Hi,
I would like to get an approximate answer in relation to how long it takes for a VMmark run to complete e.g. on a machine such as a 4 X quad-core (i.e. 16 core total)? Assume running an average (based on the VMmark results page) number of tiles for this machine i.e. 11-14 tiles simultaneously ... in approximate terms, would this be minutes, hours or days?
Thanks,
herevirtually.
Rest assured that during the core 180 minute "runtime" all the VM's are running and under load simultaneously - it during the middle 2 hours that all the measurements for VMmark scoring are made.
The tiledelay feature increases the total rampup time for each preceeding tile. So if we have a 4 tile test for example (time in minutes):
Test time line:
t00---t10-t20-t30-t40-t50
t230
--
tile0 starts at t00 ramps up until t50 and continues running until t230
tile1 starts at t10 ramps up until t50 and continues running until t230
tile 2 starts at t20 ramps up until t50 and continues running until t230
tile 3 starts at t40 ramps up until t50 and continues running until t230
The first 40 minute measurement phase starts for all 4 tiles at t80, the second at t120 and the 3rd at t160 which ends at t200
Assuming the default 10min tiledelay to stagger the startup of the VMs in each tile and the 180 minute runtime (30 minutes of warmup, 3 - 40 minute measurement intervals, and 30 minutes to rampdown), a complete VMmark run for 14 tiles would take a little over 5 hours.
Every VMmark run has the same core 3 hour "runtime" + the time to startup each tile which add to the total time to run the test.
Of course staging all 84 VMs and 14 clients and making sure everything is configured an working so the that 14 tile run will be successful can take some time - especially the first time through the process.
Hi,
Thanks for that. Referring to the staggering of the starting of VMs within each tile ... my understanding of a multi-tile VMmark result is that it is obtained with all VMs in all tiles running simultaneously. With a delay of up to 140 minutes between the start of VMs in the first tile and the start of VMs in the last tile (due to to 10 min staggered startup between VMs in each tile), can you explain / reconcile the staggered startup and my understanding of the requirement for the VMs to run simultaneously?
Thanks,
herevirtually.
Rest assured that during the core 180 minute "runtime" all the VM's are running and under load simultaneously - it during the middle 2 hours that all the measurements for VMmark scoring are made.
The tiledelay feature increases the total rampup time for each preceeding tile. So if we have a 4 tile test for example (time in minutes):
Test time line:
t00---t10-t20-t30-t40-t50
t230
--
tile0 starts at t00 ramps up until t50 and continues running until t230
tile1 starts at t10 ramps up until t50 and continues running until t230
tile 2 starts at t20 ramps up until t50 and continues running until t230
tile 3 starts at t40 ramps up until t50 and continues running until t230
The first 40 minute measurement phase starts for all 4 tiles at t80, the second at t120 and the 3rd at t160 which ends at t200
Thanks to psmith2006 for the reply and information ... I've marked the previous post as the answer to this thread.