Merge pull request #6 from hgb-bin-proteomics/develop

verbosity in Cuda Batched
hgb-bin-proteomics · Jul 24, 2023 · 645a6a1 · 645a6a1
2 parents 8aa8660 + 25afc39
commit 645a6a1
Show file tree

Hide file tree

Showing 8 changed files with 33 additions and 12 deletions.
diff --git a/DataLoader/Compare.cs b/DataLoader/Compare.cs
@@ -118,7 +118,7 @@ public static int Compare(int nrCandidates, int nrSpectra, int topN, Random r)
                                                           sValuesPtr, sIdxPtr,
                                                           csrRowoffsets.Length, csrIdx.Length,
                                                           spectraValues.Length, spectraIdx.Length,
-                                                          topN, (float) 0.02, NORMALIZE, USE_GAUSSIAN);
+                                                          topN, (float) 0.02, NORMALIZE, USE_GAUSSIAN, 0);
 
                 Marshal.Copy(resultCuda, resultArrayCuda, 0, spectraIdx.Length * topN);
 
@@ -132,7 +132,7 @@ public static int Compare(int nrCandidates, int nrSpectra, int topN, Random r)
                                                                   sValuesPtr, sIdxPtr,
                                                                   csrRowoffsets.Length, csrIdx.Length,
                                                                   spectraValues.Length, spectraIdx.Length,
-                                                                  topN, (float) 0.02, NORMALIZE, USE_GAUSSIAN, BATCH_SIZE);
+                                                                  topN, (float) 0.02, NORMALIZE, USE_GAUSSIAN, BATCH_SIZE, 0);
 
                 Marshal.Copy(resultCudaB, resultArrayCudaB, 0, spectraIdx.Length * topN);
 

diff --git a/DataLoader/Cuda.cs b/DataLoader/Cuda.cs
@@ -12,7 +12,8 @@ private static extern IntPtr findTopCandidatesCuda(IntPtr cR, IntPtr cI,
                                                            int cRL, int cNNZ,
                                                            int sVL, int sIL,
                                                            int n, float tolerance,
-                                                           bool normalize, bool gaussianTol);
+                                                           bool normalize, bool gaussianTol,
+                                                           int verbose);
 
         [DllImport(dllCuda, CallingConvention = CallingConvention.Cdecl)]
         private static extern IntPtr findTopCandidatesCudaBatched(IntPtr cR, IntPtr cI,
@@ -21,7 +22,8 @@ private static extern IntPtr findTopCandidatesCudaBatched(IntPtr cR, IntPtr cI,
                                                                   int sVL, int sIL,
                                                                   int n, float tolerance,
                                                                   bool normalize, bool gaussianTol,
-                                                                  int batchSize);
+                                                                  int batchSize,
+                                                                  int verbose);
 
         [DllImport(dllCuda, CallingConvention = CallingConvention.Cdecl)]
         private static extern int releaseMemoryCuda(IntPtr result);
@@ -108,7 +110,8 @@ public static int Cuda(int nrCandidates, int nrSpectra, int topN, Random r, bool
                                                           csrRowoffsets.Length, csrIdx.Length,
                                                           spectraValues.Length, spectraIdx.Length,
                                                           topN, (float) 0.02,
-                                                          NORMALIZE, USE_GAUSSIAN);
+                                                          NORMALIZE, USE_GAUSSIAN,
+                                                          1000);
 
                     Marshal.Copy(result, resultArray, 0, spectraIdx.Length * topN);
 
@@ -127,7 +130,8 @@ public static int Cuda(int nrCandidates, int nrSpectra, int topN, Random r, bool
                                                                  spectraValues.Length, spectraIdx.Length,
                                                                  topN, (float) 0.02,
                                                                  NORMALIZE, USE_GAUSSIAN,
-                                                                 BATCH_SIZE);
+                                                                 BATCH_SIZE,
+                                                                 1000);
 
                     Marshal.Copy(result, resultArray, 0, spectraIdx.Length * topN);
 

diff --git a/DataLoader/DataLoader.cs b/DataLoader/DataLoader.cs
@@ -46,7 +46,7 @@ public static void Main(string[] args)
                 var status = Cuda(nrCandidates, nrSpectra, topN, r, false);
                 Console.WriteLine($"Cuda routine exited with status: {status}");
             }
-            if (mode == "CudaB")
+            else if (mode == "CudaB")
             {
                 var status = Cuda(nrCandidates, nrSpectra, topN, r, true);
                 Console.WriteLine($"Cuda routine exited with status: {status}");

diff --git a/DataLoader/DeterministicCompare.cs b/DataLoader/DeterministicCompare.cs
@@ -73,7 +73,8 @@ public static int DeterministicCompare()
                                                                  csrRowoffsets.Length, csrIdx.Length,
                                                                  spectraValues.Length, spectraIdx.Length,
                                                                  topN, (float) 0.0,
-                                                                 NORMALIZE, USE_GAUSSIAN, 2);
+                                                                 NORMALIZE, USE_GAUSSIAN, 2,
+                                                                 1000);
 
                 Marshal.Copy(resultCuda, resultArrayCuda, 0, spectraIdx.Length * topN);
 

diff --git a/VectorSearch/dllmain.cpp b/VectorSearch/dllmain.cpp
@@ -61,6 +61,7 @@ float normpdf(float, float, float);
 /// <param name="tolerance">Tolerance for peak matching (float).</param>
 /// <param name="normalize">If candidate vectors should be normalized to sum(elements) = 1 (bool).</param>
 /// <param name="gaussianTol">If spectrum peaks should be modelled as normal distributions or not (bool).</param>
+/// <param name="verbose">Print info every (int) processed spectra.</param>
 /// <returns>An integer array of length sILength * n containing the indexes of the top n candidates for each spectrum.</returns>
 int* findTopCandidates(int* candidatesValues, int* candidatesIdx, 
                        int* spectraValues, int* spectraIdx, 
@@ -158,6 +159,7 @@ int* findTopCandidates(int* candidatesValues, int* candidatesIdx,
 /// <param name="normalize">If candidate vectors should be normalized to sum(elements) = 1 (bool).</param>
 /// <param name="gaussianTol">If spectrum peaks should be modelled as normal distributions or not (bool).</param>
 /// <param name="batchSize">How many spectra (int) should be searched at once.</param>
+/// <param name="verbose">Print info every (int) processed spectra.</param>
 /// <returns>An integer array of length sILength * n containing the indexes of the top n candidates for each spectrum.</returns>
 int* findTopCandidatesBatched(int* candidatesValues, int* candidatesIdx,
                               int* spectraValues, int* spectraIdx,

diff --git a/VectorSearchCUDA/dllmain.cpp b/VectorSearchCUDA/dllmain.cpp
@@ -11,7 +11,7 @@
 
 const int versionMajor = 1;
 const int versionMinor = 3;
-const int versionFix = 3;
+const int versionFix = 4;
 
 #define METHOD_EXPORTS
 #ifdef METHOD_EXPORTS
@@ -31,14 +31,16 @@ extern "C" {
                                       int, int, 
                                       int, int,
                                       int, float,
-                                      bool, bool);
+                                      bool, bool,
+                                      int);
 
     EXPORT int* findTopCandidatesCudaBatched(int*, int*,
                                              int*, int*,
                                              int, int,
                                              int, int,
                                              int, float,
                                              bool, bool,
+                                             int,
                                              int);
 
     EXPORT int releaseMemoryCuda(int*);
@@ -86,13 +88,15 @@ int CHECK_CUSPARSE(cusparseStatus_t status)
 /// <param name="tolerance">Tolerance for peak matching (float).</param>
 /// <param name="normalize">If candidate vectors should be normalized to sum(elements) = 1 (bool).</param>
 /// <param name="gaussianTol">If spectrum peaks should be modelled as normal distributions or not (bool).</param>
+/// <param name="verbose">Print info every (int) processed spectra.</param>
 /// <returns>An integer array of length sILength * n containing the indexes of the top n candidates for each spectrum.</returns>
 int* findTopCandidatesCuda(int* csrRowoffsets, int* csrColIdx, 
                            int* spectraValues, int* spectraIdx,
                            int csrRowoffsetsLength, int csrNNZ,
                            int sVLength, int sILength,
                            int n, float tolerance,
-                           bool normalize, bool gaussianTol) {
+                           bool normalize, bool gaussianTol,
+                           int verbose) {
 
     if (n >= csrRowoffsetsLength) {
         throw std::invalid_argument("Cannot return more hits than number of candidates!");
@@ -201,6 +205,10 @@ int* findTopCandidatesCuda(int* csrRowoffsets, int* csrColIdx,
         }
 
         delete[] idx;
+
+        if (verbose != 0 && (i + 1) % verbose == 0) {
+            std::cout << "Searched " << i + 1 << " spectra in total..." << std::endl;
+        }
     }
 
     // device destroy descriptors
@@ -239,14 +247,16 @@ int* findTopCandidatesCuda(int* csrRowoffsets, int* csrColIdx,
 /// <param name="normalize">If candidate vectors should be normalized to sum(elements) = 1 (bool).</param>
 /// <param name="gaussianTol">If spectrum peaks should be modelled as normal distributions or not (bool).</param>
 /// <param name="batchSize">How many spectra (int) should be searched at once.</param>
+/// <param name="verbose">Print info every (int) processed spectra.</param>
 /// <returns>An integer array of length sILength * n containing the indexes of the top n candidates for each spectrum.</returns>
 int* findTopCandidatesCudaBatched(int* csrRowoffsets, int* csrColIdx,
                                   int* spectraValues, int* spectraIdx,
                                   int csrRowoffsetsLength, int csrNNZ,
                                   int sVLength, int sILength,
                                   int n, float tolerance,
                                   bool normalize, bool gaussianTol,
-                                  int batchSize) {
+                                  int batchSize,
+                                  int verbose) {
 
     if (n >= csrRowoffsetsLength) {
         throw std::invalid_argument("Cannot return more hits than number of candidates!");
@@ -462,6 +472,10 @@ int* findTopCandidatesCudaBatched(int* csrRowoffsets, int* csrColIdx,
         CHECK_CUDA(cudaFree(dM_csrValues));
         CHECK_CUDA(cudaFree(dspgemM_csrColIdx));
         CHECK_CUDA(cudaFree(dspgemM_csrValues));
+
+        if (verbose != 0 && (i + batchSize) % verbose == 0) {
+            std::cout << "Searched " << i + batchSize << " spectra in total..." << std::endl;
+        }
     }
 
     // device destroy descriptors

diff --git a/dll/VectorSearch.dll b/dll/VectorSearch.dll
diff --git a/dll/VectorSearchCUDA.dll b/dll/VectorSearchCUDA.dll