Rick's b.log - 2009/11/27 |
|
It is the 21st of November 2024 You are 3.147.71.175, pleased to meet you! |
|
mailto:
blog -at- heyrick -dot- eu
Then came a boom. I'm afraid neither HTML nor Flash can do justice to this boom. George Lucas would wet his pants if THX could accurately reproduce this boom. The world shook and the windows rattled. It was that kind of boom. And it dragged on. And on. And we've had fighter jets scream overhead above supersonic that time a budget airliner entered French airspace with a broken radio and was heading towards a major city. The air force is allowed to break the rules and fly supersonic over land in defence of its territory. Piffle. Nothing compared to this boom.
And then, as I said. Darkness.
Then a little bit of light. Just a little, like the power was running around 80v or so. For a second. Then fifteen seconds of dark again. Then a hint of light. This cycle repeated three times before it gave up.
In a maaad panic, because I know mom hates thunderstorms and I had all my computers on, I dashed to the breaker box and pushed the big button. Then to check on mom. Then a walk around the house in the tipping rain. At that point I could not be entirely sure if it was a bolt of lightning or a massive explosion.
The rain is severe and heavy. But, oddly, no lightning. Not a flicker. I am leaning towards the lightning theory as an explosion like that in the next town would probably have wiped half of it off the map (and I mean an explosion like this) and I don't hear any sirens. Certainly no helicopters with "Sky News" painted on the side (image from the Sky Press Office).
Still dark.
I console myself that Azumi survived, so I could watch a movie off batteries. Indeed I am sitting in the kitchen shivering and writing this. I meant to put the kettle on but got sidetracked. I'll do it now.
Okay. So it's dark. And wet. And I'm worried if my equipment hasn't just been toasted. ONE bolt of lightning and it takes us out. Cheers.
It goes past midnight. I was heating water for a cup of tea using the camping gas burner. The cannister is low so it doesn't burn well. I managed, what with giving the thing a flick every thirty seconds, to bring some water to the boil for a cuppa. It was just starting to boil when...
...light! Whoo-hoo!
I wait for a few minutes, in case it is just a test. But no, the power stays on. I hear a little "bing" and I know the Livebox has just powered up the phone. So internet is back up. The PVR is okay. The satellite receiver is on. I power up Ayleigh and it starts, as had the router.
Unbelievable. It looks like the EDF cables took a near direct hit (I think it hit the cables, not the house) and nothing got toasted. I always power down in thunderstorms as I've seen what happened to a computer after a lightning strike, evidently the only thing that still worked was the floppy drive, the rest was landfill.
But today... Mom listens to Radio4 on Long Wave and tells me of impending storms. Lightning comes across as crackles on the radio. Nothing. Nothing before, nothing after. Just one massive mother------ of a bolt that was so intense it blinded the world outside. I've seen lighning. I like to stand under cover and watch nature's anger. While I've seen some pretty bright sparks, I've not seen one so bright the world turned into a white-out. This was something else. As was EDF's ability to dissipate at the risk of their equipment so ours would be okay.
If you don't fully understand the differences... The main way of accessing harddiscs and network adaptors and such is "DMA". This means Direct Memory Access (known as "BusMastering" on many PCI cards). In this mode, the processor sets up a data transfer and the hardware then gets on and does it independently of the processor. This means a harddisc can read many megabytes per second which the controller dumps directly into memory. The processor, meanwhile, can be doing other stuff. The processor will receive an interrupt when the data transfer is complete. That's like an "oi! you!" so it knows the situation.
By contrast, in PIO mode, the processor must fetch every bit of data from memory and write it to the harddisc controller (to write to the disc), or the reverse to read. As the processor is doing all this itself, the CPU use will max out, and in addition to that, the speed will drop dramatically not only due to CPU intervention but also for all the other things the computer does - clock ticks, video interrupts, timeslicing in a modern multitasker, the software playing that internet radio station...
I opened the box. Problem? All cables correct. Power supply within parameters. It writes, it verifies, so the thing can't surely be failing.
I delete the DVD drive from the hardware settings and reboot.
I finally got it to work by deleting the secondary IDE controller and the DVD writer, then rebooting for Windows to redetect the hardware.
Looking on Microsoft's WHDC, it says:
This approach FAILS for several reasons:
OMFG!
So having fixed a rather annoying quirk with Windows XP (more info below), I set about collecting some stuff to burn off a DVD-R for a friend. After a file or twenty had copied, I went into the kitchen and put the kettle on. Into the back kitchen to pass time stroking the cat, when suddenly it went intense white. Like if they dropped a nuke nearby. Then total darkness.
I'd like to tell you what my first thought was, but I didn't have one. Not a "WTF?", not a "OMG!", nothing. Well, maybe a "fuuuuuuuuu....".
DVD PIO
For a while my DVD writer was stuck in PIO mode. To give you an idea of what this means, cheap'n'cheerful old hardware talks to IDE devices in PIO mode. I'm not entirely certain what the acronym means, "Polling I/O" perhaps? I just know that an IDE interface for a 6502-based computer (clocked at 1MHz) talks 8 bit PIO to a harddisc. And I would imagine that IDE card in my A3000 that has a suspicious 6522 VIA chip works in exactly the same way, using the VIA to fake up an 8 bit IDE interface. There's a 16 bit version as well, but it is about as sophisticated as an '80s parallel printer.
Imagine burning a DVD like that.
Don't, actually. Credit to it, the 1.1GHz processor managed to do a sort-of 2× speed running flat out with the CPU usage off the chart. Thank God for buffer underrun protection, I burned a number of DVDs this way, each taking over an hour.
For the sake of completeness, DMA and Bus Mastering are not exactly identical. DMA means accessing system memory directly, however it is performed. Older hardware needed to have a specific DMA chip to handle DMA channels, so effectively it relieved the processor of the tedious read-write by doing it itself. Bus mastering took this to its logical conclusion where a device is able to effectively claim control of the data and address signals in the computer in order to permit the device to write directly to memory itself (well, via SouthBridge assistance, maybe more, as PCI isn't directly on the memory bus...). Contrast with the old days when the address signals were emitted by the processor and that was that.
Redetected. In PIO mode.
If more that six DMA transfer timeouts occur, Windows will turn off DMA and use only PIO mode on that device.
Now this might make sense on bugged hardware, but in most cases I can't see how this behaviour is useful. Will a harddisc work better in PIO mode if failure was imminent? I doubt it.
As for CD/DVD reading, tell me what's the most likely cause of DMA timeouts? Dying hardware? Nope! It's crappy discs. So you'd better pay attention to this article for if you borrow a disc from the library, try to rip it, and it has scratches large enough to cause errors, you might just find Windows decides to revert your hardware to a really slow mode...
No comments yet...
© 2009 Rick Murray |
This web page is licenced for your personal, private, non-commercial use only. No automated processing by advertising systems is permitted. RIPA notice: No consent is given for interception of page transmission. |