gwr@linus.UUCP (Gordon W. Ross) (02/06/90)
I will speak in defense of MYLEX since they've done well for me. Too bad Jay Ts and Mark Hilliard have not had such good luck. In article <190@metran.UUCP> of comp.unix.i386, jay@metran.UUCP (Jay Ts) writes: > I am using a home-built system based on Mylex's MI386-20 motherboard. > My system passed BIOS tests OK, but wouldn't boot UNIX. [It] freezes > at the "Booting the UNIX System..." message from ISC's 386/ix. [ He then describes his frustrating tech-support experience. ] > [...] I simply cannot recommend Mylex > products to *anyone* until they get their act together! This hang-during-boot problem is mentioned frequently, and is not specific to MYLEX boards. It is (as I understand it) caused by the BIOS setting the CPU status word to indicate presence of a math co-processor when in fact the BIOS is just emulating its presence. The AMI BIOS allows you to disable the "co-processor emulation" feature. I observed this problem while installing ESIX on my MYLEX (AMI BIOS) machine. With the AMI BIOS co-processor emulation disabled, UNIX boots and runs flawlessly! In article <110@gizzmo.UUCP> of comp.unix.i386, mark@gizzmo.UUCP (Mark Hilliard) writes: > This board has a NON-STANDARD caching system which is [incompatible] with > MOST dual ported memory cards. I have [extensively] tested Mylex boards > and had BAD experiences with all of the caching boards. Some companies > make custom work arounds for their cards to support the mylex board, > but only a few. I gave up on Mylex and started using AMI boards with > NO problems with caching and dual ported memory. I don't think there is any "STANDARD" caching system as far as dual-ported memory cards are concern. You should note, though that MYLEX can provide you with a replacement for one of the PALs that disables the cache for certain address ranges (you want it disabled for the address range used by any memory-mapped I/O devices). Alternatively, if your software supplier is cooperative, MYLEX does clearly document in the hardware manual how software can enable or disable the cache. The driver software could be changed to disable the cache while accessing the memory-mapped I/O device. I am very happy with the MYLEX board. I've never had difficulty with their tech. support (though I've only called once with questions). The board comes with very complete documentation written in English! -- -- Gordon W. Ross / The MITRE Corporation / Bedford, MA 01730 (U.S.A.) internet: gwr@linus.mitre.org uucp: {decvax,philabs}!linus!gwr