MuMMI

Create Training Set

Train application_name system_name processors num_mpi num_omppn input frequency executable_name scenario frequency papi_tot_cyc papi_tot_ins papi_l1_tcm papi_l2_tcm papi_fml_ins papi_br_tkn papi_l2_ldm papi_l2_stm papi_fdv_ins papi_br_cn papi_tot_iis papi_ca_shr papi_ca_cln papi_ca_itv papi_br_ntk papi_br_msp papi_fp_ins papi_res_stl papi_l1_ldm papi_l1_stm papi_l2_tcw papi_l2_dca papi_l2_dcr papi_l2_dcw papi_l1_icm papi_br_ins papi_l1_dch papi_l1_dca papi_l1_dcm papi_l1_tca papi_l1_ica papi_fp_ops papi_l2_ica papi_tlb_dm papi_vec_ins papi_l2_dcm papi_l2_icm papi_l2_tch papi_l2_tca papi_ld_ins papi_sr_ins papi_tlb_im papi_hw_int papi_l3_tcm nruntime power_sys power_cpu power_mem num_mpi num_omppn scaling_factor
Horovod CANDLE NT3 Benchmark ANL Theta 128 128 64 learning rate=0.001, batch size=70.0, epochs=128.0 1.301 nt3-hybrid-opt.py empirical 1.301 1 0.5889426989638944 0.0181733538908182 0.0020926102561664 0.0 0.0300779928337772 0.001022500276392 0.0 0.0 0.1099929178601383 0.0 0.0 0.0 0.0 0.0797773686090857 0.0070845582834634 0.0 0.1744511169949059 0.0059253921785077 0.0 0.0 0.0 0.0 0.0 0.0120762501906897 0.1419462056574483 0.0 0.2237019770821643 0.0059253921785077 0.0 0.4283899777992187 0.0 0.0 0.0046434175872291 0.0 0.0 0.0 0.0172627475685928 0.0193553578247592 0.1576647858393212 0.0630073247932225 0.0 0.0 0.0 4.81802912012e-09 136.466877 91.239748 15.24164 128 64 164401871855.077491805900646151
Horovod CANDLE NT3 Benchmark ANL Theta 128 128 64 learning rate=0.001, batch size=60.0, epochs=128.0 1.301 nt3-hybrid-opt.py empirical 1.301 1 0.5945187558133437 0.0180564788501563 0.0020238860586597 0.0 0.0303325072079261 0.0010255044365796 0.0 0.0 0.1110415604734947 0.0 0.0 0.0 0.0 0.080522945857066 0.0070307253301712 0.0 0.1767318528186164 0.0059058502551593 0.0 0.0 0.0 0.0 0.0 0.0122538599691907 0.1431365238888439 0.0 0.2221711629820414 0.0059058502551593 0.0 0.4300160678436187 0.0 0.0 0.0046506435384235 0.0 0.0 0.0 0.0173544891246638 0.0193783751833236 0.1587190159691468 0.0636161172762152 0.0 0.0 0.0 5.00681689958e-09 139.2457 93.60688 13.159091 128 64 162852311629.050778925868315086
Horovod CANDLE NT3 Benchmark ANL Theta 128 128 64 learning rate=0.001, batch size=50.0, epochs=128.0 1.301 nt3-hybrid-opt.py empirical 1.301 1 0.5873739831335535 0.0183041287297946 0.0020192552683697 0.0 0.0300441921739162 0.0010160554113278 0.0 0.0 0.1097728031627022 0.0 0.0 0.0 0.0 0.0796029753460057 0.0070844889934541 0.0 0.174980464455112 0.0060524048869086 0.0 0.0 0.0 0.0 0.0 0.012228386527379 0.1418702747718783 0.0 0.22231626964989 0.0060524048869086 0.0 0.4246028913373757 0.0 0.0 0.0046636291058412 0.0 0.0 0.0 0.0174871527272712 0.0195064079956409 0.1594509907634867 0.0639090766172227 0.0 0.0 0.0 4.95286580628e-09 137.763303 86.559021 11.718654 128 64 164791666668.034561591542204355
Horovod CANDLE NT3 Benchmark ANL Theta 128 128 64 learning rate=0.001, batch size=40.0, epochs=128.0 1.301 nt3-hybrid-opt.py empirical 1.301 1 0.5959042974661707 0.0181343265353894 0.002037564182922 0.0 0.0305278815772389 0.0010129596021141 0.0 0.0 0.1113281261166685 0.0 0.0 0.0 0.0 0.0807352164833725 0.0071047782296918 0.0 0.1746494734921178 0.0059385189894486 0.0 0.0 0.0 0.0 0.0 0.0121546176715895 0.1441306014692769 0.0 0.2274405142113707 0.0059385189894486 0.0 0.4322454666003031 0.0 0.0 0.0045785459455475 0.0 0.0 0.0 0.0173208953177355 0.0193584595006575 0.1595613130897308 0.0639001242232404 0.0 0.0 0.0 5.02350102674e-09 136.518315 83.42674 12.379731 128 64 162998791607.966697409430297968
Horovod CANDLE NT3 Benchmark ANL Theta 128 128 64 learning rate=0.001, batch size=30.0, epochs=128.0 1.301 nt3-hybrid-opt.py empirical 1.301 1 0.5901682222773964 0.0180188042738794 0.0020701255495336 0.0 0.0301899920353206 0.001008744583543 0.0 0.0 0.110302429799285 0.0 0.0 0.0 0.0 0.0800000783040664 0.0071232091300076 0.0 0.1699851722272501 0.0058975318678979 0.0 0.0 0.0 0.0 0.0 0.0123810146285546 0.1424912669516961 0.0 0.224306872678967 0.0058975318678979 0.0 0.4246484207468067 0.0 0.0 0.0047043608676593 0.0 0.0 0.0 0.0172610093900184 0.019331134939552 0.158980210620787 0.0637712838429975 0.0 0.0 0.0 5.13871793651e-09 138.854772 89.218139 14.016598 128 64 164023665516.158412199481970123
Horovod CANDLE NT3 Benchmark ANL Theta 128 128 64 learning rate=0.001, batch size=10.0, epochs=128.0 1.301 nt3-hybrid-opt.py empirical 1.301 1 0.5863512055420886 0.0180561772655115 0.0020509441296691 0.0 0.0303091416318284 0.001022471321405 0.0 0.0 0.109564478809267 0.0 0.0 0.0 0.0 0.0794716531494765 0.0070974027978311 0.0 0.1743879725830543 0.0059080674795096 0.0 0.0 0.0 0.0 0.0 0.0121637373893777 0.1432628902256641 0.0 0.2227352767267587 0.0059080674795096 0.0 0.4279690581840662 0.0 0.0 0.0046254531146262 0.0 0.0 0.0 0.0175453064080063 0.0195962505376754 0.1579105101413002 0.0632955021342772 0.0 0.0 0.0 6.81305003542e-09 143.647241 96.02649 12.130243 128 64 165865849968.08060914430171863