LCZero Benchmarks
 Share
The version of the browser you are using is no longer supported. Please upgrade to a supported browser.Dismiss

 
€
%
123
 
 
 
 
 
 
 
 
 
 
ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
 ThreadsEngine version/typeSpeed npsNeural Net NameRemark
2
RTX 3080 & 30706lc0 v0.2720x25613122911248lc0 -t 6 --backend=multiplexing --backend-opts="backend=cuda-fp16,a(gpu=0),b(gpu=1),c(gpu=0),d(gpu=1)" -w weights_11248.txt --nncache=2000000 --minibatch-size=1024
3
RTX 3080 & 2070 Super6lc0 v0.26.320x25612009611248lc0 -t 6 --backend=multiplexing --backend-opts="backend=cuda-fp16,a(gpu=0),b(gpu=1),c(gpu=0),d(gpu=1)" -w weights_11248.txt --nncache=2000000 --minibatch-size=1024
4
RTX 30904lc0 v0.26.320x25610100642850-t 4 -b cuda-fp16 --nncache=2000000 --minibatch-size=1024
5
3 x RTX 2080 TI2lc0 v0.23.1+git.6837b8320x2568081442850-t 2 --backend=demux "--backend-opts=(backend=cudnn-fp16,gpu=0),(backend=cudnn-fp16,gpu=1),(backend=cudnn-fp16,gpu=2)" --nncache=2000000 --minibatch-size=1024; go nodes 5000000
6
2x RTX TITAN4lc0 v0.20 dev (with PR 619)20x25680000--threads=4 --backend=roundrobin --nncache=10000000 --cpuct=3.0 --minibatch-size=256 --max-collision-events=64 --max-prefetch=64 --backend-opts=(backend=cudnn-fp16,gpu=0),(backend=cudnn-fp16,gpu=1) go infinite; NPS checked after 100 seconds (peak was over 100k, then it starts dropping)
7
4xV1004lc0 cuda92 cudnn714 ubuntu20x2567870010040./lc0 --backend=multiplexing --backend-opts="x(backend=cudnnhalf,gpu=0,max_batch=512),y(backend=cudnnhalf,gpu=1,max_batch=512),yy(backend=cudnnhalf,gpu=2,max_batch=512),yyy(backend=cudnnhalf,gpu=3,max_batch=512)" --no-smart-pruning --minibatch-size=1024 --threads=4
8
RTX 2070 Super & 20704lc0 v0.20 dev (with PR 619)20x2567840132332same as above
9
RTX 2070 Super & 207012lc0 v0.22.020x2567605242425ba
10
2
lc0-cudnn Batchsize=256 node-collisions=32
20x25666723kb1-256x20-2100000l
11
GTX 1070 @ stock2lc0 v0.20.320x2565415032930-t 2 --backend=cudnn-fp16 --minibatch-size=1024 --nncache=20000000; go nodes 5000000
12
mkl3lc0 v0.20.420x2565383811248--threads=3 --backend=roundrobin --nncache=10000000 --cpuct=3.0 --minibatch-size=256 --max-collision-events=64 --max-prefetch=64 --backend-opts=(backend=cudnn-fp16,gpu=0) go infinite; NPS checked at 100 seconds
13
2 x RTX 20604lc0 v0.22.020x2565241040685-t 4 -backend=demux -nncache=1000000 -minibatch-size=512 -max-prefetch=32 -backend-opts=(backend=cudnn-fp16,gpu=0),(backend=cudnn-fp16,gpu=1) go nodes 5000000
14
2 x RTX 20604lc0 v0.22.020x32052340T40B.2-106-t 4 -backend=demux -nncache=1000000 -minibatch-size=512 -max-prefetch=32 -backend-opts=(backend=cudnn-fp16,gpu=0),(backend=cudnn-fp16,gpu=1) go nodes 5000000
15
RTX Titan3lc0 v0.20.1-rc120x2565055832392--minibatch-size=512 -t 3 --backend=cudnn-fp16 --nncache=10000000; go infinite; note down NPS after 1 min
16
RTX 2080 Ti @ 338W (~1785 MHz)2
lc0 v0.20.2 (linux: fedora 29, 415.27, cuda 10.0, cudnn 7.4.2.24)
20x2565045632392-t 2 --backend=cudnn-fp16 --minibatch-size=1024 --nncache=20000000; go nodes 5000000
17
2lc0 v0.20.1-rc120x25646446--backend=cudnn-fp16 --nncache=10000000; go infinite; note down NPS after 1 min
18
RTX Titan3
lc0 v0.20.2 (linux: fedora 29, 415.27, cuda 10.0, cudnn 7.4.2.24)
20x2563939032930--minibatch-size=512 -t 3 --backend=cudnn-fp16 --nncache=2000000; go nodes 5000000
19
RTX 2070 (slight OC, +66Mhz core)lc0-v0.19.1.1-windows-cuda.zip20x2563172332085--minibatch-size=1024 -t 2 --backend=multiplexing --backend-opts="x(backend=""cudnn-fp16"",gpu=0)" ; go nodes 1000000
20
2lc0 v0.18.13143311250
21
TITAN V2lc0 v0.20.1-rc220x2563100410048--minibatch-size=512 -t 2 --backend=cudnn-fp16 --nncache=2000000; go nodes 1000000
22
6lc0-v0.19.020x2562932931748lc0 -t 6 -w weights_31748.txt --backend=multiplexing "--backend-opts="a(backend=cudnn-fp16,gpu=0,minibatch-size=512,nncache=2000000),b(backend=cudnn,gpu=1)"
23
2
lc0-v0.18.1-windows-cuda10.0-cudnn7.3-for-2080.zip
2613511250?--futile-search-aversion=0 --minibatch-size=1024 -t 2 --backend-opts="x(backend=""cudnn-fp16"",gpu=0)"
24
RTX 2070lc0 v0.26.320x2562548042850lc0 benchmark default settings
25
RTX2080(laptop)2lc0 v0.22.020x25624142T40B.2-106lc0.exe -b cudnn-fp16 -w T40B.4-160;go nodes 100000
26
2lc0-v0.20.2?2179732742--minibatch-size=512 -t 2 --backend=cudnn-fp16 --nncache=2000000; go nodes 1000000
27
RTX 20604
lc0-v0.18.1-windows-cuda10.0-cudnn7.3-for-2080.zip
20x2562141311250.\lc0 --weights=weights_run1_11248.pb.gz --threads=4 --minibatch-size=256 --allowed-node-collisions=256 --cpuct=2.8 --nncache=10000000 --backend=multiplexing --backend-opts="(backend=cudnn,gpu=0),(backend=cudnn,gpu=1)"
28
4lc0-v0.18.120x2562141311248
29
Radeon R9 390X2lc0-v0.26.3-windows10-gpu-dx1210x12819897703810lc0 benchmark
30
RTX 2060 (laptop)lc0 v0.26.320x2561752242850lc0 benchmark default settings
31
GeForce GTX 780 Ti4lc0-v0.26.020x256110001541000Default Settings, backend = cudnn-fp16, NNCacheSize = 20000000, MiniBatchSize = 1024, MaxCollisionsEvents = 1024, Threads = 4
32
RX 5700 XT2lc0-v0.25.1-windows10-gpu-dx1220x2561098142850./lc0 benchmark
33
GTX 1080 TiLC0 V17.2 dev20x256920810954GPU load 98-99%, GPU Temp <= 82°C, Fanspeed 95%, go movetime 130 000
34
GTX 1650 SUPER2lc0-v0.24.1-windows-gpu-nvidia-cuda20x256877342850./lc0 benchmark --threads=2 --backend=cudnn-fp16 --minibatch-size=512 --movetime=50000
35
GTX 10802?800011250
36
2x GTX 1060 (6GB)1lc0-win-20180526 (cuda 9.2)20x2567596kb1-256x20-2000000-t 12 --backend=multiplexing "--backend-opts=(backend=cudnn-fp16,gpu=0),(backend=cudnn-fp16,gpu=1)" --nncache=2000000 --minibatch-size=1024; go nodes 5000000
37
4LC0 Ver 17 RC2 (Windows)700510970-t 4 --minibatch-size=512 --backend=multiplexing --backend-opts=(backend=cudnn,gpu=0,max_batch=1024),(backend=cudnn,gpu=1,max_batch=1024)
38
GTX 1070Ti2lc0.20.1 (cuda)20x256649632965--nncache=8000000 --max-collision-events=256 --minibatch-size=256 --backend=multiplexing --cpuct=3.1
39
GTX 1070Ti2Lc0 v0.1720x256565711149--futile-search-aversion=0 (the equivalent of --no-smart-pruning) and otherwise default settings
40
2LC0 Ver 17 RC2 (Windows)20x256359810970--threads=1 --fpu-reduction=0.2 --cpuct=1.2 --slowmover=1.5 --move-overhead=10 --no-smart-pruning
41
GeForce GTX 1060 (3GB)lc0 v0.24.1 cuda 10.0.0 cudnn 7.4.220x2563365LS 14.3lc0 benchmark (v0.24.1)
42
GTX 1060 (6GB)2lc0 v0.24.1 (cuda 10.2)20x2563359
20x256SE-jj-9-53420000
43
GTX980m 10%underclock1lc0-win-20180522 (cuda 9.2)20x2562258kb1-256x20-2000000
44
GTX 9602lc0 v0.18.1 windows cuda20x256212311248
45
GTX980m 10%underclock220x2561855kb1-256x20-2000000--no-smart-pruning (value adjusted from 1093 to 1855 after rerun)
46
GTX 750 Ti @ 1350/1350 MHz1lc0 v0.18.1 windows cuda20x256165811258GPU Load 98%, GPU Temp 59°C, Fan Speed 39%, unknown command line flag --no-smart-pruning; -nncache 300000; go nodes 725000
47
AMD R9 Fury X (core + mem 20% overclock)
4
v0.21.2-rc3 (Manjaro, OpenCL 2.1 AMD-APP (2841.4))
20x256151842512result at 130.000 nodes, peaks at 550.000 nodes at 1880nps, then starts decreasing
48
GTX 9501
v0.16.0 custom build for Debian (cuda 9.1.85)
20x256150110687--no-smart-pruning --minibatch-size=1024 --threads=2
49
GTX 750 Ti @ stock2lc0 v0.18.1 windows cudaWhat is the "0"131411258GPU Load 100%, GPU Temp 50°C, Fan Speed 35%, unknown command line flag --no-smart-pruning
50
Nvidia Quadro K22002LC0 Ver 20.1 rc1 (Windows)20x256130432194
51
AMD RX480 (core + mem 20% overclock)
4
v0.21.2-rc3 (Manjaro, OpenCL 2.1 AMD-APP (2841.4))
20x256113342512result at 130.000 nodes, peaks at 540.000 nodes at 1385nps, then starts decreasing
52
Nvidia Geforce 840M4
v0.21.2-rc3 (Manjaro, cuda 10.1.0, cudnn 7.5.1)
20x25643342512
53
li16v0.21.2 OpenBlas (Windows)20x25628842699
54
GTX 4702lczero v0.10 OpenCL (Windows)20x25613010021
55
AMD Ryzen 3 1200 @stock4lczero v0.10 OpenBlas (Windows)20x2563510021tested using bumblebee + optirun, peaks at 570.000 nodes at 539nps, then starts decreasing, stays around 527-528nps.gp
56
57
Intel Core i7-4900MQ 4x 2,80GHz 500GB 16GB Quadro K2100M RW
LC0 Ver 20.1 rc1 (Windows)20x25632805This was performed at the starting position. Whole game average was 63k (see latest paper and talkchess clarifications)
58
TITAN V3lc0-win-20180708-cuda92-cudnn71420x25611248
59
MX1502LC0 Ver 17 RC2 (Windows)20x25611045--minibatch-size=256 --backend=cudnn
60
TITAN XP220x25610751This benchmark is done in Arena Chess. I suspect on Cutechess-cli Leela will achieve an even higher NPS.
61
(for cudnn client start with --no-smart-pruning flag and do "go nodes 130000" and wait until it finishes). For faster GPUs let it run for 1m or even 5m nodes.
×x
62
CPU
63
GeForce GTX 780 Ti
64
Have clean up for the last time! after that i will close it!
65
LCZero Benchmark Nodes/sec. GPU & CPU
66
Please put your own bench scores here in sorted NPS order if you can. If you don't know what engine type, gpu is opencl and cpu is openblas + Networks ID !
67
RTX 2080 Ti @ 1290 MHz (~169W)2-t 2 --backend=cudnn-fp16 --minibatch-size=1024 --nncache=20000000; go nodes 5000000
68
Run go infinite from start position and abort after depth 30 and report NPS output.
69
4x TPU20x256?--threads=4 --backend=roundrobin --nncache=10000000 --cpuct=3.0 --minibatch-size=256 --max-collision-events=64 --max-prefetch=64 --backend-opts=(backend=cudnn-fp16,gpu=0),(backend=cudnn-fp16,gpu=1) go infinite; NPS checked after 100 seconds (peak was over 100k, then it starts dropping)
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100