szilard / GBM-perf

Performance of various open source GBM implementations
MIT License
215 stars 28 forks source link

c5.metal & multicore scaling #41

Open szilard opened 4 years ago

szilard commented 4 years ago
Architecture:                    x86_64
CPU op-mode(s):                  32-bit, 64-bit
Byte Order:                      Little Endian
Address sizes:                   46 bits physical, 48 bits virtual
CPU(s):                          96
On-line CPU(s) list:             0-95
Thread(s) per core:              2
Core(s) per socket:              24
Socket(s):                       2
NUMA node(s):                    2
Vendor ID:                       GenuineIntel
CPU family:                      6
Model:                           85
Model name:                      Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz
Stepping:                        7
CPU MHz:                         1200.669
CPU max MHz:                     3900.0000
CPU min MHz:                     1200.0000
BogoMIPS:                        6000.00
Virtualization:                  VT-x
L1d cache:                       1.5 MiB
L1i cache:                       1.5 MiB
L2 cache:                        48 MiB
L3 cache:                        71.5 MiB
NUMA node0 CPU(s):               0-23,48-71
NUMA node1 CPU(s):               24-47,72-95
szilard commented 4 years ago

On r4.8xlarge's (Intel Xeon CPU E5-2686 v4 @ 2.30GHz) 16 physical cores (current benchmark):

Tool Time[s] 100K Time[s] 1M Time[s] 10M AUC 1M AUC 10M
h2o 14 16 90 0.762 0.776
xgboost 0.8 5.5 70 0.748 0.754
lightgbm 2.1 4.0 20 0.765 0.792
catboost 3.8 10 80 0.734 0.735

On c5.metal's 1 socket's (Intel Xeon Platinum 8275CL CPU @ 3.00GHz) 24 physical cores (no HT):

Tool Time[s] 100K Time[s] 1M Time[s] 10M AUC 1M AUC 10M
h2o 15 17 70 0.762 0.776
xgboost 0.5 3 40 0.748 0.754
lightgbm 1.9 3.7 14 0.765 0.792
catboost 6.4 18 88 0.734 0.735
szilard commented 4 years ago

Multi-core scaling (and NUMA):

Timings (horizontal axis) vs cores (physical cores, HT=hyperthreading used, NUMA=2 sockets used, ALL=HT+NUMA): Screen Shot 2020-09-21 at 3 24 28 AM

Speedup from 1 core to N physical cores (on 1 socket) (for different data sizes): Screen Shot 2020-09-21 at 9 57 31 AM

Speedup from 1 to 24 cores:

data size h2o xgboost lightgbm catboost
0.1M 1.8x 7.3x 1.5x 2.7x
1 M 8.7x 9.3x 5.3x 5.8x
10M 28x 7.1x 15x 11.6x

Hyperthreading (HT) and multi-socket (NUMA) slowdown:

Screen Shot 2020-09-21 at 3 25 24 AM Screen Shot 2020-09-21 at 3 25 33 AM

szilard commented 4 years ago

lstopo

Machine (189GB total)
  Package L#0
    NUMANode L#0 (P#0 94GB)
    L3 L#0 (36MB)
      L2 L#0 (1024KB) + L1d L#0 (32KB) + L1i L#0 (32KB) + Core L#0
        PU L#0 (P#0)
        PU L#1 (P#48)
      L2 L#1 (1024KB) + L1d L#1 (32KB) + L1i L#1 (32KB) + Core L#1
        PU L#2 (P#1)
        PU L#3 (P#49)
      L2 L#2 (1024KB) + L1d L#2 (32KB) + L1i L#2 (32KB) + Core L#2
        PU L#4 (P#2)
        PU L#5 (P#50)
      L2 L#3 (1024KB) + L1d L#3 (32KB) + L1i L#3 (32KB) + Core L#3
        PU L#6 (P#3)
        PU L#7 (P#51)
      L2 L#4 (1024KB) + L1d L#4 (32KB) + L1i L#4 (32KB) + Core L#4
        PU L#8 (P#4)
        PU L#9 (P#52)
      L2 L#5 (1024KB) + L1d L#5 (32KB) + L1i L#5 (32KB) + Core L#5
        PU L#10 (P#5)
        PU L#11 (P#53)
      L2 L#6 (1024KB) + L1d L#6 (32KB) + L1i L#6 (32KB) + Core L#6
        PU L#12 (P#6)
        PU L#13 (P#54)
      L2 L#7 (1024KB) + L1d L#7 (32KB) + L1i L#7 (32KB) + Core L#7
        PU L#14 (P#7)
        PU L#15 (P#55)
      L2 L#8 (1024KB) + L1d L#8 (32KB) + L1i L#8 (32KB) + Core L#8
        PU L#16 (P#8)
        PU L#17 (P#56)
      L2 L#9 (1024KB) + L1d L#9 (32KB) + L1i L#9 (32KB) + Core L#9
        PU L#18 (P#9)
        PU L#19 (P#57)
      L2 L#10 (1024KB) + L1d L#10 (32KB) + L1i L#10 (32KB) + Core L#10
        PU L#20 (P#10)
        PU L#21 (P#58)
      L2 L#11 (1024KB) + L1d L#11 (32KB) + L1i L#11 (32KB) + Core L#11
        PU L#22 (P#11)
        PU L#23 (P#59)
      L2 L#12 (1024KB) + L1d L#12 (32KB) + L1i L#12 (32KB) + Core L#12
        PU L#24 (P#12)
        PU L#25 (P#60)
      L2 L#13 (1024KB) + L1d L#13 (32KB) + L1i L#13 (32KB) + Core L#13
        PU L#26 (P#13)
        PU L#27 (P#61)
      L2 L#14 (1024KB) + L1d L#14 (32KB) + L1i L#14 (32KB) + Core L#14
        PU L#28 (P#14)
        PU L#29 (P#62)
      L2 L#15 (1024KB) + L1d L#15 (32KB) + L1i L#15 (32KB) + Core L#15
        PU L#30 (P#15)
        PU L#31 (P#63)
      L2 L#16 (1024KB) + L1d L#16 (32KB) + L1i L#16 (32KB) + Core L#16
        PU L#32 (P#16)
        PU L#33 (P#64)
      L2 L#17 (1024KB) + L1d L#17 (32KB) + L1i L#17 (32KB) + Core L#17
        PU L#34 (P#17)
        PU L#35 (P#65)
      L2 L#18 (1024KB) + L1d L#18 (32KB) + L1i L#18 (32KB) + Core L#18
        PU L#36 (P#18)
        PU L#37 (P#66)
      L2 L#19 (1024KB) + L1d L#19 (32KB) + L1i L#19 (32KB) + Core L#19
        PU L#38 (P#19)
        PU L#39 (P#67)
      L2 L#20 (1024KB) + L1d L#20 (32KB) + L1i L#20 (32KB) + Core L#20
        PU L#40 (P#20)
        PU L#41 (P#68)
      L2 L#21 (1024KB) + L1d L#21 (32KB) + L1i L#21 (32KB) + Core L#21
        PU L#42 (P#21)
        PU L#43 (P#69)
      L2 L#22 (1024KB) + L1d L#22 (32KB) + L1i L#22 (32KB) + Core L#22
        PU L#44 (P#22)
        PU L#45 (P#70)
      L2 L#23 (1024KB) + L1d L#23 (32KB) + L1i L#23 (32KB) + Core L#23
        PU L#46 (P#23)
        PU L#47 (P#71)
    HostBridge
      PCI 00:11.5 (SATA)
      PCI 00:17.0 (SATA)
    HostBridge
      PCIBridge
        PCIBridge
          PCIBridge
            PCI 7d:00.0 (Ethernet)
              Net "ens6"
      PCIBridge
        PCIBridge
          PCIBridge
            PCI 90:00.0 (NVMExp)
              Block(Disk) "nvme0n1"
  Package L#1
    NUMANode L#1 (P#1 94GB)
    L3 L#1 (36MB)
      L2 L#24 (1024KB) + L1d L#24 (32KB) + L1i L#24 (32KB) + Core L#24
        PU L#48 (P#24)
        PU L#49 (P#72)
      L2 L#25 (1024KB) + L1d L#25 (32KB) + L1i L#25 (32KB) + Core L#25
        PU L#50 (P#25)
        PU L#51 (P#73)
      L2 L#26 (1024KB) + L1d L#26 (32KB) + L1i L#26 (32KB) + Core L#26
        PU L#52 (P#26)
        PU L#53 (P#74)
      L2 L#27 (1024KB) + L1d L#27 (32KB) + L1i L#27 (32KB) + Core L#27
        PU L#54 (P#27)
        PU L#55 (P#75)
      L2 L#28 (1024KB) + L1d L#28 (32KB) + L1i L#28 (32KB) + Core L#28
        PU L#56 (P#28)
        PU L#57 (P#76)
      L2 L#29 (1024KB) + L1d L#29 (32KB) + L1i L#29 (32KB) + Core L#29
        PU L#58 (P#29)
        PU L#59 (P#77)
      L2 L#30 (1024KB) + L1d L#30 (32KB) + L1i L#30 (32KB) + Core L#30
        PU L#60 (P#30)
        PU L#61 (P#78)
      L2 L#31 (1024KB) + L1d L#31 (32KB) + L1i L#31 (32KB) + Core L#31
        PU L#62 (P#31)
        PU L#63 (P#79)
      L2 L#32 (1024KB) + L1d L#32 (32KB) + L1i L#32 (32KB) + Core L#32
        PU L#64 (P#32)
        PU L#65 (P#80)
      L2 L#33 (1024KB) + L1d L#33 (32KB) + L1i L#33 (32KB) + Core L#33
        PU L#66 (P#33)
        PU L#67 (P#81)
      L2 L#34 (1024KB) + L1d L#34 (32KB) + L1i L#34 (32KB) + Core L#34
        PU L#68 (P#34)
        PU L#69 (P#82)
      L2 L#35 (1024KB) + L1d L#35 (32KB) + L1i L#35 (32KB) + Core L#35
        PU L#70 (P#35)
        PU L#71 (P#83)
      L2 L#36 (1024KB) + L1d L#36 (32KB) + L1i L#36 (32KB) + Core L#36
        PU L#72 (P#36)
        PU L#73 (P#84)
      L2 L#37 (1024KB) + L1d L#37 (32KB) + L1i L#37 (32KB) + Core L#37
        PU L#74 (P#37)
        PU L#75 (P#85)
      L2 L#38 (1024KB) + L1d L#38 (32KB) + L1i L#38 (32KB) + Core L#38
        PU L#76 (P#38)
        PU L#77 (P#86)
      L2 L#39 (1024KB) + L1d L#39 (32KB) + L1i L#39 (32KB) + Core L#39
        PU L#78 (P#39)
        PU L#79 (P#87)
      L2 L#40 (1024KB) + L1d L#40 (32KB) + L1i L#40 (32KB) + Core L#40
        PU L#80 (P#40)
        PU L#81 (P#88)
      L2 L#41 (1024KB) + L1d L#41 (32KB) + L1i L#41 (32KB) + Core L#41
        PU L#82 (P#41)
        PU L#83 (P#89)
      L2 L#42 (1024KB) + L1d L#42 (32KB) + L1i L#42 (32KB) + Core L#42
        PU L#84 (P#42)
        PU L#85 (P#90)
      L2 L#43 (1024KB) + L1d L#43 (32KB) + L1i L#43 (32KB) + Core L#43
        PU L#86 (P#43)
        PU L#87 (P#91)
      L2 L#44 (1024KB) + L1d L#44 (32KB) + L1i L#44 (32KB) + Core L#44
        PU L#88 (P#44)
        PU L#89 (P#92)
      L2 L#45 (1024KB) + L1d L#45 (32KB) + L1i L#45 (32KB) + Core L#45
        PU L#90 (P#45)
        PU L#91 (P#93)
      L2 L#46 (1024KB) + L1d L#46 (32KB) + L1i L#46 (32KB) + Core L#46
        PU L#92 (P#46)
        PU L#93 (P#94)
      L2 L#47 (1024KB) + L1d L#47 (32KB) + L1i L#47 (32KB) + Core L#47
        PU L#94 (P#47)
        PU L#95 (P#95)