Folding Forum

I've read lots of positive praise on the efficiency of Intel's Clear Linux.
In many cases, Clear Linux got a 20% higher performance score than most other distros like Mint or Ubuntu.
The ISO is a mere 250MB in size.

Anyone got experience with Clear Linux?

First I want to see if the live version can run on my netbook.
If it can, I might want to know how to install the deb or RPM package on there.

I assume, that anyone who's running a program like Folding@home on dedicated hardware, would want to run it on an as efficient operating system as possible; and Clear Linux just seems to fit the bill!
Though more server based, some people said it installed fine on their desktops or laptops.
I'll update the thread with my findings soon.

My understanding is that clear linux has had is binaries compiled with optimizations for intel's cpu's, this is why it has the performance boost. Most linux distro's packages are built with compiler flags that will perform on the majority of hardware.

Weather these tweaks will filter down to fah is another question altogether.

With arch linux (and derivatives), there is a file (/etc/makepkg.conf) that you can edit and set your optimizations, then, when the "makepkg" helper script runs, it will read the file and apply your optimizations for the machine it is compiling the application on.

Note though, that compiler optimizations dont always equal better performance, in some cases, it can result in compiler failures if you dont set it up properly.

I guess it does not help FAH because the FAH cores for CPU or GPU are compiled at Stanford and not on local machine. But maybe some driver or system kernel speedup in Intel's Clear Linux could help FAH too?

From what I read, though made by intel, for intel chips, it works even better on AMD cores.
The irony...

I'm not sure if the performance spike is only on flatpack applications, or also on the deb and RPM packages.
The fact that the OS is only 250MB in size, could also be the reason.
It's like running Windows 3.11 on a modern system.
I think, looking at the size, that the size of the OS has a lot to say about background tasks (or absence thereof), being one of the reasons for higher performance.

Years back we had an issue with compiled cores that gave better performance on Intel over AMD, due to compiler optimisations. Basically the core looked for "Genuine Intel" or some other string and then used optimisations like SSE, but if it found nothing it ran without, which is somewhat slower. At the time AMD lagged behind Intel in terms of power and features, but the optimisations being called by FAH were in both cpus, I think it was SSE1. At some point someone worked out how to patch the core so all computers were genuine, which broke the licensing on the compiler. So nowadays I think they just use the best generic options possible to avoid any problems like that occurring again. Also most technologies are now generic to both manufacturers, with the main differences being in certain other features like crypto and cache sizes.