On Fri, May 30, 2014 at 9:33 PM, Warren Young <war...@etr-usa.com> wrote:
> On 5/30/2014 11:23, Stephan Beal wrote: > >> >> a) pi is using an external USB 2.0 drive here because compiling anything >> on a SD card is just too pokey. >> > > Rotating media or SSD? > Plain old cheap spinning disk. i don't want to migrate to SSD until i can migrate all my machines, because i'm afraid i'll get spoiled by the speed difference and then resent all of my spinning disk (i've got 7 or 8 external USB drives laying around here). > Did you use the same external HDD on the laptop, or did it have an unfair > advantage in its internal drive? The laptop used its internal drive, which is most certainly faster than USB2 and almost certainly has a much larger cache (it's a relatively high-end laptop intended for software development). The laptop also (with 8GB RAM) doesn't ever need to swap, whereas the Pi will now and again. So, of course this is an apples/oranges (or raspberries!) comparison. > > b) These "f-xyz" calls are the libfossil CLI tools: >> > > I saw the caveat about networking on your f-commit tool, but does it work > with local repositories? If so, would you try that test, too? I speculate > that f-acat and f-commit model a huge chunk of everyday command line Fossil > use. > Yes, it does work fine on local repos, with one minor caveat: the "current leaf" is (for reasons i don't understand) not getting recalculated properly. After a checkin, the current leaf is always (in the timeline) marked as the previous version. A rebuild fixes it, which means that the metadata is right, but that something is missing in the calculation of the leaf. It's solely a cosmetic problem, but an annoying one nonetheless. FWIW, i use it regularly for doing libf checkins. > I hope you were doing all of these tests locally, and not introducing the > network as a confounding factor. The RPi NIC is only 100 Mbit/s. Even if > you don't saturate it, packet latency will still be higher than GigE. The RPI connection is over ethernet/ssh, but the console output for these simple tests is not enough to affect the test speed. The drive is connected directly to the Pi - i would never in my life use a fossil repo on an NFS mount. > - f-acat rid:1 >> pi: 0.34s >> laptop: 0.17s >> > > Interesting, only about 2x slower. Not even that much, if I'm right in > guessing that you have some confounding variables. i found that one particularly interesting. My assumption is that the startup overhead (opening the repo db) is the major factor there, but i'm not interested enough in it to explore it further ;). So long as it runs, i'm happy. -- ----- stephan beal http://wanderinghorse.net/home/stephan/ http://gplus.to/sgbeal "Freedom is sloppy. But since tyranny's the only guaranteed byproduct of those who insist on a perfect world, freedom will have to do." -- Bigby Wolf
_______________________________________________ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
