Turbo mode is built in to a lot of CPUs now. The idea is that under load the processor sets a faster frequency until thermal throttling sets in. Apple's new Arm cpu, the M1 has a steady frequency of 2.5(?)Ghz, but a turbo speed of 4.5Ghz, which is a bigger difference than most. Most x86_64 stuff have lower turbo speeds.

On laptops, things can be very controlled. Fan speed is a function of temperature. Then, depending on your ACPI setup, cpu frequency can be throttled back to combat rising temperature. Laptop fans are puny anyhow. My laptop has an air intake in between the hinges where keyboard meets screen. Any partial airblock there, or anywhere the exhaust causes the thing to overheat, so I don't reckon I have frequency throttling. But you might have.

My Lenovo is a tower on the outside but I gather from my reading that it's a laptop inside. Mains power comes in via an external transformer block so there's no internal power unit and no fan (apart from the cpu fan).

Whatever is causing this, it's intermittent, so it must be hardware-related. I don't think it's bad memory because I did an overnight memtest run when the problem first blew up last year and it showed nothing. Overheating sounds more likely, but I wonder why it causes this software bug.

PS: is there anything in particular that I should be looking for in messages?

Your thing isn't exactly young, is it? Don't exclude dodgy connectors which have exceeded their life expectancy. Dirt builds up on the pins, plastic warps or wears away, etc, etc. Many of these plugs are tinned. Solder = lead + tin, and lead can wear away, leaving inferior contact surfaces. PCI or PCIE slots are obvious points, as the brackets sometimes affect or limit how things sit in the slots. You might solve any heating issue by the simple expedient of leaving the side off, or laying it on it's side.

Aside from all that, there is the issue of dry joints. Thermionic valves used to wear out, because the electrons went out of the cathode and were not adequately replaced over time. You could 'boost' a tired valve or CRT by overdriving it severely. The outside layer of the cathode would be shed as sparks, and fresh metal would be exposed.

The same thing goes on in dry joints. The metal in high current solder joints can change solder composition, joints get noisy, or in an extreme case, a black ring would insulate one side on the joint from the other. Classic points for this are the high current +5V & +3.3V pins soldered onto the motherboard. All joints losing reflectiveness are suspect. I did hardware repair for a living. I would routinely desolder such joints, clean dirty pins, and resolder them. In at least one case I found an insulating dry joint inside hardened adhesive heat shrink sleeving that was halting an entire production line. Only if the joint was soldered at a perfect temperature does it last.

It causes the bug because something that should be connected isn't, or something that shouldn't be connected is. An alternative is that random noise is being injected somewhere. Sorry to be so vague. I'm a hardware guy, but we can't do that work remotely, and I can't do it at all now.

They are all remote possibilities, except the dry joints. What I'm trying to say is that reliability wanes as age increases. I can't diagnose that issue from here. Nobody can. It's a poor idea to move a museum piece of a pc about anyhow. Reliability takes a nosedive.

Ok, I'll just have to put up with it. At least it only affects alsa and I don't use sound very much. I haven't had any other crashes; everything works like clockwork, except that it won't do a warm reboot any more. Actually that has been my general experience of pre-loved hardware, though yours seems to have been different. I've never used anything else.

I was just vaguely wondering if there was an explanation; I wasn't really looking for a cure. Maybe in Slackware-15 I'll switch to pulseaudio, which is the standard driver now. I've tried to avoid it because it has so many dependencies but it might prove more reliable.

hazel,

motherboard audio, in particular microphone input(s) are known to be low quality and even unreliable. I suggest you get some USB audio device for recording, there are some pretty good ones which work in Linux and won't break the bank. Also, you may want a condenser microphone, these are more sensitive and more suitable to record an instrument without being placed too close.

Edit: See here, this is a solid piece of equipment. I know, you are not in this side of pond, just to show you what I mean.

Would you be surprised if I told you that the actual computer cost me only a little more than that?

Pulse Audio is actually a cultural jerk from Alsa. Alsa is C/C++, conf files etc. Pulse really works best in X, it´s dependencies have dependencies which have their dependencies.... It presents as a point & click thing, but very messy to do anything involved.

Personally, I never recovered much respect for alsa after 2015. I had this religious thing to do - a meeting in the chosen language (one of 6) followed later by a two hour stream coming in English. I could organize it if I had HDMI with sound as normal, and hdmi with no sound and sound on the earphone socket. Some kind soul helped me with an .asoundrc, and I could drop it in and out via a script. I wouldn´t like to try that in pulse. In fact, I wouldn´t like to try it again - ever!

BTW: I´m apostrophe here in Mint while I back up Slackware. I´ve just noticed that the apostrophe (´) is a DEAD key. It actually looks more like an acute accent. So I get ¨é¨ by hitting apostrophe followed by e. It´s the same on shift_2 (=¨)

You should put that in Jeremy's thread about new discoveries! I just tried it myself. It doesn't work the same way on my machine, but if I hold down AltGr while typing the ', I get ê. Which is a circumflex, not an acute accent, so it makes no sense to me!

PS: if I do it with #/~, I get è.

So what. If your computer does what you want then the price was right, same is true about microphone. If you want good sounding recordings then a good microphone is paramount. Do you know how much studio microphones cost? They pay the price because there is no way around it.

Well, I'm not paying that much just to make the odd sound recording for my church! I'm a Christian but I'm not that good a Christian. They'll have to take what they get from me.

PulseAudio is a parasite layer between ALSA and user application. For most people it is nothing more than ALSA configuration tool, PulseAudio creators have clearly deep knowledge of how ALSA works and how to manipulate it. So you can't say ALSA sucks but PA is good, without ALSA PulseAudio is nothing. OK, PulseAudio can work also with OSS4, but still, it can not make sound in it's own.
Regarding HDMI, there is only one driver for HDMI sound, it is ALSA driver in kernel. When this driver is loaded all HDMI outputs can have sound - if the player is sending sound to it. People have this idea fix, to have sound in HDMI it must be configured default first with .asoundrc. Not true, all HDMI outputs are ready to use without it. For instance,
this command plays sound thru my TV set which is connected as third monitor. Meanwhile speakers connected to the PC continue working with other sound applications as normal.

So that means that if there is a problem with alsa (as there clearly is in my case), using pulseaudio will not solve it because it's still alsa under the hood. Fine! Then I won't use it. I'd rather not, frankly, given all the cruft that comes with it.

Correct. As I said, the generic sound chips used in motherboards do really crappy job with microphones. This is not all their fault, the signal level coming from microphone is very low and being analogue it is very susceptible to all kind of noise and distortion. This is why even a cheap USB microphone has much better sound. I'm sure you can get one for €10-15.