Half the RAM half the disk cache. More time spent in disk wait. Dunno if clang has something like -pipe to skip creating some temp files may help. Making sure you use noatime on all your mounts. Check that you're not swapping....also check the RAM speeds. Slower RAM makes a huge difference for some workloads.
On Saturday, August 6, 2016, Ian Collins <[email protected]> wrote: > Over the weekend I have re-purposed my Solaris build machine to be a > compute node. I was expecting, based on raw numbers and past experience, > it to build the current C++ code base I'm working with in 50-60% of the > time taken by either of my current build slaves. > > Reality hasn't matched expectations big time! > > An existing system, using the same motherboard but slower CPUs does the > build in about 9 minutes. After 40 minutes the build was only a little > over half done on the "fast" box... > > Both builds where using Ubuntu-16.04 lx zones with more than enough memory. > > Older box: > > # sysinfo > { > ... > "Manufacturer": "Supermicro", > "Product": "X9DRH-7TF/7F/iTF/iF", > ... > "CPU Type": "Intel(R) Xeon(R) CPU E5-2620 0 @ 2.00GHz", > "CPU Virtualization": "vmx", > "CPU Physical Cores": 2, > ... > "CPU Total Cores": 24, > "MiB of Memory": "131044", > > The newer box: > > "CPU Type": "Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz", > "CPU Virtualization": "vmx", > "CPU Physical Cores": 2, > ... > "CPU Total Cores": 32, > "MiB of Memory": "65501", > > I noticed this comparing prstat on the two hosts: > > Older: > > PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/NLWP > 33887 1005 294M 205M cpu1 1 0 0:00:04 1.6% clang/1 > 33919 1005 251M 156M cpu5 1 0 0:00:02 1.0% clang/1 > 33934 1005 238M 146M cpu0 1 0 0:00:02 0.8% clang/1 > 33949 1005 215M 123M cpu9 1 0 0:00:01 0.6% clang/1 > 33952 1005 213M 118M cpu4 1 0 0:00:01 0.6% clang/1 > 33955 1005 201M 107M cpu3 1 0 0:00:01 0.5% clang/1 > 33958 1005 200M 107M cpu10 1 0 0:00:01 0.4% clang/1 > 33963 1005 180M 82M cpu8 1 0 0:00:00 0.2% clang/1 > 33964 1005 179M 81M run 1 0 0:00:00 0.2% clang/1 > 33967 1005 179M 81M cpu6 1 0 0:00:00 0.2% clang/1 > 33186 1005 51M 38M sleep 1 0 0:00:01 0.2% ninja/1 > 33970 1005 174M 75M cpu2 1 0 0:00:00 0.1% clang/1 > > Newer: > > PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/NLWP > 59248 1005 220M 145M wait 4 0 0:00:03 0.3% clang/1 > 59221 1005 240M 165M wait 4 0 0:00:04 0.2% clang/1 > 59236 1005 208M 135M wait 4 0 0:00:02 0.2% clang/1 > 59251 1005 172M 96M wait 4 0 0:00:02 0.2% clang/1 > 59254 1005 176M 101M wait 4 0 0:00:02 0.2% clang/1 > 58811 1005 651M 576M wait 4 0 0:00:20 0.2% clang/1 > 59043 1005 353M 278M wait 4 0 0:00:08 0.2% clang/1 > 59259 1005 124M 48M wait 4 0 0:00:00 0.1% clang/1 > 59264 1005 121M 45M wait 8 0 0:00:00 0.1% clang/1 > 59265 1005 119M 42M wait 8 0 0:00:00 0.1% clang/1 > 59200 1005 194M 119M wait 4 0 0:00:02 0.1% clang/1 > 59180 1005 163M 87M wait 4 0 0:00:02 0.0% clang/1 > 59099 1005 252M 177M wait 4 0 0:00:04 0.0% clang/1 > > All the "wait" states look odd to me. > > Ideas? > -- > Ian. > -- "Genius might be described as a supreme capacity for getting its possessors into trouble of all kinds." -- Samuel Butler ------------------------------------------- smartos-discuss Archives: https://www.listbox.com/member/archive/184463/=now RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00 Modify Your Subscription: https://www.listbox.com/member/?member_id=25769125&id_secret=25769125-7688e9fb Powered by Listbox: http://www.listbox.com
