Some personal stuff - and prototypes building!


I'm not sure downclocking an idle CPU makes a lot of difference, since (I assume) it will be clock-gated (⇒ basically no dynamic power) when idle (maybe even power-gated). I'm pretty sure SoC CPUs don't run an infinite loop when idle, they just don't run at all. Correct me if I'm wrong.
Well, it's how it worked on the Pandora with its OMAP 3 chip. Maybe assuming the OMAP 5 works the same is a failure on my part, but until shown otherwise I'm assuming it's basically the same with a few headline differences.
 
I'm pretty sure SoC CPUs don't run an infinite loop when idle, they just don't run at all.
The difference is not as significant as you might expect. The wasted energy and therefore generated heat depends on which parts of the pipeline are active, the more complex the pipeline components you currently use are the more transistors will waste energy and generate heat. If you're only executing NOPs, you hardly use anything - you are already really close to not using any energy at all, especially if the microcode of the CPU is able to detect what you're trying to do.

A CPU never stops when idle, it only stops if you actively shut it down, e.g. for a standby mode.
 
I mean, the prototypes I had the chance to play with over the years were alright with the thermals (though, only running at 1 GHz iirc), but reading this reminds me how I really wouldn't mind having a fan on this thing.
It doesn't need to run all times, maybe not even much at all. But it sounds way better than throttling on time constrained tasks (games).

You don't need to tell me how this won't affect me at all. There will be many games and apps that'll unthrottled for sure. And yet we'll likely also have enough stuff hitting the performance ceiling too.
 
While we're straying into the realm of silly humorous cooling solutions...

There are tiny piezo fans that could even work at ultrasonic speeds at really tiny scales.

And fingers might get numb very quickly due to the imperceptible vibration. But it's something that'd fit in that tiny space :D
Sell more replacement parts due to people dropping theirs after their hands go numb :D


Remember the USB power capacitor that made noise? That was acting as a piezo element.

Hey, @EvilDragon let's bring back that capacitor and attach a tiny flap to it! We got our cooling solution. :D
 
Last edited:
We just need to start playing some games, processor will heat --> air will get hot --> air will start moving. If you know this no cooling solution is needed. VOILA :D
 
While we're straying into the realm of silly humorous cooling solutions...

There are tiny piezo fans that could even work at ultrasonic speeds at really tiny scales.

And fingers might get numb very quickly due to the imperceptible vibration. But it's something that'd fit in that tiny space :D
Sell more replacement parts due to people dropping theirs after their hands go numb :D


Remember the USB power capacitor that made noise? That was acting as a piezo element.

Hey, @EvilDragon let's bring back that capacitor and attach a tiny flap to it! We got our cooling solution. :D

This looks rather noisy! For sure a robust solution for really dirty/wet/greasy environments.
 
You'll always hear the airflow itself...
Yeah, from turbulences, but the sound it produces directly - as an oscilating membrane is ultrasonic.
I'm more astonished, that it can get to that amplitute or that the air doesn't restrict the amplitute more at that speed. Does it maybe rather bend in waves, but a single curve?
 
The 500MHz was removed from the kernel.org's vanilla kernel OMAP5 device tree by Ti.
One board member commented out that it may be linked to stability, or something like being in the right tolerances.

---Edit, source:
https://pyra-handheld.com/boards/threads/omap5432-at-home.75874/page-6#post-1341191

Yeah, although I've ran it for a long time on 500Mhz without issues... In theory if we can come up with a proper Frequency to voltage table, we could have a setup like the Pandora and have a full range of frequencies we can set.
 
Last edited:
I didn't find that to be quiet at all, just as loud if not louder than a normal super el cheapo fan. Not great IMO
 
And not really compact.

Those are pretty large but they can be made very tiny. Of course the amount of air moved is reduced too.
They work similarly to insect wings and insects can get pretty small.

The large ones seem to be running around 50-60hz oscillation which is great for running right off the mains.
But like instrument strings: the smaller, the higher the frequency you have to / can run them at.

I did say it was a silly idea :D

It's also possible to shape air conducts in a way that a simple oscillation will cause the air to move only one way due to compression (sound) waves blocking and bouncing the air off in one direction only with a specific resonant frequency.
The most dramatic implementation of this are "valveless pulsejet engines".
Mechanically simple but excessively noisy.
 
Anyway that was informative.

Maybe one day we'll have some MHD cooling solutions :^) .
 
Back
Top