3 ms·

I thought IDE was a nightmare, cheap and unreliable compared to SCSI. For instance, one has to fiddle around with jumper settings on each connected IDE device (

I thought IDE was a nightmare, cheap and unreliable compared to SCSI. For instance, one has to fiddle around with jumper settings on each connected IDE device (master, slave!) whereas SCSI was a lot more intelligent and stable (and obviously a lot more expensive). I still have my old Adaptec 19160U somewhere, it’s a really beautiful piece of engineering.

bogantech 5y ago
You need to jumper each SCSI device to set an address too but I guess sometimes the defaults don't clash?
- Sesse__ 5y ago
  Larger systems would usually use drive bays with 80-pin SCSI, which was the regular 68-pin SCSI, plus automated address configuration (ie., you got the address assigned to the bay), plus power, all in one connector.
  - bogantech 5y ago
    TIL! That's very cool
morsch 5y ago
IDE drives had two jumpers to set master vs slave, SCSI drives had four jumpers or dip switches to configure the bus id... And having to terminate the bus was kind of annoying. I'm sure things got more convenient (but didn't late era PATA also have auto configure for master/slave?).
- dspillett 5y ago
  > IDE drives had two jumpers to set master vs slave Or in some cases one to choose between to the two. > but didn't late era PATA also have auto configure for master/slave?) That was “cable select”, enabled with either an extra jumper or (common if the were separate master and slave jumpers) by jumpering for neither master or slave. Here one of the ground pins was not connected in the far end connector on the cable so a drive could decide what it was by inspecting the properties of that line. You could make a CS compatible cable by cutting that line on a non-CS cable. I remember this being around at least as early as 1999, and drives coming set to CS by default at least once 80-conductor cables were common, though I also remember never really trusting it and explicitly setting Master/Slave as I wanted - setting one or two jumpers was no hardship (the vast majority of drives had a little jumper diagram on the printed label so you didn't have to look up the magic anywhere), and you had a touch more cabling freedom (you didn't have to care which connector master was plugged into, it was master irrespective) which was a small mercy with wide ribbon cables.
- garaetjjte 5y ago
  It wasn't so straightforward in the early days of IDE: https://www.os2museum.com/wp/the-dual-drive-ide-hell/ https://www.os2museum.com/wp/the-dual-drive-ide-hell/
hulitu 5y ago
SCSI was more inteligent. The onlx issue people had was when 2 devices on the bus had the same SCSI id :)
vidarh 5y ago
As an Amiga user, Commodore's choice to go IDE instead of SCSI for the Amiga 4000 was almost seen like a betrayal and a sellout by some users, and it contributed to the sense of impending doom - the Amiga had long compensated for not keeping up in CPU speed increases by offloading everything, and SCSI was part of that... But of course it was also part of what made it expensive.

bogantech 5y ago

You need to jumper each SCSI device to set an address too but I guess sometimes the defaults don't clash?

Sesse__ 5y ago
Larger systems would usually use drive bays with 80-pin SCSI, which was the regular 68-pin SCSI, plus automated address configuration (ie., you got the address assigned to the bay), plus power, all in one connector.
- bogantech 5y ago
  TIL! That's very cool

morsch 5y ago

IDE drives had two jumpers to set master vs slave, SCSI drives had four jumpers or dip switches to configure the bus id... And having to terminate the bus was kind of annoying. I'm sure things got more convenient (but didn't late era PATA also have auto configure for master/slave?).

dspillett 5y ago
> IDE drives had two jumpers to set master vs slave Or in some cases one to choose between to the two. > but didn't late era PATA also have auto configure for master/slave?) That was “cable select”, enabled with either an extra jumper or (common if the were separate master and slave jumpers) by jumpering for neither master or slave. Here one of the ground pins was not connected in the far end connector on the cable so a drive could decide what it was by inspecting the properties of that line. You could make a CS compatible cable by cutting that line on a non-CS cable. I remember this being around at least as early as 1999, and drives coming set to CS by default at least once 80-conductor cables were common, though I also remember never really trusting it and explicitly setting Master/Slave as I wanted - setting one or two jumpers was no hardship (the vast majority of drives had a little jumper diagram on the printed label so you didn't have to look up the magic anywhere), and you had a touch more cabling freedom (you didn't have to care which connector master was plugged into, it was master irrespective) which was a small mercy with wide ribbon cables.
garaetjjte 5y ago
It wasn't so straightforward in the early days of IDE: https://www.os2museum.com/wp/the-dual-drive-ide-hell/ https://www.os2museum.com/wp/the-dual-drive-ide-hell/

hulitu 5y ago

SCSI was more inteligent. The onlx issue people had was when 2 devices on the bus had the same SCSI id :)

vidarh 5y ago

As an Amiga user, Commodore's choice to go IDE instead of SCSI for the Amiga 4000 was almost seen like a betrayal and a sellout by some users, and it contributed to the sense of impending doom - the Amiga had long compensated for not keeping up in CPU speed increases by offloading everything, and SCSI was part of that... But of course it was also part of what made it expensive.