Free In Situ Volume Compression Using NVENC Nick Leaf, Bob Miller, - - PowerPoint PPT Presentation

▶

Nov 23, 2023 486 likes •711 views

Free In Situ Volume Compression Using NVENC Nick Leaf, Bob Miller, and Kwan-Liu Ma UC Davis A supercomputer is a device for turning compute-bound problems into I/O-bound problems. -Ken Batcher Video Processing Unit (VPU) Dedicated ASIC

SLIDE 1

“Free” In Situ Volume Compression Using NVENC

Nick Leaf, Bob Miller, and Kwan-Liu Ma UC Davis

SLIDE 2

A supercomputer is a device for turning compute-bound problems into I/O-bound problems.

Ken Batcher

SLIDE 3

Video Processing Unit (VPU)

Dedicated ASIC
Energy efficient
Widely available
NVENC
Kepler (Titan) and later
Special-purpose API

http://on-demand.gputechconf.com/gtc/2017/presentation/s7111-abhijit-patait-nvidia-video-technologies.pdf

SLIDE 4

Y X Time X Z Y

SLIDE 5

Proving Usefulness

Quality Performance Compression ratio

SLIDE 6

Methodology

Seven volume datasets
Multiple compressors
a. Libx264, Libx265, NVENC h264, NVENC h265

Volume Pixel Format Encode Decode Statistics Render

Quantitative Qualitative

SLIDE 7

Ground Truth Examples

Supernova NCAR Plume Argon Bubble

SLIDE 8

Conversion Results: Quantitative

Conversion Only Conversion + Compression

int8 int16 int24 int32

SLIDE 9

Conversion Results: Qualitative

Raw int32 Raw int8 Compressed int8

SLIDE 10

Lossiness Comparison: Quantitative

SLIDE 11

Lossiness Comparison: Qualitative

QP 15 QP 30 QP 40

SLIDE 12

Proving Usefulness

Quality Performance Compression ratio

SLIDE 13

Compression Ratios

SLIDE 14

Proving Usefulness

Quality Performance Compression ratio

SLIDE 15

Integration with HPGMG-CUDA

High-Performance Geometric Multi-Grid solver benchmark
Multi-level hybrid CPU/GPU Finite-Volume (FV) solver
https://hpgmg.org
CUDA version: https://bitbucket.org/nsakharnykh/hpgmg-cuda
Why did we choose HPGMG-CUDA?
Worst-case encoding target = strongest case for technique
Compression integration
One new dependency: libnvidia-encode.so
Less than 100 lines in application, plus helper code

SLIDE 16

In Situ Results

SLIDE 17

Raw Data Load vs Load + Decode

SLIDE 18

Proving Usefulness

Performance Compression ratio Quality Sufficient for Vis 100:1 or better Tiny in situ impact (“Free”)

SLIDE 19

Acknowledgements

Sponsored in part by the U.S. Department of Energy via grants DE-

SC0007443 and DE-SC0012610 under program manager Lucy Nowell.

Thanks to NVIDIA for accepting my talk!

Leaf, Nick, Bob Miller, and Kwan-Liu Ma. "In situ video encoding of floating-point volume data using special-purpose hardware for a posteriori rendering and analysis." In 2017 IEEE 7th Symposium on Large Data Analysis and Visualization (LDAV), pp. 64-73. IEEE, 2017.

For more details, see

SLIDE 20

Thank you!

SLIDE 21

Datasets

Name Dimensions Min Max Argon Bubble 256x256x640 1 2.67 JHTDB QCR 10243

1.76E4

5.99E4 Marschner-Lobb 5123 1.18E-1 8.82E-1 NCAR Plume 252x252x1024 2.08E-6 6.5E1 Random 5123 1.16E-8 1 Supernova 8643 2.02E-15 1.25E-1 Visible Female 512x512x1734 4.03E3