diff options
Diffstat (limited to 'cuda-kernels/scripts')
| -rwxr-xr-x | cuda-kernels/scripts | 55 |
1 files changed, 55 insertions, 0 deletions
diff --git a/cuda-kernels/scripts b/cuda-kernels/scripts new file mode 100755 index 0000000..b9a3b50 --- /dev/null +++ b/cuda-kernels/scripts @@ -0,0 +1,55 @@ +#!/bin/bash +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp4_16 v4p_genericMatrixMultiply.cu -DSIZE=16 +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp4_32 v4p_genericMatrixMultiply.cu -DSIZE=32 +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp4_64 v4p_genericMatrixMultiply.cu -DSIZE=64 +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp4_128 v4p_genericMatrixMultiply.cu -DSIZE=128 +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp4_256 v4p_genericMatrixMultiply.cu -DSIZE=256 +./vp4_16>vp4_log16 +./vp4_32>vp4_log32 +./vp4_64>vp4_log64 +./vp4_128>vp4_log128 +./vp4_256>vp4_log256 + +grep "kernel_name.*vp_example" -A 589 vp4_log16 >vp4_16_summary +grep "kernel_name.*vp_example" -A 589 vp4_log32 >vp4_32_summary +grep "kernel_name.*vp_example" -A 589 vp4_log64 >vp4_64_summary +grep "kernel_name.*vp_example" -A 589 vp4_log128 >vp4_128_summary +grep "kernel_name.*vp_example" -A 589 vp4_log256 >vp4_256_summary + + +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp8_16 v8p_genericMatrixMultiply.cu -DSIZE=16 +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp8_32 v8p_genericMatrixMultiply.cu -DSIZE=32 +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp8_64 v8p_genericMatrixMultiply.cu -DSIZE=64 +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp8_128 v8p_genericMatrixMultiply.cu -DSIZE=128 +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp8_256 v8p_genericMatrixMultiply.cu -DSIZE=256 +./vp8_16 >vp8_log16 +./vp8_32 >vp8_log32 +./vp8_64 >vp8_log64 +./vp8_128>vp8_log128 +./vp8_256>vp8_log256 + +grep "kernel_name.*vp_example" -A 589 vp8_log16 >vp8_16_summary +grep "kernel_name.*vp_example" -A 589 vp8_log32 >vp8_32_summary +grep "kernel_name.*vp_example" -A 589 vp8_log64 >vp8_64_summary +grep "kernel_name.*vp_example" -A 589 vp8_log128 >vp8_128_summary +grep "kernel_name.*vp_example" -A 589 vp8_log256 >vp8_256_summary + +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp16_16 v16p_genericMatrixMultiply.cu -DSIZE=16 +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp16_32 v16p_genericMatrixMultiply.cu -DSIZE=32 +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp16_64 v16p_genericMatrixMultiply.cu -DSIZE=64 +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp16_128 v16p_genericMatrixMultiply.cu -DSIZE=128 +nvcc --gpu-architecture=compute_70 --gpu-code=compute_70 -lcudart -g -o vp16_256 v16p_genericMatrixMultiply.cu -DSIZE=256 +./vp16_16>vp16_log16 +./vp16_32>vp16_log32 +./vp16_64>vp16_log64 +./vp16_128>vp16_log128 +./vp16_256>vp16_log256 + +grep "kernel_name.*vp_example" -A 589 vp16_log16 >vp16_16_summary +grep "kernel_name.*vp_example" -A 589 vp16_log32 >vp16_32_summary +grep "kernel_name.*vp_example" -A 589 vp16_log64 >vp16_64_summary +grep "kernel_name.*vp_example" -A 589 vp16_log128 >vp16_128_summary +grep "kernel_name.*vp_example" -A 589 vp16_log256 >vp16_256_summary + +mv vp* VPlog/ +make clean |