[Robman]

Took that laptop all apart again, some cooling fins at the edges of the heatsink were clogged, nothing too horrible though.
I used better Arctic Silver paste, drilled some more vent holes and got the temp lowered by 15c.

I tried to set the CPU power percentage lower in the power management settings but it's kinda screwy, I'll explain.
If the CPU percentage is set to 100% - The cpu throttles from 2.1Ghz to 1Ghz then to 500Mhz.
Now, if I set the percentage to 99% - The CPU goes no higher than 1Ghz, throttling down to 500Mhz.
I just wanted the fckn thing to not clock so high.....AMD pisses me off.

With the CPU set to run at 2.1Ghz - 1Ghz - 500Mhz I now get 60c at idle, 83c full load for half hour.
With the CPU set at the 99% mark, it runs no higher than 1ghz and gets no hotter than 65c.

So, I was thinking with my little cooling improvements that I would set the CPU to go no higher than say 1.8Ghz
and that would help keep it cooler while maintaining some reasonable performance.

But do you think I can get it to go no higher than 1.8Ghz? No.... of course not, because that would be what I actually want.

[scaryred24]

The strangest thing my PC had done to date was that it somehow decides to throw it hands up on my secondary HDD and not read anything from it. It also decided to be highly unstable since it needed that HDD for it to function properly too. A simple power cycle took care of that.

[SpooferJahk]

Resetting randomly not too long after I assembled it, almost like it lost power or something to that effect. Found that one of the metal contacts of the power supply was slightly bent and bending it back into place fixed the issue.

[Altered Reality]

I remember two particularly bad things happening with computers I used.

One was way back when I had an Amiga 500. It had a faulty drive which didn't read floppies correctly, it buzzed for a couple of seconds and then the computer went back to show the Kickstart picture. Being 9 at the time, and having just read what formatting was, I cried my eyes out because I had concluded that the computer would automatically format whatever floppy I put in.
Of course, when my dad brought the computer back to the store (it was still under warranty) and the technician replaced the floppy drive, I heard an explanation of what was actually going on. Back home, with a brand new floppy drive installed for free, I was one happy Amiga user.

The other was in 2004, when I used a Pentium 3 at 800 MHz and I had just installed a second-hand GeForce 5900 to play Doom 3 with. Practically every Direct3D or OpenGL application I ran caused the computer to lock: the image would freeze on the screen, it wouldn't respond to control and I'd have to hard reset it. After trying various versions of the drivers to no avail, and reinstalling the OS to no avail, I happened to read that a GeForce 5900 would automatically underclock itself if the auxiliary molex connector was unplugged. I turned off the machine, unplugged that connector, turned it back on and launched some games. They were absolutely stable and the computer never locked up again.

[Mennuz]

I had a old 5970 card, which is 2 5870's on one card. It worked fine, but after a while, after a driver update, it refused to work. I managed to figure out something, putting a setting in the bios on a state the graphics card didn't understand, so it went in a fail-safe mode so it could display a error message. But the card was under-performing like crazy. After a year or so i found out the bridge chip between the to cards was fried and i was running on one card for a long time. It was a miracle that thing didn't fry my whole rig.

[Kathy]

My videocard overheated after 30+ hours of playing and I couldn't game even several minutes in 3d without having artifacts all over the place. Next day, since it was winter, I opened actual windows so that it got cold and I could finish the game.

[oasiz]

Bought a replacement asrock motherboard back in the P4 days since the old one had tons of issues. New one also had issues where it didn't detect some pci bus devices correctly (integrated NIC being one).

Got mad, put the HDDs back to my other system for a few days.. decided to try again with a separate PCI card. Put the HDD back in again, powered on and the board started making bright flashes and caught on fire, tons of smoke poured out and later on there was a 10cm scorch around the edge. It did give one explanation as to why the NIC wasn't working in the first place..
No other component was faulty, I managed to re-use them later when I got a replacement board for the burnt one.

[Person of Color]

Mennuz, on 08 July 2016 - 01:56 AM, said:

I had a old 5970 card, which is 2 5870's on one card. It worked fine, but after a while, after a driver update, it refused to work. I managed to figure out something, putting a setting in the bios on a state the graphics card didn't understand, so it went in a fail-safe mode so it could display a error message. But the card was under-performing like crazy. After a year or so i found out the bridge chip between the to cards was fried and i was running on one card for a long time. It was a miracle that thing didn't fry my whole rig.

Those dual GPU cards are bullshit. Fifty pounds of crap in a ten pound box. None of them last or work reliably in the long term, be it AMD or Nvidia. It wouldn't surprise me if after being asked to reinitialize one last time it just up and shat it's pants.

[Mark]

I thought my trusty old computer might be finally giving in to old age this morning. Shortly after the first bios loading screen the internal speaker started rapidly beeping non stop and would boot no further. I shut it down, removed the cover and pulled out and re-seated all cards, mem sticks and connectors. Same problem. But this time I noticed that when I hit the F2 key to go into bios setup the beeping stopped. Without making any changes I exited bios setup and the computer booted just fine and I used it all day. Tonight I decided to tackle the problem again. I measured backup battery voltage and it was on the edge at 2.99 volts. I replaced it and rebooted to reset my bios settings. After I did that I rebooted and still the same problem.

Another issue that cropped up at the same time was the Windows Help and Support screen was there at windows bootup time. I could not close it out except with Task Manager. I decided to take a break from the hardware issue to look up this Windows issue. In the first article it mentions that the Help window pops up with the F1 key which I never knew. All of a sudden I got excited. I looked at my F1 key and it was stuck in the pressed position. That was the problem the whole time. Then it dawned on me that my cat walked across my keyboard shortly before I booted up for the first time today.

GAAAHHH!!!!

PS Also, I checked the error log in the bios and it made no mention of a keyboard problem.

[Person of Color]

Check it for swollen caps - and not just the motherboard. Take all the old shit out and look at the caps. If they're swollen, it's time. You must junk the old bastard.

[Mark]

It was just a stuck key. But yes, its getting old and I'll be buying a new system some time this year.

[TDRR]

My PC, had a Debian 9 and Windows 7 dual-boot, once i had to reset it because Firefox simply wanted to eat up all my RAM and hog up my CPU, like as usual, however when i booted it up again, i was presented with a GRUB2 Rescue screen, so i just put my Windows 7 CD on the disc tray and formatted C: (Which made me lose the couple episodes i was going to release for Duke and my 320gb of por- games i mean) a couple days later i discovered that i could have rescued it with some commands, then i facepalmed really hard.

A lot of these are incredibly weird, what gave me a chuckle was how the disc simply blew up in the tray.

[High Treason]

Here's an odd one that comes down to a design flaw - for whatever reason German hardware companies in the 1990s seemed to love making broken installers, as I've yet to find one that works properly, ambitious as hell hardware every time but the software to support it is always questionable in some way.

The Terratec Maestro 32/96 uses an EEPROM to store the card's settings, things like the I/O ports, IRQs, DMAs and stuff like that. Early Plug&Pray crap. To program the EEPROM reliably requires that CPU Caches and ROM Shadowing are disabled, but this is not discussed in the documentation for the card - and it actually does have good documentation - so you wouldn't know other than the configuration program gives it away. Unfortunately this tool carries out its tasks in the wrong order, attempting to program the EEPROM and then disabling access to caches (which is dubious in itself as doing this in software isn't entirely reliable) as is even shown by the bullet list it runs through as it goes about its job. As a result the application will often lock up as the progress bar for programming the EEPROM ends, otherwise it will appear to work but will write corrupted data to the EEPROM or else it will appear to work but do nothing at all. The card is complicated, using 2 DMA channels, at least 5 I/O addresses and 3 IRQs, it generally won't work well if at all should the EEPROM not be properly programmed and the Windows configuration tool won't run if the chip isn't already programmed correctly. The Windows tool is part of the Terratec mixer which itself won't run if the chip isn't properly programmed, meaning you're stuck in DOS for doing this stuff. The card also reverts to the DOS mixer settings if you run a DOS application in Windows, but you can't use the DOS mixer in Windows because it is the same TERRATEC.EXE that runs in Windows 95 all the time - that one EXE literally does everything.


Needless to say I got things to do what they should but then forgot to run the caches back on, resulting in half an hour of wondering why my lovely new K5 was running so slow all of a sudden.

Otherwise, it's a good card and it makes up for the problems if you can get it to work. Case and point;
https://www.youtube....h?v=BL7ynTYteg4

-----

The system I have it installed in exhibits another odd behaviour. If I install a SCSI card (which I have) then it does what you'd expect it to do, unless you hit CTRL+ALT+DELETE on a failed boot, this will somehow corrupt the BIOS EEPROM on the motherboard, writing FF's to the end of the EEPROM for whatever reason. I got around it by removing the "Flash Voltage" jumper from the board, meaning the chip cannot be written to, but I've no idea why the system attempts this in the first place. All BIOS versions for this board also hide the "Memory Hole at 15-16MB" option in the CMOS setup, but leave it enabled by default so MODBIN is required. Being an older board I suppose it is entirely possible they just assumed nobody would install that much RAM, but it had me scratching my head for a while as to why POST showed my 32MB EDO count up only for Windows to report I had only 15MB installed. I doubt I really need 32MB, but it was there on the desk and I have the 512K of cache to stomach it, so why the hell not.

I love working with old machines for this stuff though, trying to figure things out and get it all to work keeps my brain ticking over.

@TDRR; Funny, 320GB is almost exactly the same size (319.92GB) as my por- err.. games folder. Probably not healthy, but fuck it.

[Minigunner]

Though not particularly interesting, my computer's story is certainly bad. Got transported in my sister's car, which had poor suspension, plus my computer was not restrained or even packaged, so it probably moved around a bit. I got it running, and only a couple minutes in my secondary drive got the Click. My main drive, however, has been slowly dying over the 10 months since, with occasional read errors leading to corrupted sectors and in some cases completely fucked files that cause a BSOD when attempting to read them. I have to run a disk check pretty much every time I restart (I usually put my computer in Hibernate instead).

[Kawa]

Any of you guys remember Visual Basic 3? I used it back in the good ol' days. It had a Screen object you could reference to find out things like resolution. At one point in my early life I was trying to make a screensaver that involved bugs moving up the screen at random intervals. Somehow, a multiplication by Screen.TwipsPerPixelY bluescreened the system.

I would've been more sad by the total loss of all the code work on that screensaver if it wasn't so little yet.

But I can do you one weirder. Remember ROM BASIC? You turn on your Commodore 64 or Apple ][ and you get a READY > prompt? I was building my first Pentium and apparently I'd gotten a cable very very wrong somewhere. Mind you, this was a brand new mobo with a brand new CPU. The only old things from my previous system were the disk drives and cards.

Imagine my surprise.

[High Treason]

I've seen a few 486 boards do that. I've never looked into it, but I'd hazard a guess it was a left over from when some systems did run DOS, BASIC, CP/M or whatever from ROM. There are funny things in some early Pentium BIOSes and that would qualify as one of them, likely carried over from earlier systems. My 66MHz Pentium has a BIOS which contains references to a 486DX3 and some clock speeds that never actually existed outside of testing. That machine qualifies as strange by its mere existence though, really, Socket 4 never did take off.

[Kawa]

Well, it was an early Pentium. 99.999% likely to have had the FDIV bug too.

Doesn't explain the other thing though -- that was somewhat later.

[TDRR]

Imagine having to boot Ubuntu from a live usb every now and then to format the drive so you can keep it working, lol.

Why not simply buy a new drive though? I don't know the specs of that PC but it would be a shame to see it go just because of the hard drive.