OS Images |
os_Image_1(1)
|
Hardware Description |
hw_1
|
Number of Systems |
1
|
SW Environment |
non-virtual
|
Tuning |
BIOS Settings: - NUMA nodes per socket = NPS4
- Determinism Control = Manual
- Determinism Slider = Power
- cTDP Control = Manual
- cTDP = 240
- Package Power Limit Control = Manual
- Package Power Limit = 240
- L1 Stream HW Prefetcher = Disable
- L2 Stream HW Prefetcher = Disable
- xGMI Link Max speed = 18Gbps
|
Notes |
notes
|
|
JVM Instances |
jvm_Backend_1(64)
|
OS Image Description |
os_1
|
Tuning |
- cpupower -c all frequency-set -g performance
- tuned-adm profile throughput-performance
- echo 960000 > /proc/sys/kernel/sched_rt_runtime_us
- echo 40000000 > /proc/sys/kernel/sched_latency_ns
- echo 1000 > /proc/sys/kernel/sched_migration_cost_ns
- echo 800000000 > /proc/sys/kernel/sched_min_granularity_ns
- echo 200000000 > /proc/sys/kernel/sched_wakeup_granularity_ns
- echo 4000 > /proc/sys/kernel/sched_nr_migrate
- 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/kernel/numa_balancing
- echo 0 > /proc/sys/vm/numa_stat
- echo always > /sys/kernel/mm/transparent_hugepage/enabled
- echo always > /sys/kernel/mm/transparent_hugepage/defrag
- ulimit -n 1024000
- UserTasksMax=970000
- DefaultTasksMax=970000
|
Notes |
None
|
Parts of Benchmark |
Backend
|
JVM Instance Description |
jvm_1
|
Command Line |
-Xms31g -Xmx31g -Xmn29g -XX:AllocatePrefetchInstr=2 -XX:+UseParallelGC -XX:ParallelGCThreads=16 -XX:LargePageSizeInBytes=2m -XX:-UseAdaptiveSizePolicy -XX:+AlwaysPreTouch -XX:+UseLargePages -XX:SurvivorRatio=30 -XX:TargetSurvivorRatio=95 -XX:MaxTenuringThreshold=15 -XX:InlineSmallCode=11k -XX:MaxGCPauseMillis=300 -XX:LoopUnrollLimit=200 -XX:AdaptiveSizeMajorGCDecayTimeScale=12 -XX:AdaptiveSizeDecrementScaleFactor=2 -XX:+UseTransparentHugePages -XX:-UseFastStosb -XX:+UseUnalignedLoadStores -XX:+UseXMMForArrayCopy -XX:+UseXMMForObjInit -XX:+UseFPUForSpilling -XX:ParGCArrayScanChunk=3000 -XX:TLABAllocationWeight=75
|
Tuning |
None
|
Notes |
Used numactl to affinitize each Backend JVM to 8 Core / 16 Threads in every H12DST-B node in SuT - Group1: numactl --physcpubind=0-7,128-135 --localalloc
- Group2: numactl --physcpubind=8-15,136-143 --localalloc
- Group3: numactl --physcpubind=16-23,144-151 --localalloc
- Group4: numactl --physcpubind=24-31,152-159 --localalloc
- Group5: numactl --physcpubind=32-39,160-167 --localalloc
- Group6: numactl --physcpubind=40-47,168-175 --localalloc
- Group7: numactl --physcpubind=48-55,176-183 --localalloc
- Group8: numactl --physcpubind=56-63,184-191 --localalloc
- Group9: numactl --physcpubind=64-71,192-199 --localalloc
- Group10: numactl --physcpubind=72-79,200-207 --localalloc
- Group11: numactl --physcpubind=80-87,208-215 --localalloc
- Group12: numactl --physcpubind=88-95,216-223 --localalloc
- Group13: numactl --physcpubind=96-103,224-231 --localalloc
- Group14: numactl --physcpubind=104-111,232-239 --localalloc
- Group15: numactl --physcpubind=112-119,240-247 --localalloc
- Group16: numactl --physcpubind=120-127,248-255 --localalloc
|
|