Hacker News Reader

Comment by nlh | Hacker News Reader

The article says Linode has "predictably greater CPU performance". Am I missing something? It looks like Digital Ocean is the winner in every single benchmark other than Apache Static Page Serving.

(I'm assuming lower is better for encoding time.)

View on HN

The X264 encoding graph shows "frames per second" and if it's higher then it's better. And probably he measured the other encondings the same way but mislabeled the graphs.

Last I had to work on a video transcoder using ffmpeg, most codecs were not easily parallelized. That's what might have happened here, even though Linode has 4 cores, core vs core comparison it gets beaten.

Your reply makes sense. I'd note also that running 4 parallel encoding jobs would expose the CPU difference.

spyder is correct, the x264 graph was mislabeled. I've updated the label. It should have read "Frames per Second" not "Seconds". However, the mp3 and flac graphs are correctly labeled. DigitalOcean still outperforms Linode on mp3 and flac, but that's probably (guessing) because the processer are not waiting for the disk and the x264 encoding is using multiple cores where mp3 and flac encoding is only using one core.

For some of the media encoding, presumably higher is better. x264 for sure: when talking about frames encoded per second, more would be better. the rest of the benchmarks are all "higher is better" so perhaps the same is true for seconds of (mp3|flac) encoded in a period. they are certainly ambiguous though.

My thought exactly. Unless I'm misunderstanding, the only benchmark in which Linode came out in front was the Apache static page one and that was the closest of all the tests. Based on this DigitalOcean is a run-away winner which is a bit surprising.

Reply on news.ycombinator.com