OS Images |
os_Image_1(1)
|
Hardware Description |
hw_1
|
Number of Systems |
1
|
SW Environment |
Non-virtual
|
Tuning |
- Workload Profile=High Performance Compute(HPC)
- Thermal Configuration=Maximum Cooling
- Determinism Control=Manual
- Performance Determinism=Power Deterministic
- Memory Patrol Scrubbing=Disabled
- Numa Memory Domains Per Socket(NPS)=One Memory Domain Per Socket
- Last-Level Cache(LLC) As NUMA Node=Disabled
- L1 Stream HW Prefetcher=Disabled
- L2 Stream HW Prefetcher=Disabled
- Minimum Processor Idle Power Core C-State=C6 state
|
Notes |
None
|
|
JVM Instances |
jvm_Backend_1(4)
|
OS Image Description |
os_1
|
Tuning |
- ulimit -n 1024000
- tuned-adm profile throughput-performance
- echo 950000 > /proc/sys/kernel/sched_rt_runtime_us
- echo 100000000 > /proc/sys/kernel/sched_latency_ns
- echo 80000 > /proc/sys/kernel/sched_migration_cost_ns
- echo 600000 > /proc/sys/kernel/sched_min_granularity_ns
- echo 100000 > /proc/sys/kernel/sched_wakeup_granularity_ns
- echo 10000 > /proc/sys/vm/dirty_expire_centisecs
- echo 1500 > /proc/sys/vm/dirty_writeback_centisecs
- echo 40 > /proc/sys/vm/dirty_ratio
- echo 10 > /proc/sys/vm/dirty_background_ratio
- echo 10 > /proc/sys/vm/swappiness
- echo 0 > /proc/sys/vm/numa_stat
- echo 0 > /proc/sys/kernel/numa_balancing
- echo always > /sys/kernel/mm/transparent_hugepage/enabled
- echo always > /sys/kernel/mm/transparent_hugepage/defrag
|
Notes |
None
|
Parts of Benchmark |
Backend
|
JVM Instance Description |
jvm_1
|
Command Line |
-Xms122g -Xmx122g -Xmn119g -server -XX:MetaspaceSize=256m -XX:AllocatePrefetchInstr=2 -XX:LargePageSizeInBytes=2m -XX:-UsePerfData -XX:-UseAdaptiveSizePolicy -XX:+AlwaysPreTouch -XX:+UseLargePages -XX:+UseParallelGC -XX:SurvivorRatio=90 -XX:TargetSurvivorRatio=99 -XX:ParallelGCThreads=32 -XX:MaxTenuringThreshold=15 -XX:InitialCodeCacheSize=25m -XX:InlineSmallCode=10k -XX:MaxGCPauseMillis=200 -XX:+UseCompressedOops -XX:ObjectAlignmentInBytes=32 -XX:+UseTransparentHugePages -XX:CompileThresholdScaling=15
|
Tuning |
Used numactl to affinitize four Backend JVMs to physical cores in a NUMA node - numactl --physcpubind=0-15,64-79 --localalloc
- numactl --physcpubind=16-31,80-95 --localalloc
- numactl --physcpubind=32-47,96-111 --localalloc
- numactl --physcpubind=48-63,112-127 --localalloc
|
Notes |
None
|
|