So everyone knows, that on Linux and any other open source OS, everything you think you want to change you can change. Anything from removing some code in a Graphics Driver or writing your own personal CPU scheduler.
However it seems that a lot of the Linux "Know it All's" ( that is meant as a compliment and not as an insult) do not really post every tune-able or personal code out there to increase performance. I read another Linux Forum, about how users have their own partition setup to increase performance off of their hard disks, however they did not release their partition scheme.
It seems that the only thing that is mostly released in terms of modification in the Linux World is themes, and not performance improving ideas or scripts or anything along those lines.
So if you look at it at that way, we are more like Mac and Windows users, where all that we modify is a theme, and nothing internally.
So how about we change that?
Post what, you modified to your Linux install to increase performance. This is a broad term so anything from compile flags, to how you increased boot times, to ways you got faster FPS, to custom toolchaines to custom patches to custom sysctl.conf's and the like. Anything that you did ( i.e. changed or "modified" ) too make things faster from stock. Faster latencies, higher throughput, etc.. etc..
Here is what I did to increase performance on my lappy via sysctl.conf:
Network
sysctl -w net.core.rmem_max = 12388608
sysctl -w net.core.wmem_max = 97777216
sysctl -w net.core.rmem_default = 85536
sysctl -w net.core.wmem_default = 65536
sysctl -w net.ipv4.tcp_rmem = '4096 87380 12388608'
sysctl -w net.ipv4.tcp_wmem = '4096 65536 12388608'
sysctl -w net.ipv4.tcp_mem = '8388608 8388608 8388608'
sysctl -w net.ipv4.route.flush = 1
sysctl -w net.ipv4.tcp_window_scaling = 1
sysctl -w net.ipv4.tcp_slow_start_after_idle = 0
sysctl -w net.core.optmem_max = 25165824
sysctl -w net.core.netdev_max_backlog = 3100
sysctl -w net.ipv4.icmp_echo_ignore_broadcasts = 1
sysctl -w net.ipv4.tcp_keepalive_time = 260
sysctl -w net.ipv4.tcp_keepalive_probes = 5
sysctl -w net.ipv4.tcp_keepalive_intvl = 15
sysctl -w net.ipv4.neigh.default.gc_thresh1 = 256
sysctl -w net.ipv4.neigh.default.gc_thresh2 = 1024
sysctl -w net.ipv4.neigh.default.gc_thresh3 = 2048
sysctl -w net.ipv4.neigh.default.gc_interval = 60
sysctl -w net.ipv4.neigh.default.gc_stale_time = 120
sysctl -w net.ipv4.tcp_low_latency = 1
sysctl -w net.ipv4.tcp_timestamps = 0
System security
sysctl -w kernel.shmmni= 4096
sysctl -w kernel.shmmax= 17179869184
sysctl -w kernel.shmall= 8388608
sysctl -w kernel.sem= 250 256000 32 4096
sysctl -w kernel.msgmni= 16384
sysctl -w kernel.msgmax= 65536
sysctl -w kernel.msgmnb= 65536
Increase number of open files
sysctl -w fs.file-max= 18980
General
sysctl -w kernel.sched_child_runs_first = 0
sysctl -w vm.nr_hugepages = 0
sysctl -w vm.nr_hugepages_mempolicy = 0
sysctl -w vm.hugepages_treat_as_movable = 0
sysctl -w vm.nr_overcommit_hugepages = 0
Kernel modules add to boot time, things that you know that will be constant like, file systems, wifi cards, webcam sd card reader and cpu arch should be set as a built in, this can increase boot time by as much as 3 seconds.
But as far as kernel patches. The two that work the greatest together are the BFS patch and the low latency patch. Applications open faster, text documents open faster. Video playback while having other applications open in the background is smooth without frame drops. But of course throughput suffers, but for Desktop use it is unnoticeable for the most part. Only when trying to open multiple 8meg+ images can you tell of the bad throughput.
And this is not a hate thread, if I come off like a jerk I am sorry. And please excuse how this is written, English is my first language but I speak slangish more then English.
*sidenote
The person who came up with the 20,000 letter captcha is not my friend at the moment.