Metrics (Intel) - phoronix/hmmerThis test searches through the Pfam database of profile hidden markov models. The search finds the domain structure of Drosophila Sevenless protein.
sh - pid 31393 On_CPU 0.893 On_Core 7.144 IPC 1.295 Retire 0.703 (70.3%) FrontEnd 0.101 (10.1%) Spec 0.030 (3.0%) Backend 0.166 (16.6%) Elapsed 9.75 Procs 12 Maxrss 254K Minflt 73985 Majflt 0 Inblock 0 Oublock 8 Msgsnd 0 Msgrcv 0 Nsignals 0 Nvcsw 4869 (90.8%) Nivcsw 494 Utime 69.220512 Stime 0.436867 Start 520469.51 Finish 520479.26
The On_CPU is ~90% and it has a large number of voluntary context switches (90%). While the amount of I/O blocks recorded isn’t large, this benchmark has characteristics of one that waits on either I/O or other external item and hence may have limitations to scaling. The IPC of 1.30 and relatively high retirement rate put this middle of range with some backend stalls and some front end stalls.
Metrics (AMD) - phoronix/hmmersh - pid 17723 On_CPU 0.538 On_Core 8.604 IPC 1.503 FrontCyc 0.136 (13.6%) BackCyc 0.101 (10.1%) Elapsed 8.53 Procs 20 Maxrss 490K Minflt 133420 Majflt 0 Inblock 0 Oublock 8 Msgsnd 0 Msgrcv 0 Nsignals 0 Nvcsw 7886 (52.5%) Nivcsw 7132 Utime 72.693009 Stime 0.702369 Start 530601.54 Finish 530610.07
AMD platform shows a dropping of On_CPU as number of cores goes up correlating to scaling issues. The IPC when running is higher for AMD than Intel.
Process Tree - phoronix/hmmerProcess Tree
31393) sh 31394) hmmer 31395) hmmpfam 31396) hmmpfam 31397) hmmpfam 31398) hmmpfam 31399) hmmpfam 31400) hmmpfam 31401) hmmpfam 31402) hmmpfam 31403) hmmpfam 31404) hmmpfam
The process tree is simple.
About this graph
All the cores are kept consistently busy though at ~90%.
The IPC is also fairly consistent.
About this graph
Shows ~15% of time in backend stalls and ~10% in frontend stalls.
Next steps: None