GPU Kernel Information Aggregated by Name
| kernel_name | kernel_count | kernel_duration (us) | model_duration_percentage | kernel_flops | kernel_dram_read_bytes | kernel_dram_write_bytes | kernel_achieved_occupancy (%) | kernel_arithmetic_intensity (flops/byte) | kernel_arithmetic_throughput (GFlops) | kernel_memory_bound |
|---|
kernel_name | kernel_count | kernel_duration (us) | model_duration_percentage | kernel_flops | kernel_dram_read_bytes | kernel_dram_write_bytes | kernel_achieved_occupancy (%) | kernel_arithmetic_intensity (flops/byte) | kernel_arithmetic_throughput (GFlops) | kernel_memory_bound |
|---|---|---|---|---|---|---|---|---|---|---|
| void cudnn::detail::implicit_convolve_sgemm<float, float, 1024, 5, 5, 3, 3, 3, 1, true, false, true>(int, int, int, float const*, int, float*, float*, kernel_conv_params, int, float, float, int, float*, float*, int, int) | 19 | 388.67 | 14.65 | 0 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | true |
| void cudnn::detail::implicit_convolve_sgemm<float, float, 128, 5, 5, 3, 3, 3, 1, true, false, true>(int, int, int, float const*, int, float*, float*, kernel_conv_params, int, float, float, int, float*, float*, int, int) | 1 | 20.67 | 0.78 | 0 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | true |
| void cudnn::detail::pooling_fw_4d_kernel<float, float, cudnn::detail::averpooling_func<float>, 1, false>(cudnnTensorStruct, float const*, cudnnTensorStruct, float*, cudnnPoolingStruct, float, float, int, cudnn::reduced_divisor, cudnn::reduced_divisor) | 0 | 8.00 | 0.30 | 0 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | true |
| void cudnn::detail::pooling_fw_4d_kernel<float, float, cudnn::detail::maxpooling_func<float, (cudnnNanPropagation_t)0>, 0, false>(cudnnTensorStruct, float const*, cudnnTensorStruct, float*, cudnnPoolingStruct, float, float, int, cudnn::reduced_divisor, cudnn::reduced_divisor) | 2 | 21.00 | 0.79 | 0 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | true |
| void cudnn::winograd::generateWinogradTilesKernel<0, float, float>(cudnn::winograd::GenerateWinogradTilesParams<float, float>) | 10 | 53.00 | 2.00 | 0 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | true |
| void mshadow::cuda::MapPlanKernel<mshadow::sv::saveto, 8, mshadow::expr::Plan<mshadow::expr::SliceExp<mshadow::Tensor<mshadow::gpu, 3, float>, mshadow::gpu, float, 3, 2>, float>, mshadow::expr::Plan<mshadow::Tensor<mshadow::gpu, 3, float>, float> >(mshadow::expr::Plan<mshadow::expr::SliceExp<mshadow::Tensor<mshadow::gpu, 3, float>, mshadow::gpu, float, 3, 2>, float>, int, mshadow::Shape<2>, mshadow::expr::Plan<mshadow::Tensor<mshadow::gpu, 3, float>, float>) | 15 | 68.15 | 2.57 | 0 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | true |
| void mxnet::op::mxnet_op::mxnet_generic_kernel<mxnet::op::mxnet_op::op_with_req<mxnet::op::mshadow_op::identity, 1>, float*, float*>(int, float*, float*) | 0 | 4.00 | 0.15 | 0 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | true |
| void op_generic_tensor_kernel<2, float, float, float, 256, (cudnnGenericOp_t)0, (cudnnNanPropagation_t)0, (cudnnDimOrder_t)0, 0>(cudnnTensorStruct, float*, cudnnTensorStruct, float const*, cudnnTensorStruct, float const*, float, float, float, float, dimArray, reducedDivisorArray, bool) | 27 | 133.17 | 5.02 | 0 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | true |
| void op_generic_tensor_kernel<2, float, float, float, 256, (cudnnGenericOp_t)8, (cudnnNanPropagation_t)0, (cudnnDimOrder_t)0, 1>(cudnnTensorStruct, float*, cudnnTensorStruct, float const*, cudnnTensorStruct, float const*, float, float, float, float, dimArray, reducedDivisorArray, bool) | 33 | 147.67 | 5.57 | 0 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | true |
| volta_scudnn_winograd_128x128_ldg1_ldg4_relu_tile148t_nt_v1 | 7 | 138.33 | 5.22 | 0 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | true |
Showing 1 to 10 of 10 entries