Molecular dynamics package for chemical simulations. Shows a high amount of time in the backend where both CPU and memory contribute.
AMD metrics show we only spend ~50% of the time on the CPU. There is a moderate amount of I/O so likely waiting there. There is some floating point but much less than other codes such as namd.
elapsed 482.655
on_cpu 0.476 # 7.61 / 16 cores
utime 3565.437
stime 108.786
nvcsw 72583 # 86.38%
nivcsw 11441 # 13.62%
inblock 267168 # 553.54/sec
onblock 751504 # 1557.02/sec
cpu-clock 3674433961350 # 3674.434 seconds
task-clock 3674472081380 # 3674.472 seconds
page faults 5786497 # 1574.783/sec
context switches 86216 # 23.464/sec
cpu migrations 2090 # 0.569/sec
major page faults 2125 # 0.578/sec
minor page faults 5784372 # 1574.205/sec
alignment faults 0 # 0.000/sec
emulation faults 0 # 0.000/sec
branches 944150288240 # 31.083 branches per 1000 inst
branch misses 13146281378 # 1.39% branch miss
conditional 715415461727 # 23.552 conditional branches per 1000 inst
indirect 36389905918 # 1.198 indirect branches per 1000 inst
cpu-cycles 15950171457573 # 2.07 GHz
instructions 30275006717964 # 1.90 IPC
slots 31897501627062 #
retiring 10256456755117 # 32.2% (32.2%)
-- ucode 11587443035 # 0.0%
-- fastpath 10244869312082 # 32.1%
frontend 1782822670565 # 5.6% ( 5.6%)
-- latency 1041193668954 # 3.3%
-- bandwidth 741629001611 # 2.3%
backend 19397722726257 # 60.8% (60.8%)
-- cpu 8131906328178 # 25.5%
-- memory 11265816398079 # 35.3%
speculation 442612899648 # 1.4% ( 1.4%)
-- branch mispredict 424617941550 # 1.3%
-- pipeline restart 17994958098 # 0.1%
smt-contention 17876810980 # 0.1% ( 0.0%)
cpu-cycles 15914711735792 # 2.07 GHz
instructions 30248693256001 # 1.90 IPC
instructions 10090949159980 # 24.975 l2 access per 1000 inst
l2 hit from l1 152624079844 # 18.65% l2 miss
l2 miss from l1 16245693931 #
l2 hit from l2 pf 68637530034 #
l3 hit from l2 pf 3233438783 #
l3 miss from l2 pf 27522248586 #
instructions 10078247012013 # 67.501 float per 1000 inst
float 512 79 # 0.000 AVX-512 per 1000 inst
float 256 124 # 0.000 AVX-256 per 1000 inst
float 128 680293984159 # 67.501 AVX-128 per 1000 inst
float MMX 0 # 0.000 MMX per 1000 inst
float scalar 0 # 0.000 scalar per 1000 inst
Intel metrics
elapsed 832.937
on_cpu 0.731 # 11.69 / 16 cores
utime 9542.274
stime 198.469
nvcsw 99850 # 79.04%
nivcsw 26479 # 20.96%
inblock 6552
onblock 2191480
cpu-clock 9741000680111 # 9741.001 seconds
task-clock 9741049746853 # 9741.050 seconds
page faults 7930295 # 814.111/sec
context switches 130264 # 13.373/sec
cpu migrations 12544 # 1.288/sec
major page faults 764 # 0.078/sec
minor page faults 7929531 # 814.032/sec
alignment faults 0 # 0.000/sec
emulation faults 0 # 0.000/sec
branches 2682985873307 # 65.358 branches per 1000 inst
branch misses 16668425938 # 0.62% branch miss
conditional 2682985889723 # 65.358 conditional branches per 1000 inst
indirect 406967729044 # 9.914 indirect branches per 1000 inst
slots 63845246869628 #
retiring 34233072745943 # 53.6% (53.6%)
-- ucode 1863060964672 # 2.9%
-- fastpath 32370011781271 # 50.7%
frontend 8219305390172 # 12.9% (12.9%)
-- latency 6780205223367 # 10.6%
-- bandwidth 1439100166805 # 2.3%
backend 19478016926828 # 30.5% (30.5%)
-- cpu 5033487906332 # 7.9%
-- memory 14444529020496 # 22.6%
speculation 2127848416049 # 3.3% ( 3.3%)
-- branch mispredict 2019288962047 # 3.2%
-- pipeline restart 108559454002 # 0.2%
smt-contention 0 # 0.0% ( 0.0%)
cpu-cycles 22207184550713 # 1.67 GHz
instructions 32171673180975 # 1.45 IPC
l2 access 432584423400 # 14.223 l2 access per 1000 inst
l2 miss 150307005673 # 34.75% l2 missNo process information as the program currently hangs with a parallel make.