The spirit of this thread has been me trying to show that overclocking via the BIOS (which _is_ software) does the same exact thing as overclocking with software running under the OS: they both write to the MSR, and thus overclocking via your BIOS is _not_ superior to overclocking via software running under your OS. The latter is often called the "software solution" as if the BIOS isn't software.
I've been doing research online trying to support my guess that both methods write to the MSR and are thus really the same. I don't claim to know for sure they both do, but I did find what looks like evidence of that. The big one being the Intel documents saying that the way the CPU clock speed is controlled is by writing to the MSR, which further affects the PLL systems, which determines the CPU speed.
Well, I've found an exception to this, which goes against my initial guess, and therefore I find it exciting!
You see, a friend of mine is thinking about building a custom computer. Haven't bought any parts yet. He's asked me for some help, and the two of us have been researching hardware.
It wasn't too long until I found this: ASUS's TPU and EPU (thank you Newegg for not only having such great prices, but for also being such a wealth of knowledge!). Here's a link were some people discuss it: http://www.rage3d.com/board/showthread.php?t=33975240
Some ASUS motherboards actually have a processor on the board dedicated to controlling the speed of the CPU! So here, the normal system of the CPU sending a code to a PLL chip isn't at play, but rather special hardware is controlling the PLL system (or maybe the CPU _can_ still talk to the PLL, but there's this other hardware for the purpose present).
So if there's special hardware present that can control clock speeds, and this hardware looks to the BIOS settings to see how to do the job, _in_that_situation_ I can see overclocking via the BIOS actually being superior to overclocking via software under your OS, provided that there is no way software under you OS can control the special software, and that the hardware does a better job of overclocking than possible by sending what you have written to the MSR to a PLL chip.
But in a typical scenario, where there is no special overclocking hardware present, it looks like my initial guess still stands that overclocking via the BIOS and overclocking via software under the OS both write to the MSR, are doing the same thing, and thus one is not better than the other.
This is something I'd like to have straight, so further input on this would be appreciated. If anyone is thoroughly convinced that overclocking through the BIOS is superior to overclocking via software under the OS, please say why, even if that reason is simply personal experience, with no known reason behind it.
Also want to note that at least one person (from the above link) likes manual overclocking better than trusting ASUS's TPU, which puts us back at square one. Is overclocking from the BIOS really superior?
Of course, I wrote this intending to reach out to Linux users on computers that don't have an option to overclock from the BIOS (or whatever their system firmware is), so whether one is better than the other isn't even choice to be had. It's just that when I delve into what's actually going on, I'm trying to see if there even is a difference, which naturally makes me wonder about people saying that going through the BIOS is superior.