From af4e4f92dc9e5da160eb6c6870a7b38b863f1c6c Mon Sep 17 00:00:00 2001
From: Joseph Redmon <pjreddie@gmail.com>
Date: Tue, 28 Oct 2014 02:45:06 +0000
Subject: [PATCH] getting rid of sub_arrays, nvidia driver memory leak
---
src/mini_blas.h | 54 +++++++++++++++++++++++++++++++++++++++++++++---------
1 files changed, 45 insertions(+), 9 deletions(-)
diff --git a/src/mini_blas.h b/src/mini_blas.h
index 34f15de..5d5e715 100644
--- a/src/mini_blas.h
+++ b/src/mini_blas.h
@@ -1,3 +1,5 @@
+#include "opencl.h"
+
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,
@@ -6,15 +8,47 @@
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,
- 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);
+
+#ifdef GPU
+void axpy_ongpu(int N, float ALPHA, cl_mem X, int INCX, cl_mem Y, int INCY);
+void copy_ongpu(int N, cl_mem X, int INCX, cl_mem Y, int INCY);
+void scal_ongpu(int N, float ALPHA, cl_mem X, int INCX);
+void im2col_ongpu(cl_mem data_im, int batch,
+ int channels, int height, int width,
+ int ksize, int stride, int pad, cl_mem data_col);
+
+void col2im_gpu(float *data_col, int batch,
+ int channels, int height, int width,
+ int ksize, int stride, int pad, float *data_im);
+void col2im_ongpu(cl_mem data_col, int batch,
+ int channels, int height, int width,
+ int ksize, int stride, int pad, cl_mem data_im);
+
+void im2col_gpu(float *data_im, int batch,
+ int channels, int height, int width,
+ int ksize, int stride, int pad, float *data_col);
+
+void gemm_ongpu_offset(int TA, int TB, int M, int N, int K, float ALPHA,
+ cl_mem A_gpu, int a_off, int lda,
+ cl_mem B_gpu, int b_off, int ldb,
+ float BETA,
+ cl_mem C_gpu, int c_off, int ldc);
+
+void gemm_ongpu(int TA, int TB, int M, int N, int K, float ALPHA,
+ cl_mem A_gpu, int lda,
+ cl_mem B_gpu, int ldb,
+ float BETA,
+ cl_mem C_gpu, int ldc);
+#endif
+
+void im2col_cpu(float* data_im, int batch,
+ int channels, int height, int width,
+ int ksize, int stride, int pad, float* data_col);
+
+void col2im_cpu(float* data_col, int batch,
+ int channels, int height, int width,
+ int ksize, int stride, int pad, float* data_im);
+
void test_blas();
void gemm_gpu(int TA, int TB, int M, int N, int K, float ALPHA,
@@ -28,5 +62,7 @@
float BETA,
float *C, int ldc);
void axpy_cpu(int N, float ALPHA, float *X, int INCX, float *Y, int INCY);
+void copy_cpu(int N, float *X, int INCX, float *Y, int INCY);
void scal_cpu(int N, float ALPHA, float *X, int INCX);
+float dot_cpu(int N, float *X, int INCX, float *Y, int INCY);
void test_gpu_blas();
--
Gitblit v1.10.0