更新第三节

Author: YouQixiaowu

src(pyCUDA speeds up your python code)/HelloWorld.py renamed to src(pyCUDA speeds up your python code)/1_Hello_World.py

@@ -0,0 +1,12 @@
+import pycuda.autoinit
+from pycuda.compiler import SourceModule
+
+kernel_code = r"""
+__global__ void print_id(void)
+{
+    printf("blockIdx.x = %d; threadIdx.x = %d;\n", blockIdx.x, threadIdx.x);
+}
+"""
+mod = SourceModule(kernel_code)
+print_id = mod.get_function("print_id")
+print_id(grid=(2,), block=(3,1,1))

src(pyCUDA speeds up your python code)/2_Print_ID.py

@@ -0,0 +1,11 @@
+import numpy
+num = 4
+A = numpy.random.rand(num)
+B = numpy.random.rand(num)
+C = A + B
+print('A=', A)
+print('B=', B)
+print('C=', C)
+
+
+

src(pyCUDA speeds up your python code)/3_Array_Add.py

@@ -0,0 +1,14 @@
+import pycuda.autoinit
+import pycuda.gpuarray as gpuarray
+import numpy
+num = 4
+A = numpy.random.rand(num)
+B = numpy.random.rand(num)
+A_GPU = gpuarray.to_gpu(A.astype(numpy.float32))
+B_GPU = gpuarray.to_gpu(B.astype(numpy.float32))
+C_GPU = A_GPU + B_GPU
+C = C_GPU.get()
+print('A=', A)
+print('B=', B)
+print('C=', C)
+

src(pyCUDA speeds up your python code)/3_Array_Add_GPU_easy.py

@@ -0,0 +1,29 @@
+import pycuda.autoinit
+import pycuda.driver as drv
+from pycuda.compiler import SourceModule
+import numpy
+mod = SourceModule(r"""
+void __global__ add(const float *x, const float *y, float *z)
+{
+const int n = blockDim.x * blockIdx.x + threadIdx.x;
+z[n] = x[n] + y[n];
+}
+""")
+add = mod.get_function("add")
+num = 6
+A = numpy.random.rand(num).astype(numpy.float32)
+B = numpy.random.rand(num).astype(numpy.float32)
+C = numpy.zeros(num).astype(numpy.float32)
+A_GPU = drv.mem_alloc(A.nbytes)
+B_GPU = drv.mem_alloc(B.nbytes)
+C_GPU = drv.mem_alloc(C.nbytes)
+drv.memcpy_htod(A_GPU, A)
+drv.memcpy_htod(B_GPU, B)
+add(A_GPU, B_GPU, C_GPU, grid=(2, 1), block=(4,1,1))
+drv.memcpy_dtoh(C, C_GPU)
+A_GPU.free()
+B_GPU.free()
+C_GPU.free()
+print('A=', A)
+print('B=', B)
+print('C=', C)

src(pyCUDA speeds up your python code)/3_Array_Add_GPU_hard.py

@@ -0,0 +1,25 @@
+import pycuda.autoinit
+from pycuda.compiler import SourceModule
+import pycuda.gpuarray as gpuarray
+import numpy
+mod = SourceModule(r"""
+void __global__ add(const float *x, const float *y, float *z)
+{
+    const int n = blockDim.x * blockIdx.x + threadIdx.x;
+    z[n] = x[n] + y[n];
+}
+""")
+add = mod.get_function("add")
+num = 6
+A = numpy.random.rand(num)
+B = numpy.random.rand(num)
+C = numpy.zeros(num)
+A_GPU = gpuarray.to_gpu(A.astype(numpy.float32))
+B_GPU = gpuarray.to_gpu(B.astype(numpy.float32))
+C_GPU = gpuarray.to_gpu(B.astype(numpy.float32))
+add(A_GPU, B_GPU, C_GPU, grid=(2,), block=(4,1,1))
+C = C_GPU.get()
+print('A=', A)
+print('B=', B)
+print('C=', C)
+