On Fri, Feb 11, 2005 at 06:31:56AM +0000, walter fry wrote: > In my bootstrap listing there is a line "ATI Radeon IGP Northbridge not > fully tested" What does that mean? I hope it is not merely to raise my > frustration level, it is high enough allready.
If it's on your AMD-based laptop, it's a notice that the chipset your laptop uses is basically not very well tested by the kernel developers (and therefore may not work too well.) I recall some people mentioning this at a euglug meeting or another. The northbridge on a motherboard controls: - What kinds of memory are supported (SDRAM, DDR SDRAM, bus speeds, etc) - The AGP slot, if you've got one - The PCI slots, if you've got them - The PCI busses - Onboard video, if it's part of the northbridge If yours goes bad, your computer is a paperweight. If your OS doesn't support the northbridge, it won't run on the machine. Fortunately, there are generic standards for how these things are supposed to work, and most of the time, the chipmakers almost sortof actually even pretend to follow those standards. Some of the chipsets aren't made to be used in those generic compatibility modes, so the makers don't test very well. And of course, if there's a bug in their design, do you think they recall the defective parts? No, of course not! They just update the Windows driver for the chipset and maybe tell the Linux people if they feel like it. It's called the northbridge because it connects directly to the CPU and RAM and PCI cards. There's also a southbridge in the modern chipset, in all but a few cases. It controls: - Serial, parallel, keyboard, and mouse ports - USB - The ISA bus (you have one even if you don't have ISA slots) - IDE drives - Floppy drives - Ethernet, sometimes - Audio ports, sometimes You can pretty much count on the chipmakers to make at least one incredibly stupid blunder in any chipset generation. Usually, these problems are found quickly by Linux kernel people because lots of people have the chipsets and it's usually pretty easy to get the chipmakers to tell kernel developers how to fix these problems. Even so, the fixes usually involve stupid things like "reset the chip every few minutes", as was required for one ethernet chip that comes to mind. My old Duron system had a VIA 686B southbridge would even cause serious data corruption if you didn't do something braindead to keep it from happening. Someone suggested I'd have 15% faster disc access if I had a non-broken chipset. For most files, "before I could count to one" was fast enough in the real world. VIA seems to think so too. I guess when they put "Designed for Windows XP" on a product, anymore, they're not kidding. So anyway, given that there are established standards for these things, and they're often deviant from the standard in some braindead and broken manner, the Linux kernel complains if your chipset is one that is recognized as "nobody's actually really made sure this works without strange bugs and crashes yet." In the case of your AMD-based laptop, this is why many people said you would have been better off with an Intel laptop. (AMD laptops are much less common than Intel. Very different situation from desktops, where AMD chips are now preferred by many to Intel.) Hope that sheds some light on the state of the art (they call it art!?) _______________________________________________ EUGLUG mailing list [email protected] http://www.euglug.org/mailman/listinfo/euglug
