3 ms·

Sutherland was something like a cofounder of Sun Labs, so I guess he was likely to develop some interest. I'm afraid I couldn't track down anything more specifi

by chalst 7y ago

Sutherland was something like a cofounder of Sun Labs, so I guess he was likely to develop some interest. I'm afraid I couldn't track down anything more specific, but I'll keep my eyes peeled. An interesting point I did find: the SPARC architecture was later extended so that register windows could be saved and restored other than via SUB calls, allowing instruction reordering and their use for context switching. The Register window article doesn't talk about Sun's experience or the issue of reordering instructions.

throwaway_pdp09 7y ago
I didn't know he was a sun co-founder. I remember him saying that he had an interest in geometry rather than graphics, which related to chip design - but that link is to me a bit nebulous, and it was a long time ago anyway. Sun would not advertise an horrifically expensive design mistake, so no surprise you can't find much. I've picked up a fair bit from random reading around over the years so I can't remember where much of it came from. Perhaps email Mr. Sutherland and just ask? Worst that can happen is he doesn't respond. (thanks for the bit about using other than SUB calls, I didn't know).
- jecel 7y ago
  Not Sun but Sun Labs. Ivan's Sutherland, Sproull and Associates was bought by Sun in 1990 to become the seed of the new Sun Labs. About register windows, overflows and underflows generated traps to the operating system, and for the combination of Sparc version 8 and SunOS that meant thousands of clock cycles. That was improved in later products. Berkeley's RISC I to IV all had register windows but RISC-V doesn't with the argument that we have far better compilers now. Altera's NIOS processor had register windows which were dropped in NIOS II because it made the processor smaller without reducing the performance too much. The AMD 29000 had a more flexible register window scheme and the Itanium a very complex scheme. Computer history is often more like a spiral than a line and old ideas that have become bad might be good again in the future. With out-of-order execution and register renaming you might once again get better performance out of a binary with register windows.

by chalst 7y ago

throwaway_pdp09 7y ago

I didn't know he was a sun co-founder. I remember him saying that he had an interest in geometry rather than graphics, which related to chip design - but that link is to me a bit nebulous, and it was a long time ago anyway. Sun would not advertise an horrifically expensive design mistake, so no surprise you can't find much. I've picked up a fair bit from random reading around over the years so I can't remember where much of it came from. Perhaps email Mr. Sutherland and just ask? Worst that can happen is he doesn't respond. (thanks for the bit about using other than SUB calls, I didn't know).

jecel 7y ago
Not Sun but Sun Labs. Ivan's Sutherland, Sproull and Associates was bought by Sun in 1990 to become the seed of the new Sun Labs. About register windows, overflows and underflows generated traps to the operating system, and for the combination of Sparc version 8 and SunOS that meant thousands of clock cycles. That was improved in later products. Berkeley's RISC I to IV all had register windows but RISC-V doesn't with the argument that we have far better compilers now. Altera's NIOS processor had register windows which were dropped in NIOS II because it made the processor smaller without reducing the performance too much. The AMD 29000 had a more flexible register window scheme and the Itanium a very complex scheme. Computer history is often more like a spiral than a line and old ideas that have become bad might be good again in the future. With out-of-order execution and register renaming you might once again get better performance out of a binary with register windows.