leirbag28 wrote:
Well Thats All folks! I sold it all yesterday for a Measly $20
and Included a PCI SCSI 3 contoller with an extra SCSI 2 9gb 7200RPM HD and also added a 450 Watt Pentium 4 PSU for $15.
I know I got ripped off, but I needed the cash right at the moment.
Don't feel too bad. You can get 200GB 7200RPM ATA or SATA drives these days for only twice that, which, 'sadly' enough, will likely trounce those simply for platter densities and large caches. With NCQ on recent SATA hardware, or software patches that simulate it (NetBSD has, for example), you can get close to one of SCSI's other major advantages.
I love SCSI myself, but after comparing throughput, I'll take a 100+GB drive over previous generation hardware in almost all cases.
...
Now, when it comes to popping wide devices on narrow chains, to be
absolutely assured of compatible cabling, it pays to use 50<->68 adapters which include "high-byte termination." 'Many' drives will function without it, but I had the luck to land some 4GB IBMs back in the day with no onboard termination whatsoever, and same decidedly couldn't cope with having half their bus floating. Got the wide side of the bus appropriately terminated, and bam, instantly recognized and working.
If you want to use narrow devices on wide chains, technically one's supposed to opt for 68<->50 pin adapters (unterminated, naturally) along the bus rather than placing two terminators at separate physical offsets. In practice, you can often just use something like a 2940UW which is built to cope with however they designed the two 'halves' to the bus. (I was once told that these designs are slightly more technical cheats -- more of an on-chip bridge versus a straight-through bus -- than just applying switchable termination, but I've never looked at an AIC chip pinout to find out if that's really the case.)
...
Kudos to all the
real gurus who've finally settled what really happens with a slow device on a fast bus.
