I have a pci-e based storage working at close to 1000mb/s (Mac Pro). I do lots of GCC/make compilations. What if you want things even faster than with a pcie? Then ramdisk is your friend.
My results:
- With Regular SSD: 2 minutes and 37 seconds (on my co-worker’s machine w/Fedora)
- With my pci-e disk: 2 minutes and 35 seconds (my Mac w/Fedora)
- With Ramdisk: 1 minute and 15 seconds (my Mac w/Fedora)
So we have ~50% improvement in compilation time. For me, it’s worth the effort.
To set one up: Follow this guide
I.e:
[code language='bash']
sudo mkdir /mnt/ramdisk
sudo mount -t tmpfs -o size=2G tmpfs /mnt/ramdisk
[/code]To see how much of it is still available:
df -h
You might want to write an rsync script to copy folders onto the ramdisk that you want fast-access to. Ex a git repo that you compile frequently, some application binaries that you run. But obviously avoid putting data onto the disk.
Hi,
I once tried your solution for a program that compiles in about 30 seconds. Ramdisk did not speed this up (i wad suprised). Also I was scared to keep the data at the ram disk all the time in case of crash. I gues I would have to setup something that would dump everything from ramdisk each time i make changes.
LikeLiked by 1 person
Hmm, yea, for some things I found the ramdisk doesn’t do much.
Sometimes Linux caches files and a ramdisk doesn’t appear to speed things up :-|. Kinda depends I guess. I found that my CPU is quite a bottleneck as well. My next machine will have at least 6 cpu cores.
LikeLike
Id like that kind of machine. But also, i tried this solution on a windows machine, that even though running on an ssd, the project files were located at a refular hdd. Thats why I was suprised it didnt speed up at all.
LikeLike