From 5ef74c2031a040f30a670dc7d60790fc6a9ec720 Mon Sep 17 00:00:00 2001
From: Joseph Redmon <pjreddie@gmail.com>
Date: Fri, 02 May 2014 22:20:34 +0000
Subject: [PATCH] Slowly refactoring and pushing to GPU

---
 src/mini_blas.h |   41 +++++++++++++++++++++++++++++++----------
 1 files changed, 31 insertions(+), 10 deletions(-)

diff --git a/src/mini_blas.h b/src/mini_blas.h
index 46a37d3..34f15de 100644
--- a/src/mini_blas.h
+++ b/src/mini_blas.h
@@ -1,11 +1,32 @@
-void pm(int M, int N, double *A);
-void gemm(int TA, int TB, int M, int N, int K, double ALPHA, 
-                    double *A, int lda, 
-                    double *B, int ldb,
-                    double BETA,
-                    double *C, int ldc);
-void im2row(double *image, int h, int w, int c, int size, int stride, double *matrix);
-void im2col(double *image, int h, int w, int c, int size, int stride, double *matrix);
-void im2col_cpu(double* data_im, const int channels,
+void pm(int M, int N, float *A);
+void gemm(int TA, int TB, int M, int N, int K, float ALPHA, 
+                    float *A, int lda, 
+                    float *B, int ldb,
+                    float BETA,
+                    float *C, int ldc);
+float *random_matrix(int rows, int cols);
+void time_random_matrix(int TA, int TB, int m, int k, int n);
+void im2col_gpu(float* data_im, const int channels,
         const int height, const int width, const int ksize, const int stride,
-        double* data_col);
+        float* data_col);
+void im2col_cpu(float* data_im, const int channels,
+        const int height, const int width, const int ksize, const int stride,
+        float* data_col);
+void col2im_cpu(float* data_col, const int channels,
+        const int height, const int width, const int ksize, const int stride,
+        float* data_im);
+void test_blas();
+
+void gemm_gpu(int TA, int TB, int M, int N, int K, float ALPHA, 
+        float *A, int lda, 
+        float *B, int ldb,
+        float BETA,
+        float *C, int ldc);
+void gemm_cpu(int TA, int TB, int M, int N, int K, float ALPHA, 
+                    float *A, int lda, 
+                    float *B, int ldb,
+                    float BETA,
+                    float *C, int ldc);
+void axpy_cpu(int N, float ALPHA, float *X, int INCX, float *Y, int INCY);
+void scal_cpu(int N, float ALPHA, float *X, int INCX);
+void test_gpu_blas();

--
Gitblit v1.10.0