Re:NEON is the new black: fast JPEG optimization on ARM servers
by baybal2 on Friday April 13, 2018 @05:53PM
(#16834210)
Cyclone V board with onboard network, some ram, and flash costs $7k a piece. 2 years ago, we benchmarked an off-the-shelf jpeg transcoder IP to be ~30-40 times faster than libjpeg on top of the line 8 core xeon.
[ Parent
| on HN ]
Re:NEON is the new black: fast JPEG optimization on ARM servers
by sannee on Saturday April 14, 2018 @02:50AM
(#16836234)
Oooh, that's pretty awesome then. I was pretty sceptical after reading about somewhat "meh" results for deep learning applications a few weeks ago.
Re:NEON is the new black: fast JPEG optimization on ARM servers
by baybal2 on Saturday April 14, 2018 @05:23AM
(#16836617)
> I was pretty sceptical after reading about somewhat "meh" results for deep learning applications a few weeks ago.
To begin with, nobody even bothered yet to make an optimised HDL design for this purposes.
This is unlike for all kinds of media transcoders which are the bread and butter of the SoC industry. An image transcoding asic you find in cellphone SoC these days can easily max the I/O ceiling, which means they can transcode faster than the SoC's memory interface can work.
Re:NEON is the new black: fast JPEG optimization on ARM servers
by greglindahl on Saturday April 14, 2018 @12:16AM
(#16835827)
How about comparing to the best library? A few alternatives are mentioned in this discussion.
Re:NEON is the new black: fast JPEG optimization on ARM servers
by baybal2 on Saturday April 14, 2018 @05:18AM
(#16836609)
To compare that, I need to run benchmarks on their lib. Our original measurements were in Mb/s, and not the time per testset.
Adding to that, our fpga boards are also programmed to do very high quality Lanzo scaling with gamma correction and a sharpening algo by default. Compression settings are on the quality side
Re:NEON is the new black: fast JPEG optimization on ARM servers
by thecompilr on Saturday April 14, 2018 @12:24PM
(#16838008)
jpegtran does not transcode. It performs lossless optimization of the Huffman coefficients.