Find research datasets worth reusing
Search datasets from major research repositories and use ShareScore to quickly assess how well each record supports discovery, access, and reuse.
107
datasets available to search
ShareScore release 0.7.1
Dataset results
107 results for “Energy Consumption”
Working time, energy throughput and value added embodied in production, consumption and trade by subsectors for the US, the EU, China and rest of the world (2011)
<p>This repository contains the data needed to reproduce the results in:</p> <p>Pérez-Sánchez, L., Velasco-Fernández, R., Giampietro, M., The international division of labor and embodied working time in trade for the US, the EU and China, Ecological Economics. <a href="http://doi.org/10.1016/j.ecolecon.2020.106909">https://doi.org/10.1016/j.ecolecon.2020.1069097</a></p> <p>Sources of data are specified in the dataset (under tab "references")</p> <p> </p>
MSP430FR5969 Basic Block Worst Case Energy Consumption (WCEC) and Worst Case Execution Time (WCET) dataset
<p>This dataset contains around 30 000 basic blocks whose energy consumption and execution time have been measured in isolation on the MSP430FR5969 microcontroller, at 1MHz. Basic blocks were executed in a worst case scenario regarding the MSP430 FRAM cache and CPU pipeline. The dataset creation process is described thoroughly in [1].</p> <h2>Folder structure</h2> <p>This dataset is composed of the following files:</p> <ul> <li><code>basic_blocks.tar.xz</code> contains all basic blocks (BB) used in the dataset, in a custom JSON format,</li> <li><code>data.csv</code>/<code>data.xlsx</code> contains the measured energy consumption and execution time for each basic block</li> </ul> <p>We first details how the <code>basic_blocks.tar.gz</code> archive is organized, and then present the CSV/XSLX spreadsheet format.</p> <h2>Basic Blocks</h2> <p>We extracted the basic blocks from a subset of programs of the <em>AnghaBench</em> benchmark suite [2]. The <code>basic_blocks.tar.gz</code> archive consist of the extracted basic blocks organized as <code>json</code> files. Each <code>json</code> file correspond to a C source file from <em>AnghaBench</em>, and is given a unique identifier. An example json (<code>137.json</code>) is available here:</p> <div> <pre><code>{ "extr_pfctl_altq.c_pfctl_altq_init": [ # Basic block 1 [ # Instruction 1 of BB1 [ "MOV.W", "#queue_map", "R13" ], # Instruction 2 of BB1 [ "MOV.B", "#0", "R14" ], # Instruction 3 of BB1 [ "CALL", "#hcreate_r", null ] ], # Basic block 2 [ .... ] ] }</code></pre> </div> <p>The json contains a dict with only one key pointing to an array of basic blocks. This key is the name of the original C source file in <em>AnghaBench</em> from which the basic blocks were extracted (here <a href="https://github.com/brenocfg/AnghaBench/blob/master/freebsd/sbin/pfctl/extr_pfctl_altq.c_pfctl_altq_init.c">extr_pfctl_altq.c_pfctl_altq_init.c</a>). The array contains severals basic blocks, which are represented as an array of instructions, which are themselves represented as an array <code>[OPCODE, OPERAND1, OPERAND2]</code>.</p> <p>Then, each basic block can be identified uniquely using two ids : its <em>file id</em> and its <em>offset</em> in the file (<code>id=<file id>_<offset></code>). In our example, the basic block 1 can be identified by the json file id (<code>137</code>) and its offset in the file (<code>0</code>). Its ID is <code>137_0</code>. This ID is used to make the mapping between a basic block and its energy consumption/execution time, with the <code>data.csv</code>/<code>data.xlsx</code> spreadsheet.</p> <h2>Energy Consumption and Execution Time</h2> <p>Energy consumption and execution time data are stored in the <code>data.csv</code> file. Here is the extract of the csv file corresponding to the basic block <code>137_0</code>. The spreadsheet format is described below.</p> <pre><code>bb_id;nb_inst;max_energy;max_time;avg_time;avg_energy;energy_per_inst;nb_samples;unroll_factor 137_0;3;8.77;7.08;7.04;8.21;2.92;40;50</code></pre> <p>Spreadsheet format :</p> <ul> <li><code>bb_id</code>: the unique identifier of a basic block (cf. <a>Basic Blocks</a>)</li> <li><code>nb_inst</code>: the number of instructions in the basic block</li> <li><code>max_energy</code>: the maximum energy comsumption (in nJ) measured during the experiment</li> <li><code>max_time</code>: the maximum execution time (in us) measured during the experiment</li> <li><code>avg_time</code>: the average execution time (in us) measured during the experiment</li> <li><code>avg_energy</code>: the average energy comsumption (in nJ) measured during the experiment</li> <li><code>energy_per_inst</code>: the average energy consumption per instruction (correspond to <code>avg_energy</code>/<code>nb_inst</code>)</li> <li><code>nb_samples</code>: how much time the basic block energy consumption/execution time has been measured</li> <li><code>unroll_factor</code>: how much time the basic block was unrolled (cf Basic Block Unrolling)</li> </ul> <h3>Basic Block Unrolling</h3> <p>To measure the energy consumption and execution time of the msp430, we need to be able to handle the scale difference between the measurement tool and the basic block execution time. This is achieved by duplicating the basic block multiple times while making sure to keep the worst-case memory layout as explained in the paper. The number of time the basic block has been duplicated is called the <code>unroll_factor</code>.</p> <p>Values of energy and time are always given per basic block, so they have already been divided by the unroll factor.</p> <p> </p> <h2>Dataset description</h2> <h3>Features</h3> <p>The selected features after PCA analysis for both energy and time model are listed here: <strong> MOV.W_Rn_Rn, MOV.W_X(Rn)_X(Rn), CALL, MOV.B_#N_Rn, ADD.W_Rn_Rn, MOV.W_@Rn_Rn, MOV.W_X(Rn)_Rn, ADD.W_#N_Rn, PUSHM.W_#N_Rn, MOV.W_X(Rn)_ADDR, CMP.W_#N_Rn, MOV.W_&ADDR_X(Rn), MOV.W_Rn_X(Rn), BIS.W_Rn_Rn, RLAM.W_#N_Rn, SUB.W_#N_Rn, MOV.W_&ADDR_Rn, MOV.W_#N_X(Rn), CMP.W_Rn_Rn, BIT.W_ADDR_Rn, MOV.W_@Rn_X(Rn), ADD.W_#N_X(Rn), MOV.W_#N_Rn, AND.W_Rn_Rn, MOV.W_Rn_ADDR, SUB.W_Rn_Rn, MOV.W_ADDR_Rn, MOV.W_X(Rn)_&ADDR, MOV.W_ADDR_ADDR, JMP, ADD_#N_Rn, BIS.W_Rn_X(Rn), SUB_Rn_Rn, MOV.W_ADDR_X(Rn), ADDC_#N_X(Rn), MOV.B_Rn_Rn, CMP.W_X(Rn)_X(Rn), ADD_Rn_Rn, nb_inst, INV.W_Rn_, NOP__, ADD.W_X(Rn)_X(Rn), ADD.W_Rn_X(Rn), MOV.B_@Rn_Rn, BIS.W_X(Rn)_X(Rn), MOV.B_#N_X(Rn), MOV.W_#N_ADDR, AND.W_#N_ADDR, SUBC_X(Rn)_X(Rn), BIS.W_#N_X(Rn), SUB.W_X(Rn)_X(Rn), AND.B_#N_Rn, ADD_X(Rn)_X(Rn), MOV.W_@Rn_ADDR, MOV.W_&ADDR_ADDR, ADDC_Rn_Rn, AND.W_#N_X(Rn), SUB_#N_Rn, RRUM.W_#N_Rn, AND_ADDR_Rn, CMP.W_X(Rn)_ADDR, MOV.B_#N_ADDR, ADD.W_#N_ADDR, CMP.B_#N_Rn, SXT_Rn_, XOR.W_Rn_Rn, CMP.W_@Rn_Rn, ADD.W_@Rn_Rn, ADD.W_X(Rn)_Rn, AND.W_Rn_X(Rn), CMP.B_Rn_Rn, AND.W_X(Rn)_X(Rn), BIC.W_#N_Rn, BIS.W_#N_Rn, AND.B_#N_X(Rn), MOV.B_X(Rn)_X(Rn), AND.W_@Rn_Rn, MOV.W_#N_&ADDR, BIS.W_Rn_ADDR, SUB.W_X(Rn)_Rn, SUB.W_Rn_X(Rn), SUB_X(Rn)_X(Rn), MOV.B_@Rn_X(Rn), CMP.W_@Rn_X(Rn), ADD.W_X(Rn)_ADDR, CMP.W_Rn_X(Rn), BIS.W_@Rn_X(Rn), CMP.B_X(Rn)_X(Rn), RRC.W_Rn_, MOV.W_@Rn_&ADDR, CMP.W_#N_X(Rn), ADDC_X(Rn)_Rn, CMP.W_X(Rn)_Rn, BIS.W_X(Rn)_Rn, SUB_X(Rn)_Rn, MOV.B_X(Rn)_Rn, MOV.W_ADDR_&ADDR, AND.W_#N_Rn, RLA.W_Rn_, INV.W_X(Rn)_, XOR.W_#N_Rn, SUB.W_Rn_ADDR, BIC.W_#N_X(Rn), MOV.B_X(Rn)_ADDR, ADD_#N_X(Rn), SUB_Rn_X(Rn), MOV.B_&ADDR_Rn, MOV.W_Rn_&ADDR, ADD_X(Rn)_Rn, AND.W_X(Rn)_Rn, PUSHM.A_#N_Rn, RRAM.W_#N_Rn, AND.W_@Rn_X(Rn), BIS.B_Rn_X(Rn), SUB.W_@Rn_Rn, CLRC__, CMP.W_#N_ADDR, XOR.W_Rn_X(Rn), MOV.B_Rn_ADDR, CMP.B_X(Rn)_Rn, BIS.B_Rn_Rn, BIS.W_X(Rn)_ADDR, CMP.B_#N_X(Rn), CMP.W_Rn_ADDR, XOR.W_X(Rn)_Rn, MOV.B_Rn_X(Rn), ADD.B_#N_Rn</strong></p> <h3>Code</h3> <p>The trained machine learning model, tests, and local explanation code can be generated and found here: <a href="https://colab.research.google.com/drive/1zaCIkbKDRDtwzhY-r3Iff0OpOSkOUibW?usp=sharing" target="_blank" rel="noopener">WORTEX Machine learning code</a> </p> <h2>Acknowledgment</h2> <p>This work has received a French government support granted to the Labex CominLabs excellence laboratory and managed by the National Research Agency in the “Investing for the Future” program under reference ANR-10-LABX-07-01</p> <h2>Licensing</h2> <p>Copyright 2024 Hector Chabot Copyright 2024 Abderaouf Nassim Amalou Copyright 2024 Hugo Reymond Copyright 2024 Isabelle Puaut</p> <p>Licensed under the Creative Commons Attribution 4.0 International License</p> <h2>References</h2> <p>[1] Reymond, H., Amalou, A. N., Puaut, I. “WORTEX: Worst-Case Execution Time and Energy Estimation in Low-Power Microprocessors using Explainable ML” in 22nd International Workshop on Worst-Case Execution Time Analysis (WCET 2024) </p> <p>[2] Da Silva, Anderson Faustino, et al. “Anghabench: A suite with one million compilable C benchmarks for code-size reduction.” 2021 IEEE/ACM International Symposium on Code Generation and Optimization (CGO). IEEE, 2021.</p>
Investigating dynamics between energy use and socio-demographic characteristics in spatial modeling of residential energy consumption
<p>Files represent datasets (2017 Residential Building Stock Assessment and American Community Survey 2012-2017 5-year estimate) and R-code associated with the analysis. </p>
Block-wise sparse matrix-vector product dataset and convolutional neural nets for estimating the run time and energy consumption of the sparse matrix-vector product
<p><strong>Introduction</strong></p> <p><strong>SpMV-CNN</strong> is a set of Convolutional Neural Networks (CNNs) that provide accurate estimations of the performance and energy consumption of the SpMV kernel. The proposed CNN-based models use a block-wise approach to make the CNN architecture independent of the matrix size. These models cat be trained to estimate run time as well as total, package and DRAM energy consumption at different processor frequencies.</p> <p><strong>Prerequisites</strong></p> <p><strong>SpMV-CNN</strong> requires Python3 with the following packages:</p> <pre><code>keras==2.1.6 tensorflow==1.8.0 h5py==2.7.1 matplotlib==2.1.1 scikit-learn==0.19.1 </code></pre> <p><strong>Obtaining the dataset</strong></p> <p>The execution time and energy consumption data corresponding to the SpMV operation on a set of sparse matrices from the SuiteSparse Matrix Collection have been obtained on an Intel Xeon E5-2630 core running at frequencies 1.2, 1.6, 2.0, 2.4 GHz. The energy consumption measurements are obtained via the Intel RAPL interface and gathered at three different levels (total, package and DRAM, where total = package + DRAM) for this specific processor.</p> <p>The <code>spmv-cnn-dataset.tgz</code> archive contains the whole dataset, including the following HDF5 files:</p> <pre><code>$ tree . |-- test | |-- f_1200000_b250 | | |-- output_2cubes_sphere_1200000.h5 | | |-- output_apache2_1200000.h5 | | |-- output_bcsstk36_1200000.h5 | | |-- output_cfd1_1200000.h5 | | |-- output_cfd2_1200000.h5 | | |-- output_ct20stif_1200000.h5 | | |-- output_denormal_1200000.h5 | | |-- output_Dubcova2_1200000.h5 | | |-- output_Dubcova3_1200000.h5 | | |-- output_ecology2_1200000.h5 | | |-- output_gyro_1200000.h5 | | |-- output_gyro_k_1200000.h5 | | |-- output_msc10848_1200000.h5 | | |-- output_msc23052_1200000.h5 | | |-- output_nasasrb_1200000.h5 | | |-- output_nd3k_1200000.h5 | | |-- output_offshore_1200000.h5 | | |-- output_oilpan_1200000.h5 | | |-- output_olafu_1200000.h5 | | |-- output_parabolic_fem_1200000.h5 | | |-- output_qa8fm_1200000.h5 | | |-- output_raefsky4_1200000.h5 | | |-- output_s3dkq4m2_1200000.h5 | | |-- output_s3dkt3m2_1200000.h5 | | |-- output_ship_001_1200000.h5 | | |-- output_ship_003_1200000.h5 | | |-- output_shipsec1_1200000.h5 | | |-- output_shipsec5_1200000.h5 | | |-- output_shipsec8_1200000.h5 | | |-- output_smt_1200000.h5 | | |-- output_thermomech_dM_1200000.h5 | | |-- output_thread_1200000.h5 | | `-- output_vanbody_1200000.h5 | |-- f_1600000_b250 | | |-- output_2cubes_sphere_1600000.h5 | | |—- ... | | `-- output_vanbody_1600000.h5 | |-- f_2000000_b250 | | |-- output_2cubes_sphere_2000000.h5 | | |—- ... | | `-- output_vanbody_2000000.h5 | `-- f_2400000_b250 | |-- output_2cubes_sphere_2400000.h5 | |—- ... | `-- output_vanbody_2400000.h5 |-- test_pagerank | |-- f_1200000_b250 | | |-- output_adaptive_1200000.h5 | | |-- output_cit-HepPh_1200000.h5 | | |-- output_delaunay_n22_1200000.h5 | | |-- output_email-Enron_1200000.h5 | | |-- output_email-EuAll_1200000.h5 | | |-- output_europe_osm_1200000.h5 | | |-- output_hugebubbles-00020_1200000.h5 | | |-- output_rgg_n_2_24_s0_1200000.h5 | | |-- output_road_usa_1200000.h5 | | |-- output_Stanford_1200000.h5 | | |-- output_wb-edu_1200000.h5 | | |-- output_web-BerkStan_1200000.h5 | | |-- output_web-Google_1200000.h5 | | |-- output_web-NotreDame_1200000.h5 | | |-- output_wiki-Talk_1200000.h5 | | `-- output_wiki-Vote_1200000.h5 | |-- f_1600000_b250 | | |-- output_adaptive_1600000.h5 | | |—- ... | | `-- output_wiki-Vote_1600000.h5 | |-- f_2000000_b250 | | |-- output_adaptive_2000000.h5 | | |—- ... | | `-- output_wiki-Vote_2000000.h5 | `-- f_2400000_b250 | |-- output_adaptive_2400000.h5 | |—- ... | `-- output_wiki-Vote_2400000.h5 `-- train |-- merged_energy_train_shuffle_f1200000_250.h5 |-- merged_energy_train_shuffle_f1600000_250.h5 |-- merged_energy_train_shuffle_f2000000_250.h5 `-- merged_energy_train_shuffle_f2400000_250.h5 </code></pre> <p>The matrices contained in the merged training files (<code>merged_energy_train_shuffle_fXX00000_250.h5</code>) are the following:</p> <pre><code>$ tree . |-- output_af_0_k101_1200000.h5 |-- output_af_1_k101_1200000.h5 |-- output_af_2_k101_1200000.h5 |-- output_af_3_k101_1200000.h5 |-- output_af_4_k101_1200000.h5 |-- output_af_5_k101_1200000.h5 |-- output_af_shell10_1200000.h5 |-- output_af_shell1_1200000.h5 |-- output_af_shell2_1200000.h5 |-- output_af_shell3_1200000.h5 |-- output_af_shell4_1200000.h5 |-- output_af_shell5_1200000.h5 |-- output_af_shell6_1200000.h5 |-- output_af_shell7_1200000.h5 |-- output_af_shell8_1200000.h5 |-- output_af_shell9_1200000.h5 |-- output_atmosmodd_1200000.h5 |-- output_atmosmodj_1200000.h5 |-- output_atmosmodl_1200000.h5 |-- output_audikw_1_1200000.h5 |-- output_BenElechi1_1200000.h5 |-- output_bmw3_2_1200000.h5 |-- output_bmw7st_1_1200000.h5 |-- output_bmwcra_1_1200000.h5 |-- output_bone010_1200000.h5 |-- output_boneS01_1200000.h5 |-- output_boneS10_1200000.h5 |-- output_bundle_adj_1200000.h5 |-- output_cage14_1200000.h5 |-- output_cage15_1200000.h5 |-- output_circuit5M_1200000.h5 |-- output_circuit5M_dc_1200000.h5 |-- output_CO_1200000.h5 |-- output_consph_1200000.h5 |-- output_CoupCons3D_1200000.h5 |-- output_crankseg_1_1200000.h5 |-- output_crankseg_2_1200000.h5 |-- output_CurlCurl_2_1200000.h5 |-- output_CurlCurl_3_1200000.h5 |-- output_CurlCurl_4_1200000.h5 |-- output_dielFilterV2real_1200000.h5 |-- output_dielFilterV3real_1200000.h5 |-- output_Emilia_923_1200000.h5 |-- output_ESOC_1200000.h5 |-- output_F1_1200000.h5 |-- output_F2_1200000.h5 |-- output_Fault_639_1200000.h5 |-- output_Freescale1_1200000.h5 |-- output_Freescale2_1200000.h5 |-- output_FullChip_1200000.h5 |-- output_G3_circuit_1200000.h5 |-- output_Ga10As10H30_1200000.h5 |-- output_Ga19As19H42_1200000.h5 |-- output_Ga3As3H12_1200000.h5 |-- output_Ga41As41H72_1200000.h5 |-- output_Ge87H76_1200000.h5 |-- output_Ge99H100_1200000.h5 |-- output_Geo_1438_1200000.h5 |-- output_gsm_106857_1200000.h5 |-- output_Hardesty3_1200000.h5 |-- output_hood_1200000.h5 |-- output_Hook_1498_1200000.h5 |-- output_human_gene1_1200000.h5 |-- output_human_gene2_1200000.h5 |-- output_inline_1_1200000.h5 |-- output_JP_1200000.h5 |-- output_kkt_power_1200000.h5 |-- output_ldoor_1200000.h5 |-- output_Long_Coup_dt0_1200000.h5 |-- output_Long_Coup_dt6_1200000.h5 |-- output_mat_104_10000_1200000.h5 |-- output_mat_104_1000_1200000.h5 |-- output_mat_104_5000_1200000.h5 |-- output_mat_112_10000_1200000.h5 |-- output_mat_112_1000_1200000.h5 |-- output_mat_112_5000_1200000.h5 |-- output_mat_120_10000_1200000.h5 |-- output_mat_120_1000_1200000.h5 |-- output_mat_120_5000_1200000.h5 |-- output_mat_128_10000_1200000.h5 |-- output_mat_128_1000_1200000.h5 |-- output_mat_128_5000_1200000.h5 |-- output_mat_16_10000_1200000.h5 |-- output_mat_16_1000_1200000.h5 |-- output_mat_16_5000_1200000.h5 |-- output_mat_24_10000_1200000.h5 |-- output_mat_24_1000_1200000.h5 |-- output_mat_24_5000_1200000.h5 |-- output_mat_32_10000_1200000.h5 |-- output_mat_32_1000_1200000.h5 |-- output_mat_32_5000_1200000.h5 |-- output_mat_40_10000_1200000.h5 |-- output_mat_40_1000_1200000.h5 |-- output_mat_40_5000_1200000.h5 |-- output_mat_48_10000_1200000.h5 |-- output_mat_48_1000_1200000.h5 |-- output_mat_48_5000_1200000.h5 |-- output_mat_56_10000_1200000.h5 |-- output_mat_56_1000_1200000.h5 |-- output_mat_56_5000_1200000.h5 |-- output_mat_64_10000_1200000.h5 |-- output_mat_64_1000_1200000.h5 |-- output_mat_64_5000_1200000.h5 |-- output_mat_72_10000_1200000.h5 |-- output_mat_72_1000_1200000.h5 |-- output_mat_72_5000_1200000.h5 |-- output_mat_80_10000_1200000.h5 |-- output_mat_80_1000_1200000.h5 |-- output_mat_80_5000_1200000.h5 |-- output_mat_8_10000_1200000.h5 |-- output_mat_8_1000_1200000.h5 |-- output_mat_8_5000_1200000.h5 |-- output_mat_88_10000_1200000.h5 |-- output_mat_88_1000_1200000.h5 |-- output_mat_88_5000_1200000.h5 |-- output_mat_96_10000_1200000.h5 |-- output_mat_96_1000_1200000.h5 |-- output_mat_96_5000_1200000.h5 |-- output_memchip_1200000.h5 |-- output_ML_Laplace_1200000.h5 |-- output_mouse_gene_1200000.h5 |-- output_msdoor_1200000.h5 |-- output_m_t1_1200000.h5 |-- output_nd12k_1200000.h5 |-- output_nd24k_1200000.h5 |-- output_nd6k_1200000.h5 |-- output_nlpkkt120_1200000.h5 |-- output_nlpkkt80_1200000.h5 |-- output_PFlow_742_1200000.h5 |-- output_pwtk_1200000.h5 |-- output_rajat31_1200000.h5 |-- output_RM07R_1200000.h5 |-- output_Rucci1_1200000.h5 |-- output_Serena_1200000.h5 |-- output_Si34H36_1200000.h5 |-- output_Si41Ge41H72_1200000.h5 |-- output_Si87H76_1200000.h5 |-- output_SiO2_1200000.h5 |-- output_sls_1200000.h5 |-- output_StocF-1465_1200000.h5 |-- output_TEM152078_1200000.h5 |-- output_TEM181302_1200000.h5 |-- output_thermal2_1200000.h5 |-- output_tmt_sym_1200000.h5 |-- output_torso1_1200000.h5 |-- output_Transport_1200000.h5 |-- output_TSOPF_FS_b300_c2_1200000.h5 |-- output_TSOPF_FS_b300_c3_1200000.h5 |-- output_TSOPF_RS_b2383_1200000.h5 |-- output_TSOPF_RS_b2383_c1_1200000.h5 |-- output_TSOPF_RS_b678_c2_1200000.h5 `-- output_x104_1200000.h5 f_1600000_b250 |-- output_af_0_k101_1600000.h5 |—- ... `-- output_x104_1600000.h5 f_2000000_b250 |-- output_af_0_k101_2000000.h5 |—- ... `-- output_x104_2000000.h5 f_2400000_b250 |-- output_af_0_k101_2400000.h5 |—- ... `-- output_x104_2400000.h5 </code></pre> <p><strong>Creating your own dataset</strong></p> <p>If you wish to create your own training/testing dataset on a different target architecture you need to take the following steps:</p> <ol> <li> <p>Build the SpMV driver:</p> <ol> <li> <p>Go to <code>cd SpMV-driver/src</code></p> </li> <li> <p>Edit makefile and set the PAPI and HDF5 install prefixes.</p> </li> <li> <p>Build the driver via <code>make.</code></p> </li> </ol> </li> <li> <p>Run the SpMV driver: </p> <p><code>./driver <arg0> <arg1> ...</code></p> <p>List of driver arguments:</p> <pre><code>matrix = audikw_1.rb # Input matrix in rb format reps = 10000 # Number of repetitions of the operation to avoid overhead block_size_ini = 250 # Minimum block size block_size_end = 1000 # Maximum block size increment = 250 # Increment between block sizes base = 0 # Starting nnz of the matrix freq = [2400000, 2000000, 1600000, 1200000] # Operating frequency sym = 1 # If 1 the matrix is symmetric. If 0 the matrix is no-symmetric.</code></pre> <p>Example:</p> <p><code>numactl --membind 0 taskset -c 0 ./src/driver audikw_1.rb 10000 250 1000 250 0 2400000</code></p> <p>Note that <code>numactl</code> and <code>taskset</code> utilities are used to guarantee both NUMA and process-to-core affinity.</p> </li> <li> <p>Generating the dataset:</p> <ol> <li> <p>Edit the <code>SpMV-driver/run_all.sh</code> and uncomment the line <code>matrices =</code> in order to launch the driver for Train_symmetric / Train_noSymmetric / Test_symmetric / Test_noSymmetric matrices.</p> </li> <li> <p>Edit the 3rd parameter in the command SpMV-driver/run_driver.sh: 1 for symmetric matrices, 2 for unsymmetric matrices.</p> </li> <li> <p>Edit the command in SpMV-driver/run_driver.sh to select the input parameters of the driver as explained before.</p> </li> <li> <p>Run <code>SpMV-driver/run_all.sh</code> to obtain <code>hdf5</code> files that will create the dataset.</p> </li> </ol> </li> <li> <p>Merging the dataset:</p> <p>Run the script</p> <p><code>python3 SpMV-driver/merge_train_matrices.py /path/to/hdf5/matrix/files /output/path</code></p> <p>to obtain a single <code>hdf5</code> file containing all data from individual <code>hdf5</code> files obtained in the previous step. This merged file is the training dataset.</p> </li> </ol> <p><strong>Hyperparameter search</strong></p> <p>The script <code>spmv_cnn_hyperas.py</code> performs the hyperparameter search via the Hyperas tool. This script requires the hdf5 file dataset in the directory <code>dataset/train/</code> and produces both a <code>best_model_*.json</code> and <code>best_run_*.json </code>files in the <code>results/models/</code> directory containing the model structure and hyperparameters of the best performing configuration.</p> <p>This script can be invoked in the following way:</p> <p><code>python3 spmv_cnn_hyper.py 2400000 Time</code></p> <p>where <code>2400000</code> is the operating processor frequency (2.4 GHz) at which the dataset was generated and <code>Time</code> the modeled metric. According to the labels in the dataset, the hyperparameter search can also be performed with the <code>Energy</code>, <code>EPKG</code> and <code>EDRAM</code> metrics, corresponding to the energy measured by the Intel RAPL counters from our Intel Xeon Haswell core. In our case, however, we only search hyperparameters for the <code>Time</code> and <code>Energy</code> metrics at 2.4 GHz. Other metrics and frequencies inherit the best performing model and settings from the previous configuration.</p> <p><strong>Training</strong></p> <p>The script <code>spmv_cnn_train.py</code> performs the training on the best performing models obtained on the previous step. For that, it uses both the <code>best_model_*.json</code> and <code>best_run_*.json</code> files obtained in the hyperparameter search.</p> <p>This script can be invoked in the following way:</p> <p><code>python3 spmv_cnn_train.py 2400000 Time</code></p> <p>where <code>2400000</code> is the operating processor frequency (2.4 GHz) and <code>Time</code> the modeled metric. The training should be performed per metric and frequency. The training produces a file that contains the trained weights, so the model is ready for performing inference (testing).</p> <p><strong>Testing</strong></p> <p>The script <code>spmv_cnn_test.py</code> performs the test on the set of testing matrices involved in the SpMV operation.</p> <p>This script can be invoked in the following way:</p> <p><code>python3 spmv_cnn_test.py 2400000 Time</code></p> <p>where <code>2400000</code> is the operating processor frequency (2.4 GHz) and <code>Time</code> the modeled metric. The test should be performed per metric and frequency. The training produces two files in the <code>results/tests/</code> directory:</p> <ul> <li><code>Pred_*.txt</code>: This file contains the real measurements and the predictions obtained by the CNN for the individual vpos blocks of the testing matrices.</li> <li><code>Test_*.txt</code>: This file summarizes the information of <code>Pred_*.txt</code> file, showing the average relative error among the blocks of each test matrix and the total relative error, which is computed by summing up the real measurements and the predictions for all the blocks of a same matrix and computing the relative error upon those values.</li> </ul> <p><em>Note that this testing step and the two previous steps (hyperparameter search and training) can be performed at once using the <code>run.sh</code> script.</em></p> <p><strong>References</strong></p> <p>Publications describing <strong>SpMV-CNN-Model</strong>:</p> <ul> <li>Barreda, M., Dolz, M.F., Castaño, M.A. et al. Performance modeling of the sparse matrix–vector product via convolutional neural networks. J Supercomputing (2020). <a href="https://doi.org/10.1007/s11227-020-03186-1">https://doi.org/10.1007/s11227-020-03186-1</a></li> </ul> <p><strong>Acknowledgments</strong></p> <p>The <strong>SpMV-CNN-Model</strong> research has been partially supported by:</p> <ul> <li> <p>Project TIN2017-82972-R <strong>“Agorithmic Techniques for Energy-Aware and Error-Resilient High Performance Computing”</strong> funded by the Spanish Ministry of Economy and Competitiveness (2018-2020).</p> </li> <li> <p>Project CDEIGENT/2017/04 <strong>“High Performance Computing for Neural Networks” </strong>funded by the Valencian Government.</p> </li> <li> <p>Project UJI-A2019-11 <strong>“Energy-Aware High Performance Computing for Deep Neural Networks”</strong> funded by the Universitat Jaume I.</p> </li> </ul>
Mining API Interactions to Analyze SoftwareRevisions for the Evolution of Energy Consumption (MSR'2021 Dataset)
<p><strong>Motivation</strong></p> <p>This repository contains the data-set used as a basis for our MSR'2021 paper <em>Mining API Interactions to Analyze Software Revisions for the Evolution of Energy Consumption</em>.</p> <p><strong>Description of the dataset</strong></p> <p>The dataset is stored in a file <em>msr_2021_dataset.csv</em> and contains the following data:</p> <ul> <li>id - an individual identifier</li> <li>sampleNr - a number identifying the group this sample relates to</li> <li>name - the name of the library examined</li> <li>className - the class name as an abbreviation</li> <li>method - the name of the executed method</li> <li>duration - duration of method execution</li> <li>durationAdjusted - duration after alignment between method trace and energy profile</li> <li>energyConsumption - computed energy consumption</li> <li>watts - recorded wattage</li> <li>`package-names` - per package uAPI profile</li> <li>uApi - the computed uAPI profile value</li> </ul> <p>The files <em>joule_anova_posthoc_result.csv</em> and <em>uAPI_anova_posthoc_result.csv</em> contain the results of the ANOVA and Tukey HSD posthoc analysis to determine accuracy and F1-score of the presented approach.</p> <p> </p> <p><strong>License</strong></p> <p>Creative Commons CC-BY</p>
Optimised household consumption profiles through a smart building energy mangement system TABEDE
<p>In the context of the TABEDE project (<a href="https://www.tabede.eu/">https://www.tabede.eu/</a>) several synthetic profiles simulating the consumption and generation of residential buildings, whose appliances were controlled by our proposed Energy Management System (i.e., the TABEDE solution), were simulated. Their construction process was characterised by the following:</p> <ul> <li>Consumption profiles were generated via a bottom-up approach capable of emulating the consumption of individual household appliances. These last ones correspond to the most used appliances in the UK, which were randomly distributed among the buildings based on their average utilisation rate and ownership observed in residential buildings in the country.</li> <li>The physics in terms of heat exchange between neighbouring buildings and the environment were considered, together with the size of the buildings and their physical characteristics. A total of 66 houses and apartments, according to 8 type or building archetypes were created.</li> <li>PV generation profiles were generated according to the meteorological condition of the simulated day.</li> </ul> <p>Together with this, the profiles feature how the TABEDE solution optimised the flexible part of the consumption (i.e., appliances that were controllable by the solution and whose consumption could be shifted in time without sacrificing user comfort) to minimize the electricity bill of the buildings.</p> <p>The information contained in the actual database features the following variables:</p> <ul> <li>TABEDE penetration: percentage of buildings owning the TABEDE solution. Buildings with TABEDE will observe their flexible consumption being optimised.</li> <li>PV penetration: percentage of buildings with a PV system installed on them.</li> <li>Simulation day: one day in summer (19/06/2019) featuring the highest solar radiation of the year, and a day in winter (19/12/2019) with the lowest.</li> <li>Batteries: whether the PV systems is installed alongside household batteries.</li> </ul> <p>Details on the formulation can be found in: <a href="https://urldefense.com/v3/__https:/www.energy-proceedings.org/an-intelligent-infrastructure-for-enabling-demand-response-ready-buildings/__;!!La4veWw!khYBEaeJY85mX5yQUrp0PwoXcg5U10dEdgZ296hONYGyBS5xg91Z8MoDUQy34a4f9Lo$">https://www.energy-proceedings.org/an-intelligent-infrastructure-for-enabling-demand-response-ready-buildings/</a></p>
Dataset: The effects of class balance on the training energy consumption of logistic regression models
<p>Two synthetic datasets for binary classification, generated with the Random Radial Basis Function generator from WEKA. They are the same shape and size (104.952 instances, 185 attributes), but the "balanced" dataset has 52,13% of its instances belonging to class c0, while the "unbalanced" one only has 4,04% of its instances belonging to class c0. Therefore, this set of datasets is primarily meant to study how class balance influences the behaviour of a machine learning model.</p>
Demonstration Cases - Simulation data of energy consumption of residential building typologies
<p>The dataset is about the energy analysis for retrofit strategies of 5 building typologies and the EDEA project located in 3 climates zones in Europe: South (Madrid), Central (Berlin) and North (Helsinki).<br> The dataset includes:<br> (1) Open Document Spreadsheet (.ods) file with the results of Heating Consumption (kWh/m2·year) and Cooling Consumption (kWh/m2·year) for the five buildings, in three locations and for several scenarios:<br> - Locating external new insulation in walls and roof.<br> - Replacing Windows.<br> - Combination strategies: locating new insulation layers and replacing the existing windows.<br> - Installing solar protection devices.</p>
Dataset: energy consumption patterns in a science and technology park
<p>This dataset contains time series of energy consumption and external temperature for a group of buildings in a science and technology park, from 2018 to 2022, that is suitable for the development of algorithms to improve energy efficiency and for the early detection of energy consumption peaks based on night-time outdoor temperatures.</p> <p>Time series of power, energy consumption and external temperature for group of tertiary buildings, from 2018-01-01 to 2022-09-15.</p> <p>Peaks in electricity consumption are a major concern for building owners, especially during summer, when external temperatures are high, and users demand air conditioning. Owners may face high costs, observe increased risk of overheating in energy intensive equipment, and may exceed the power threshold set out in the electricity supply contract.</p> <p>Several effective "peak-shaving" strategies can be put in place, such as a higher temperature set-point (which implies a temporary reduction of comfort levels), switching off low-priority processes, and starting the cooling process earlier than usual.</p> <p>This dataset can be used to develop algorithms for early detection of peaks in energy consumption, based on external temperatures measured during the night.</p> <p> </p>
A Systematic Review on Techniques and Approaches to Estimate Mobile Software Energy Consumption (SUSCOM Dataset)
<p>Dataset and replication data for the systematic review entitled "A Systematic Review on Techniques and Approaches \\to Estimate Mobile Software Energy Consumption".</p>
21 coffee makers energy consumption dataset
<p>This dataset represents the use of 21 coffee machines over time, with each line representing an energy consumption event of a specific machine.</p>
Replication data for: "How does Docker affect energy consumption? Evaluating workloads in and out of Docker containers"
<p>Database of raw power measurements and energy summaries for our Docker energy tests.</p> <p>Please cite us if you use this dataset.</p> <p>Schema</p> <pre><code>CREATE TABLE configuration( name TEXT PRIMARY KEY, description TEXT ); CREATE TABLE experiment( name TEXT PRIMARY KEY, description TEXT ); CREATE TABLE run( id PRIMARY KEY, configuration TEXT REFERENCES configuration(name) ON DELETE CASCADE ON UPDATE CASCADE, experiment TEXT REFERENCES experiment(name) ON DELETE CASCADE ON UPDATE CASCADE ); CREATE TABLE measurement( run REFERENCES run(id) ON DELETE CASCADE ON UPDATE CASCADE, timestamp REAL NOT NULL, -- Unix timestamp in milliseoncds power REAL NOT NULL ); CREATE TABLE energy( id PRIMARY KEY REFERENCES run(id), configuration TEXT REFERENCES configuration(name) ON DELETE CASCADE ON UPDATE CASCADE, experiment TEXT REFERENCES experiment(name) ON DELETE CASCADE ON UPDATE CASCADE, energy REAL NOT NULL, started REAL NOT NULL, ended REAL NOT NULL, elapsed_time REAL NOT NULL -- in milliseconds );</code></pre>
Dataset for Analysis of the Overhead Crane Energy Consumption Using Different Container Loading Strategies in Urban Logistics Hubs
<p>The purpose of this dataset is to enable the replication of the research results presented in the article: Kłodawski Michał, Jachimowski Roland, & Chamier-Gliszczyński Norbert, 2024. „Analysis of the Overhead Crane Energy Consumption Using Different Container Loading Strategies in Urban Logistics Hubs”. Energies 17: 1–24. https://doi.org/10.3390/en17050985 - published online: 2024-02-20, which discusses the application of simulation in solving the problem of the overhead crane energy consumption using different container loading strategies in Urban Logistics Hubs.</p> <p>Dataset contains:</p> <ul> <li>Readme.txt: description of the dataset.</li> <li>Data_Crane.xlsx: Contains the input data used in the model for estimating crane energy consumption.</li> <li>Results_01.csv: Contains output data - Simulation results of energy consumption, and total average energy recovery for each scenario.</li> <li>Results_02.csv: Contains output data - Simulation results - mean values from the results of all scenario replications.</li> </ul> <p>The dataset was created as part of the E-Laas project (Energy optimal urban logistics As A Service).<br>Project implemented as part of the call ERA-NET Cofund Urban Accessibility and Connectivity (ENUAC China Call) organized by JPI Urban Europe and the National Natural Science Foundation of China (NSFC). This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 875022.<br> E-Laas project is carried out in an international consortium. Project coordinator in Europe: Chalmers University of Technology (Sweden), project coordinator in China: Shanghai University (China), consortium members: Tsinghua University (China), Warsaw University of Technology (Poland), cooperation partners: Stockholms stad, Trafikkontoret (Sweden), ParkUnload (Spain), Metropolis GZM (Poland), Shanghai Urban-Rural Construction and Transportation Department (China), Volvo Group Trucks Technology and Operations (Sweden).<br>- The Chinese part of the project is funded by National Natural Science Foundation of China.<br>- The Swedish part of the project is funded by Swedish Energy Agency.<br>- The Polish part of the project is funded by the National Science Centre, Poland (project no. 2022/04/Y/ST8/00134). The value of the co-financing is PLN 878,107.00. Project duration 27/04/2023 - 26/04/2026 (36 months).</p>
Energy consumption data from an office building, waterpark and warehouse in Slovenia and energy production data from a PV Plan (900KW)
<p>The first dataset included 9-month hourly energy data from a waterpark, a warehouse and high-rise office buildings. The second dataset includes 10-year hourly energy production data from a PV plant (900KW). Both datasets refer to Ljubljana, Slovenia. </p>
Low-energy Museum Storage Buildings: Climate, Energy Consumption and Air Quality. Data Set for Final Data Report
<p>The 43 txt-files included in this dataset relate to the report: Ryhl-Svendsen, Jensen, Bøhm, and Klenz Larsen (2012): <em>Low-energy Museum Storage Buildings: Climate, Energy Consumption and Air Quality. UMTS Research Project 2007</em>–<em>2011: Final Data Report</em>, Kgs. Lyngby: National Museum of Denmark, 122 pp.</p> <p>The document <a href="https://zenodo.org/api/files/145584b0-46b5-4341-8b02-7dfea90fa97c/00_List-of-data-files.pdf?versionId=a3e9691f-6e73-4a7b-aaab-c8ccee7c419b">00_List-of-data-files.pdf</a> contain a full list of the data files with a description of their structure and content, and is the key to how the individual data files relate to the report. </p> <p>The research project focussed on four modern museum storage facilities in Denmark, for which the indoor climate, air quality, and the energy consumption of the climate control systems was measured at several locations, typically for a period of between two and four years. The storage facilities were Museum of Southwest Jutland’s storage building in Ribe (‘Ribe’), The Shared Storage Facility at The Centre for Preservation of Cultural Heritage in Vejle (‘Vejle’), The Joint Storage Facility for museums in East Jutland/ Museum Østjylland (‘Randers’), and from The National Museum of Denmark the storage building Hall P at the Ørholm Storage Facility (‘Ørholm’). For description of the sites, monitoring campaigns, and graphed data, the report should be consulted.</p> <p>For completeness, the report is included with the dataset (<a href="https://zenodo.org/api/files/145584b0-46b5-4341-8b02-7dfea90fa97c/Report_low-energy-museum-storage-buildings.pdf?versionId=44097d39-775b-4031-9e07-6978c68912a9">Report_low-energy-museum-storage-buildings.pdf</a>).</p>
NATCONSUMERS - main factors and attitudes behind energy consumption
<p><strong>NATCONSUMERS</strong>’ key aim is to develop an economically and technologically feasible, advanced and complex user-centred framework to help decrease domestic energy consumption.</p> <p>Within this framework, the project has conducted surveys to find the most relevant drivers behind energy saving attitude.</p> <p>Four countries were chosen for the survey: the UK, Hungary, Italy and Denmark. In each of the four countries, a sample of 1,000 individuals aged 18-65 were surveyed. This sample size was deemed the most cost effective to provide a nationally representative sample. People living in shared accommodation – those in student campuses, residential care homes, sheltered housing or military barracks – were excluded from the sample as these people have limited or no control over energy use in their residence. Anyone working for the advertising or marketing industry or within the energy industry was also excluded from the sample as this could provide a conflict of interests which would bias their responses. Data collection happened in March and April 2016, conducted by Ipsos in the United Kingdom, Denmark and Italy, and by NRC in Hungary.</p> <p> </p> <p>The project D3.3 deliverable summarizes the main findings from the survey.</p> <p>http://natconsumers.eu/?wpdmdl=1582</p> <p> </p> <p>The project D4.2 deliverable presents how this data could be used in NATCONSUMERS counselling.</p> <p>http://natconsumers.eu/?wpdmdl=1586</p> <p> </p> <p>The project final deliverable D7.4 gives a summary how these data sources need to be processed and applied in a user-centred energy advice system.</p> <p>http://natconsumers.eu/?wpdmdl=1709</p> <p> </p> <p> </p>
Upcycling food ingredients from orange by-products by hot air-microwave drying. Impact on energy consumption.
<p>Currently industrial citrus by-products represent a relevant environmental issue. The main aim of this work was the chemical characterization of the different bioactive compounds obtained after hot air-microwave drying (HAD+MW) of orange by-products, and their further conversion into three <strong>upcycled </strong>ingredients with health-related benefits: aqueous extract, ethanolic extract and <strong>dietary fibre</strong>. Total phenolics, antioxidant capacity, individual phenolic acids, flavonoids, limonin and carotenoids were monitored during blanching and colour extraction steps by analysing fresh by-products and process co-products: an aqueous extract rich in polyphenols and an ethanolic extract rich in carotenoids. After drying, the resulting fibre was characterized in terms of chemical composition, soluble and insoluble dietary fibre content and particle size. Technological properties and colour were compared to those of commercial citrus fibre. Energy and time consumption were compared with conventional hot air drying (HAD). Most polyphenols (50-65 %) and limonin (70 %) were extracted during the blanching step. 86 % of carotenoids were removed by soaking in ethanol. The orange fibre obtained had 71.9 g DF/ 100 g and antioxidant properties (205 mg TE/ Kg<sub>dm</sub>). Whiteness, water retention capacity and oil retention capacity were similar to commercial citrus fibre. HAD+MW reduced drying time and energy consumption by up to 50 % compared to HAD.</p>
Categorical variables based on cross country household survey on energy consumption
<p>The data used in this file was collected via two large-scale surveys conducted in Italy, Switzerland and the Netherlands. A total of 6,138 responses were recorded, containing information on socio-demographic and socio-psychological characteristics, dwelling and household characteristics, technologies and energy services used, and their metered electricity consumption. There were a large number of missing responses for metered electricity consumption in the Netherlands, leading to an under-representation of data from this country. The survey responses were used to construct newly defined energy efficiency indicators, and energy service indicators. This allows two distinct factors to be separated: service consumption, and energy efficiency relative to the demanded service. Firstly, dwelling characteristics and survey responses related to energy services (e.g. floorspace, ownership of specific appliances and number of lightbulbs), were regressed to the collected metered electricity data. For each household, this allowed us to calculate the expected lighting and appliance electricity demand based on the level service that the household demanded, which is referred to as lighting and appliance service demand indicators. The idea is that a larger house, or a house with more appliances for example is expected to use more electricity. Relative to this expected electricity demand energy efficiency can be calculated. All variables are categorised in categorical variables deducted based on the questions asked in the two surveys. The survey responses were clustered based on the lighting service demand, appliance service demand and the efficiency gap (k-means clustering with Jaccard dissimilarity measure) which is described in Edelenbosch, Miu et al (2022). Translating observed household energy behaviour to agent-based technology choices in an integrated modelling framework. <em>Iscience</em> (accepted).</p>
Seasonal analysis comparison of three air-cooling systems in terms of thermal comfort, air quality and energy consumption for school buildings in Mediterranean climates
<p>Efficient air-cooling systems for hot climatic conditions, such as Southern Europe, are required in the context of nearly Zero Energy Buildings, nZEB. Innovative air-cooling systems such as regenerative indirect evaporative coolers, RIEC and desiccant regenerative indirect evaporative coolers, DRIEC, can be considered an interesting alternative to direct expansion air-cooling systems, DX. The main aim of the present work was to evaluate the seasonal performance of three air-cooling systems in terms of air quality, thermal comfort and energy consumption in a standard classroom. Several annual energy simulations were carried out to evaluate these indexes for four different climate zones in the Mediterranean area. The simulations were carried out with empirically validated models. The results showed that DRIEC and DX improved by 29.8% and 14.6% over RIEC regarding thermal comfort, for the warmest climatic conditions, Lampedusa and Seville. However, DX showed an energy consumption three and four times higher than DRIEC for these climatic conditions, respectively. RIEC provided the highest percentage of hours with favorable indoor air quality for all climate zones, between 46.3% and 67.5%. Therefore, the air-cooling systems DRIEC and RIEC have a significant potential to reduce energy consumption, achieving the user’s thermal comfort and improving indoor air quality.</p>
Comparability and Reproducibility in HPC Applications' Energy Consumption Characterization
<p>The computational power of HPC systems continues to grow, and improving their energy efficiency is a critical issue for the field in the face of climate change and energy crises. One major aspect of energy optimization lies in the applications run on the systems themselves. In this work, we are looking into comparing energy consumption between different systems using a characterization process based on the recent energy characterization paper as a reference and starting point for other data centers to assess their application’s energy patterns. We demonstrated that we could use the methods from the starting paper, replicate the findings, and extend the work to more applications and more systems. Our work acts as a proof of concept for a repository of HPC applications’ energy patterns in our future work.<br><br>This is the collection of jobscripts, data, and python scripts used in the paper.</p>
ScienceDex guides
Understand access before you commit
These curated guides explain access requirements, typical timelines, costs, and reuse considerations for widely used research datasets.
Allen Brain Atlas
Allen Brain Atlas is an Allen Institute collection of brain map atlases, datasets, APIs, and analysis tools covering mouse, human, and non-human primate brain resources.
Annotated Behaviour and Observability Dataset (ABODe)
ABODe is a University of Edinburgh DataShare dataset for behavior classification in group-housed mice using home-cage video, identities, bounding boxes, ground-plate positions, and annotator labels.
DANDI Archive for NWB datasets
DANDI is a BRAIN Initiative archive for publishing and sharing neurophysiology data, including electrophysiology, optophysiology, and behavioral data packaged as NWB and related standards.
International Brain Laboratory public data
The International Brain Laboratory public data releases expose standardized mouse decision-making experiments, including Neuropixels recordings, widefield calcium imaging, behavior, and session metadata accessed through the ONE API.
OpenNeuro
OpenNeuro is a free, open platform for sharing neuroimaging datasets, with public search, dataset pages, and download paths for web, S3, DataLad, and the OpenNeuro CLI.