_i] - _m[_j]; + if (_i == 0) + _i = 16; + else + _i--; + if (k < 0) + k += _m1; + _m[_j] = k; + if (_j == 0) + _j = 16; + else + _j--; + x[count + 1] = _dm1 * (double)k; + + + k = _m[_i] - _m[_j]; + if (_i == 0) + _i = 16; + else + _i--; + if (k < 0) + k += _m1; + _m[_j] = k; + if (_j == 0) + _j = 16; + else + _j--; + x[count + 2] = _dm1 * (double)k; + + + k = _m[_i] - _m[_j]; + if (_i == 0) + _i = 16; + else + _i--; + if (k < 0) + k += _m1; + _m[_j] = k; + if (_j == 0) + _j = 16; + else + _j--; + x[count + 3] = _dm1 * (double)k; + } + } + } + + /*---------------------------------------------------------------------------- + PRIVATE METHODS + ------------------------------------------------------------------------ */ + + private void initialize(int seed) + { + // First the initialization of the member variables; + _m1 = (one << mdig - 2) + ((one << mdig - 2) - one); + _m2 = one << mdig / 2; + _dm1 = 1.0 / (double)_m1; + + int jseed, k0, k1, j0, j1, iloop; + + this.seed = seed; + + _m = new int[17]; + + jseed = System.Math.Min(System.Math.Abs(seed), _m1); + if (jseed % 2 == 0) + --jseed; + k0 = 9069 % _m2; + k1 = 9069 / _m2; + j0 = jseed % _m2; + j1 = jseed / _m2; + for (iloop = 0; iloop < 17; ++iloop) + { + jseed = j0 * k0; + j1 = (jseed / _m2 + j0 * k1 + j1 * k0) % (_m2 / 2); + j0 = jseed % _m2; + _m[iloop] = j0 + _m2 * j1; + } + _i = 4; + _j = 16; + } + } +} diff --git a/tests/src/JIT/Performance/CodeQuality/SciMark/SOR.cs b/tests/src/JIT/Performance/CodeQuality/SciMark/SOR.cs new file mode 100644 index 0000000..bc33c16 --- /dev/null +++ b/tests/src/JIT/Performance/CodeQuality/SciMark/SOR.cs @@ -0,0 +1,53 @@ +/// +/// This is a port of the SciMark2a Java Benchmark to C# by +/// Chris Re (cmr28@cornell.edu) and Werner Vogels (vogels@cs.cornell.edu) +/// +/// For details on the original authors see http://math.nist.gov/scimark2 +/// +/// This software is likely to burn your processor, bitflip your memory chips +/// anihilate your screen and corrupt all your disks, so you it at your +/// own risk. +/// + + +using System; + +namespace SciMark2 +{ + public class SOR + { + public static double num_flops(int M, int N, int num_iterations) + { + double Md = (double)M; + double Nd = (double)N; + double num_iterD = (double)num_iterations; + + return (Md - 1) * (Nd - 1) * num_iterD * 6.0; + } + + public static void execute(double omega, double[][] G, int num_iterations) + { + int M = G.Length; + int N = G[0].Length; + + double omega_over_four = omega * 0.25; + double one_minus_omega = 1.0 - omega; + + // update interior points + // + int Mm1 = M - 1; + int Nm1 = N - 1; + for (int p = 0; p < num_iterations; p++) + { + for (int i = 1; i < Mm1; i++) + { + double[] Gi = G[i]; + double[] Gim1 = G[i - 1]; + double[] Gip1 = G[i + 1]; + for (int j = 1; j < Nm1; j++) + Gi[j] = omega_over_four * (Gim1[j] + Gip1[j] + Gi[j - 1] + Gi[j + 1]) + one_minus_omega * Gi[j]; + } + } + } + } +} \ No newline at end of file diff --git a/tests/src/JIT/Performance/CodeQuality/SciMark/SciMark.csproj b/tests/src/JIT/Performance/CodeQuality/SciMark/SciMark.csproj new file mode 100644 index 0000000..6583073 --- /dev/null +++ b/tests/src/JIT/Performance/CodeQuality/SciMark/SciMark.csproj @@ -0,0 +1,55 @@ + + + + + Debug + AnyCPU + 2.0 + {95DFC527-4DC1-495E-97D7-E94EE1F7140D} + Exe + Properties + 512 + {786C830F-07A1-408B-BD7F-6EE04809D6DB};{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC} + $(ProgramFiles)\Common Files\microsoft shared\VSTT\11.0\UITestExtensionPackages + ..\..\ + 7a9bfb7d + SciMark + + + + + + pdbonly + true + + + + False + + + + + + + + + + + + + + + + + + + + + + $(JitPackagesConfigFileDirectory)benchmark\project.json + $(JitPackagesConfigFileDirectory)benchmark\project.lock.json + + + + + diff --git a/tests/src/JIT/Performance/CodeQuality/SciMark/SparseCompRow.cs b/tests/src/JIT/Performance/CodeQuality/SciMark/SparseCompRow.cs new file mode 100644 index 0000000..02bdea6 --- /dev/null +++ b/tests/src/JIT/Performance/CodeQuality/SciMark/SparseCompRow.cs @@ -0,0 +1,56 @@ +/// +/// This is a port of the SciMark2a Java Benchmark to C# by +/// Chris Re (cmr28@cornell.edu) and Werner Vogels (vogels@cs.cornell.edu) +/// +/// For details on the original authors see http://math.nist.gov/scimark2 +/// +/// This software is likely to burn your processor, bitflip your memory chips +/// anihilate your screen and corrupt all your disks, so you it at your +/// own risk. +/// + + +using System; + +namespace SciMark2 +{ + public class SparseCompRow + { + // multiple iterations used to make kernel + // have roughly same granulairty as other + // Scimark kernels + public static double num_flops(int N, int nz, int num_iterations) + { + /* Note that if nz does not divide N evenly, then the + actual number of nonzeros used is adjusted slightly. + */ + int actual_nz = (nz / N) * N; + return ((double)actual_nz) * 2.0 * ((double)num_iterations); + } + + /// + /// computes a matrix-vector multiply with a sparse matrix + /// held in compress-row format. If the size of the matrix + /// in MxN with nz nonzeros, then the val[] is the nz nonzeros, + /// with its ith entry in column col[i]. The integer vector row[] + /// is of size M+1 and row[i] points to the begining of the + /// ith row in col[]. + public static void matmult(double[] y, double[] val, int[] row, int[] col, double[] x, int NUM_ITERATIONS) + { + int M = row.Length - 1; + + for (int reps = 0; reps < NUM_ITERATIONS; reps++) + { + for (int r = 0; r < M; r++) + { + double sum = 0.0; + int rowR = row[r]; + int rowRp1 = row[r + 1]; + for (int i = rowR; i < rowRp1; i++) + sum += x[col[i]] * val[i]; + y[r] = sum; + } + } + } + } +} \ No newline at end of file diff --git a/tests/src/JIT/Performance/CodeQuality/SciMark/Stopwatch.cs b/tests/src/JIT/Performance/CodeQuality/SciMark/Stopwatch.cs new file mode 100644 index 0000000..b82dd5d --- /dev/null +++ b/tests/src/JIT/Performance/CodeQuality/SciMark/Stopwatch.cs @@ -0,0 +1,109 @@ +/// +/// This is a port of the SciMark2a Java Benchmark to C# by +/// Chris Re (cmr28@cornell.edu) and Werner Vogels (vogels@cs.cornell.edu) +/// +/// For details on the original authors see http://math.nist.gov/scimark2 +/// +/// This software is likely to burn your processor, bitflip your memory chips +/// anihilate your screen and corrupt all your disks, so you it at your +/// own risk. +/// + + +using System; + +namespace SciMark2 +{ + /// + /// Provides a stopwatch to measure elapsed time. + /// + /// + /// Roldan Pozo + /// + /// + /// 14 October 1997, revised 1999-04-24 + /// + /// + public class Stopwatch + { + private bool _running; + private double _last_time; + private double _total; + + /// + /// R + /// eturn system time (in seconds) + /// + public static double seconds() + { + return (System.DateTime.Now.Ticks * 1.0E-7); + } + + public virtual void reset() + { + _running = false; + _last_time = 0.0; + _total = 0.0; + } + + public Stopwatch() + { + reset(); + } + + /// + /// + /// Start (and reset) timer + /// + public virtual void start() + { + if (!_running) + { + _running = true; + _total = 0.0; + _last_time = seconds(); + } + } + + /// + /// + /// Resume timing, after stopping. (Does not wipe out accumulated times.) + /// + public virtual void resume() + { + if (!_running) + { + _last_time = seconds(); + _running = true; + } + } + + /// + /// + /// Stop timer + /// + public virtual double stop() + { + if (_running) + { + _total += seconds() - _last_time; + _running = false; + } + return _total; + } + + /// + /// + /// return the elapsed time (in seconds) + /// + public virtual double read() + { + if (_running) + { + _total += seconds() - _last_time; + _last_time = seconds(); + } + return _total; + } + } +} \ No newline at end of file diff --git a/tests/src/JIT/Performance/CodeQuality/SciMark/kernel.cs b/tests/src/JIT/Performance/CodeQuality/SciMark/kernel.cs new file mode 100644 index 0000000..5c827f0 --- /dev/null +++ b/tests/src/JIT/Performance/CodeQuality/SciMark/kernel.cs @@ -0,0 +1,444 @@ +/// +/// This is a port of the SciMark2a Java Benchmark to C# by +/// Chris Re (cmr28@cornell.edu) and Werner Vogels (vogels@cs.cornell.edu) +/// +/// For details on the original authors see http://math.nist.gov/scimark2 +/// +/// This software is likely to burn your processor, bitflip your memory chips +/// anihilate your screen and corrupt all your disks, so you it at your +/// own risk. +/// + + +using Microsoft.Xunit.Performance; +using System; + +namespace SciMark2 +{ + public static class kernel + { + [Benchmark] + public static void benchFFT() + { + SciMark2.Random R = new SciMark2.Random(Constants.RANDOM_SEED); + int N = Constants.FFT_SIZE; + long Iterations = 20000; + + double[] x = RandomVector(2 * N, R); + foreach (var iteration in Benchmark.Iterations) + { + using (iteration.StartMeasurement()) + { + innerFFT(x, Iterations); + } + } + validateFFT(N, x); + } + + private static void innerFFT(double[] x, long Iterations) + { + for (int i = 0; i < Iterations; i++) + { + FFT.transform(x); // forward transform + FFT.inverse(x); // backward transform + } + } + + private static void validateFFT(int N, double[] x) + { + const double EPS = 1.0e-10; + if (FFT.test(x) / N > EPS) + { + throw new Exception("FFT failed to validate"); + } + } + + public static double measureFFT(int N, double mintime, Random R) + { + // initialize FFT data as complex (N real/img pairs) + double[] x = RandomVector(2 * N, R); + + long cycles = 1; + Stopwatch Q = new Stopwatch(); + while (true) + { + Q.start(); + innerFFT(x, cycles); + Q.stop(); + if (Q.read() >= mintime) + break; + + cycles *= 2; + } + + validateFFT(N, x); + + // approx Mflops + return FFT.num_flops(N) * cycles / Q.read() * 1.0e-6; + } + + [Benchmark] + public static void benchSOR() + { + int N = Constants.SOR_SIZE; + SciMark2.Random R = new SciMark2.Random(Constants.RANDOM_SEED); + int Iterations = 20000; + double[][] G = RandomMatrix(N, N, R); + + foreach (var iteration in Benchmark.Iterations) + { + using (iteration.StartMeasurement()) + { + SOR.execute(1.25, G, Iterations); + } + } + } + + public static double measureSOR(int N, double min_time, Random R) + { + double[][] G = RandomMatrix(N, N, R); + Stopwatch Q = new Stopwatch(); + int cycles = 1; + while (true) + { + Q.start(); + SOR.execute(1.25, G, cycles); + Q.stop(); + if (Q.read() >= min_time) + break; + + cycles *= 2; + } + + // approx Mflops + return SOR.num_flops(N, N, cycles) / Q.read() * 1.0e-6; + } + + [Benchmark] + public static void benchMonteCarlo() + { + SciMark2.Random R = new SciMark2.Random(Constants.RANDOM_SEED); + int Iterations = 40000000; + foreach (var iteration in Benchmark.Iterations) + { + using (iteration.StartMeasurement()) + { + MonteCarlo.integrate(Iterations); + } + } + } + + public static double measureMonteCarlo(double min_time, Random R) + { + Stopwatch Q = new Stopwatch(); + + int cycles = 1; + while (true) + { + Q.start(); + MonteCarlo.integrate(cycles); + Q.stop(); + if (Q.read() >= min_time) + break; + + cycles *= 2; + } + + // approx Mflops + return MonteCarlo.num_flops(cycles) / Q.read() * 1.0e-6; + } + + [Benchmark] + public static void benchSparseMult() + { + int N = Constants.SPARSE_SIZE_M; + int nz = Constants.SPARSE_SIZE_nz; + int Iterations = 100000; + SciMark2.Random R = new SciMark2.Random(Constants.RANDOM_SEED); + + double[] x = RandomVector(N, R); + double[] y = new double[N]; + int nr = nz / N; // average number of nonzeros per row + int anz = nr * N; // _actual_ number of nonzeros + double[] val = RandomVector(anz, R); + int[] col = new int[anz]; + int[] row = new int[N + 1]; + + row[0] = 0; + for (int r = 0; r < N; r++) + { + // initialize elements for row r + + int rowr = row[r]; + row[r + 1] = rowr + nr; + int step = r / nr; + if (step < 1) + step = 1; + // take at least unit steps + + for (int i = 0; i < nr; i++) + col[rowr + i] = i * step; + } + + foreach (var iteration in Benchmark.Iterations) + { + using (iteration.StartMeasurement()) + { + SparseCompRow.matmult(y, val, row, col, x, Iterations); + } + } + } + + public static double measureSparseMatmult(int N, int nz, double min_time, Random R) + { + // initialize vector multipliers and storage for result + // y = A*y; + + double[] x = RandomVector(N, R); + double[] y = new double[N]; + + // initialize square sparse matrix + // + // for this test, we create a sparse matrix wit M/nz nonzeros + // per row, with spaced-out evenly between the begining of the + // row to the main diagonal. Thus, the resulting pattern looks + // like + // +-----------------+ + // +* + + // +*** + + // +* * * + + // +** * * + + // +** * * + + // +* * * * + + // +* * * * + + // +* * * * + + // +-----------------+ + // + // (as best reproducible with integer artihmetic) + // Note that the first nr rows will have elements past + // the diagonal. + + int nr = nz / N; // average number of nonzeros per row + int anz = nr * N; // _actual_ number of nonzeros + + + double[] val = RandomVector(anz, R); + int[] col = new int[anz]; + int[] row = new int[N + 1]; + + row[0] = 0; + for (int r = 0; r < N; r++) + { + // initialize elements for row r + + int rowr = row[r]; + row[r + 1] = rowr + nr; + int step = r / nr; + if (step < 1) + step = 1; + // take at least unit steps + + for (int i = 0; i < nr; i++) + col[rowr + i] = i * step; + } + + Stopwatch Q = new Stopwatch(); + + int cycles = 1; + while (true) + { + Q.start(); + SparseCompRow.matmult(y, val, row, col, x, cycles); + Q.stop(); + if (Q.read() >= min_time) + break; + + cycles *= 2; + } + + // approx Mflops + return SparseCompRow.num_flops(N, nz, cycles) / Q.read() * 1.0e-6; + } + + [Benchmark] + public static void benchmarkLU() + { + int N = Constants.LU_SIZE; + SciMark2.Random R = new SciMark2.Random(Constants.RANDOM_SEED); + int Iterations = 2000; + + double[][] A = RandomMatrix(N, N, R); + double[][] lu = new double[N][]; + for (int i = 0; i < N; i++) + { + lu[i] = new double[N]; + } + int[] pivot = new int[N]; + + foreach (var iteration in Benchmark.Iterations) + { + using (iteration.StartMeasurement()) + { + for (int i = 0; i < Iterations; i++) + { + CopyMatrix(lu, A); + LU.factor(lu, pivot); + } + } + } + + validateLU(N, R, lu, A, pivot); + } + + public static void validateLU(int N, SciMark2.Random R, double[][] lu, double[][] A, int[] pivot) + { + // verify that LU is correct + double[] b = RandomVector(N, R); + double[] x = NewVectorCopy(b); + + LU.solve(lu, pivot, x); + + const double EPS = 1.0e-12; + if (normabs(b, matvec(A, x)) / N > EPS) + { + throw new Exception("LU failed to validate"); + } + } + public static double measureLU(int N, double min_time, Random R) + { + // compute approx Mlfops, or O if LU yields large errors + + double[][] A = RandomMatrix(N, N, R); + double[][] lu = new double[N][]; + for (int i = 0; i < N; i++) + { + lu[i] = new double[N]; + } + int[] pivot = new int[N]; + + Stopwatch Q = new Stopwatch(); + + int cycles = 1; + while (true) + { + Q.start(); + for (int i = 0; i < cycles; i++) + { + CopyMatrix(lu, A); + LU.factor(lu, pivot); + } + Q.stop(); + if (Q.read() >= min_time) + break; + + cycles *= 2; + } + + validateLU(N, R, lu, A, pivot); + + return LU.num_flops(N) * cycles / Q.read() * 1.0e-6; + } + + private static double[] NewVectorCopy(double[] x) + { + int N = x.Length; + + double[] y = new double[N]; + for (int i = 0; i < N; i++) + y[i] = x[i]; + + return y; + } + + private static void CopyVector(double[] B, double[] A) + { + int N = A.Length; + + for (int i = 0; i < N; i++) + B[i] = A[i]; + } + + private static double normabs(double[] x, double[] y) + { + int N = x.Length; + double sum = 0.0; + + for (int i = 0; i < N; i++) + sum += System.Math.Abs(x[i] - y[i]); + + return sum; + } + + private static void CopyMatrix(double[][] B, double[][] A) + { + int M = A.Length; + int N = A[0].Length; + + int remainder = N & 3; // N mod 4; + + for (int i = 0; i < M; i++) + { + double[] Bi = B[i]; + double[] Ai = A[i]; + for (int j = 0; j < remainder; j++) + Bi[j] = Ai[j]; + for (int j = remainder; j < N; j += 4) + { + Bi[j] = Ai[j]; + Bi[j + 1] = Ai[j + 1]; + Bi[j + 2] = Ai[j + 2]; + Bi[j + 3] = Ai[j + 3]; + } + } + } + + private static double[][] RandomMatrix(int M, int N, Random R) + { + double[][] A = new double[M][]; + for (int i = 0; i < M; i++) + { + A[i] = new double[N]; + } + + for (int i = 0; i < N; i++) + for (int j = 0; j < N; j++) + A[i][j] = R.nextDouble(); + return A; + } + + private static double[] RandomVector(int N, Random R) + { + double[] A = new double[N]; + + for (int i = 0; i < N; i++) + A[i] = R.nextDouble(); + return A; + } + + private static double[] matvec(double[][] A, double[] x) + { + int N = x.Length; + double[] y = new double[N]; + + matvec(A, x, y); + + return y; + } + + private static void matvec(double[][] A, double[] x, double[] y) + { + int M = A.Length; + int N = A[0].Length; + + for (int i = 0; i < M; i++) + { + double sum = 0.0; + double[] Ai = A[i]; + for (int j = 0; j < N; j++) + sum += Ai[j] * x[j]; + + y[i] = sum; + } + } + } +} -- 2.7.4