>It now boots SunOS 2.0, 3.2, and 3.5 cleanly. You can install from the distribution tapes. The SCSI emulation code is still shakey for tapes, however.

SCSI was always shakey for all devices. As soon as they make the SCSI emulation code shakey for disks as well as tapes, the emulation will be perfect!

I sometimes feel like I'm the only person who has never had big problems with SCSI, even with external cases and chaining a bunch of drives.

Though it sure is fun to talk about the ritual of the two terminations.

You're not the only one. I've used SCSI with Classic Macs, SGI/IRIX and Sun/SunOS and never had a problem.

On Windows/Linux/FreeBSD/x86 I've had many problems setting things up and getting decent throughput and I think it's just I've always had cheap/old/dubious SCSI cards.

It seemed usually only a "problem" initially when setting things up. I had a homebuilt system with:

- Adaptec AHA-2940U2W w/ internal and external ports

- IBM UltraStar 9 GB (!) U2W 25L1910

- SCSI Plextor CD burner (before DVDs existed)

- IOmega ZIP External SCSI drive

- Plextor External SCSI CD changer

- SCSI scanner

- Internal TEAC SCSI 3.5" floppy drive

- Internal SCSI tape drive

For the era, SCSI worked pretty well. That was back in the day when I had an early Zalman copper CPU heatsink with its 7k RPM, 1U-server-loud Delta fan.

„There are perfectly valid technical reasons why one has to sacrifice a young goat to get SCSI to work correctly.”

As with real estate: Termination, Termination and ... Termination.

I feel like I got a more practical education on transmission lines struggling with SCSI cables than taking electromagnetics courses.

Plug in two 40 MB SCSI external hard disks and metadisk (like RAID 0) them together. Good ol' daisy-chaining those giant Centronics/DB50 to DD-50 expensive/proprietary cables. Just don't kick a hard drive's wall wort or it'll be fsck-ing for the next hour.

PS: I was the sysadmin for a lab with a Sun 4, Sun SPARCstation 20, AIX, SCO and HP-UX boxes. There were some Linux and Windows 3.1/NT boxen too. All networked together, each had a valid IP from a routable Class C and named after Simpson characters.

Much different but will probably appeal to the same crowd: a sparcstation5 fpga workalike http://temlib.org/site/?page_id=14

Hey, that's really, really cool.

Temlib was mentioned in one of the comments here last year: https://news.ycombinator.com/item?id=21303446

I wonder if it's been used on a Pano Logic G2.

Awesome, now I'd love to be able to run (via emulator) a NeWS (Network extensible Window System) (networked display postscript) workstation... Google searches have turned up nothing...

I think a Sun 2 is too old for that. You need a Sun 3, Sun 4 or early SPARCstation to run NeWS.

I had a Sun 3/60, then later on a SparcStation 10 (bought them both used back in the 90's.) I loved Sun equipment. Never played around wit a Sun 2 but sounds fun.

>So I wrote this to learn about the MMU and, well SCSI. My goal is to eventually create an FPGA version of the Sun-2.

Okay - THAT I will want to buy, if it ever materializes.

Also see:

https://github.com/sigurbjornl/emulator-sun-2

Interesting work. Last commit was over 12 months ago though...

Are you... worried that it isn't keeping up with the rapid feature evolution of 68010 workstations?

All code bitrots, and no code is perfect the first time (or ever?).

If this code doesn't work on your computer you just need to run it in an appropriate emulator.

Code itself is one of the very few things that doesn't rot. Your execution environment might, but that's what emulators like this are for. :P

Hardware can be finished. If this faithfully implements the Sun 2, further changes defeat its purpose.

Not saying you're wrong, but faithful hardware emulation is really hard. The VICE team are still working hard on perfecting a Commodore 64 emulator and recently completed their implementation of the VSP bug, which has been discussed elsewhere here on HN[0].

[0] https://news.ycombinator.com/item?id=5314959

But that mostly means it wasn't ok (as in not faithfully emulating the hw bug) to begin with, and that it was super hard to fix, not that the VICE code rot'ed over time.