src: https://gist.github.com/Brainiarc7/6d6c3f23ea057775b72c52817759b25c

Building Tensorflow from source on Ubuntu 16.04LTS for maximum performance.

라고 하는데 정말 일까나..

2080ti cuda benchmark by using NVIDIA sample

ref url: https://www.pugetsystems.com/labs/hpc/TitanXp-vs-GTX1080Ti-for-Machine-Learning-937/

nbody -benchmark -numbodies=256000

2080Ti

--------------------------------------

> Windowed mode

> Simulation data stored in video memory

> Single precision floating point simulation

> 1 Devices used for simulation

MapSMtoCores for SM 7.5 is undefined. Default to use 64 Cores/SM

GPU Device 0: "GeForce RTX 2080 Ti" with compute capability 7.5

> Compute 7.5 CUDA device: [GeForce RTX 2080 Ti]

number of bodies = 256000

256000 bodies, total time for 10 iterations: 1225.979 ms

= 534.561 billion interactions per second

= 10691.212 single-precision GFLOP/s at 20 flops per interaction

Nvidia driver: 410.73

cuda: 9.0

g/c: 2080ti

cpu: 8700k

m/b: ga-z370-hd3

ram: dominator 3466kHz(XMP applied)

My previous GPU

- Titan Xp

--------------------------------------

> Windowed mode

> Simulation data stored in video memory

> Single precision floating point simulation

> 1 Devices used for simulation

gpuDeviceInit() CUDA Device [0]: "Graphics Device

> Compute 6.1 CUDA device: [Graphics Device]

number of bodies = 256000

256000 bodies, total time for 10 iterations: 1611.591 ms

= 406.654 billion interactions per second

= 8133.082 single-precision GFLOP/s at 20 flops per interaction

(The range of GLOP/s was from 8066.468 to 8133.082)

Nvidia driver: 375.26

cuda: 8.0

g/c: titan xp

cpu: 8700k

m/b: ga-z370-hd3

ram: dominator 3466kHz(XMP applied)

* GTX 680 showed "1606.946 single-precision GFLOP/s at 20 flops per interaction"

부모가 되면 안다는 것 중 적어도 하나는

아이를 생각하면

좋았던 기억과 미안했던 기억만이

번갈아 떠오른 다는 것 같다.