On Wed, 28 Nov 2001, Preston L. Bannister wrote: > Adam has pretty nicely summarized my point of view.
me too, actually. Because ... > If the motherboard EEPROMs were commonly 512KB or bigger then > the LinuxBIOS approach would be incredibly cool. yep, that's the problem. When we started this project 2 years ago, I was pretty sure the economics of things like MP3 players -- MP3s use basically all the flash, or so the flash people tell me -- would have pushed the cheapest flash size up, so that motherboards would routinely have 1MB or larger flash by 2001. Kind of like the cheapest DRAM and cheapest disk keep growing in size. We just figured that 1MB and up FLASH would be the rule. ACPI seemed to be something that would push that along. Continued presence of 256KB FLASH on mainboards is a real disappointment. I really did not count on 256KB flash hanging around so long. Still worse, Linux is getting so big that 512KB can't even hold a kernel with networking and IDE both turned on! So, until we get those motherboards with big flash, we're stuck with the intermediate boot program of the type Adam is doing. For clusters, however, we will continue to only use motherboards that have large flash or a 32-DIP socket. > Now from browsing around the embedded Linux web pages, it > seems that fitting a kernel into 256KB (compressed) is not > impossible, but then I have not gone through the exercise. I'd like to hear more :-) Is it possible? We've actually discussed putting 2.0 into flash, but then you have different problems. ron
