Interesting numbers.
#curl on my Linux machine can download a large file from http://localhost at 5.0GiB/sec. Pointing to the file:// version of the exact same file "only" increases the speed to .8.8GiB/sec.
Discussion
Loading...
Post
@bagder You know what this means. You need to develop a Webserver with performance and reliability similar to curl.
@bagder I guess you are testing sendfile, I'm almost surprised it is not closer.
@bagder what web server are you using for the test? Any idea whether curl or the web server is the bottleneck?
@KimmoAhokas curl spends less than 40% CPU on this, Apache2 is at 100%...
@bagder Ha, suspected something like that 🤣. I wonder if there even is any web server that could match the curl performance in such a test.
bonfire.cafe
A space for Bonfire maintainers and contributors to communicate
Automatic federation enabled