Home | Notifications | New Note | Local | Federated | Search | Logout
Note Detail
Reply to @david_chisnall@infosec.exchange
Sun Microdevil Pte Ltd@koakuma@uwu.social (2026-08-17 01:43:56)
@david_chisnall In SPARC, at least, there's the #ASI_PNF loads that would eat the trap and return zeros instead (though it doesn't set any status codes in the CCR)
I've yet to see any compilers emitting it, I suppose partly cus you can't really tell between a failed load and a successful one that just happens to return zero :char_nachoneko_baa:
---Reply---
David Chisnall (*Now with 50% more sarcasm!*)@david_chisnall@infosec.exchange (2026-08-17 01:53:18)
@koakuma
SPARC had a few things for working with Smalltalk-like tagged integers. I wonder if Smalltalk on SPARC used this. You could load the values unconditionally early on in a function and then use them after the tag checks.
Reply
---Replies---
Sun Microdevil Pte Ltd@koakuma@uwu.social (2026-08-17 02:20:53)
@david_chisnall The latter part is exactly how the v9 ISA book recommends it to be used (though it uses branches as an example), if my memory serves me right
I dunno how much of an impact it would have on modern heavily-optimizing implementations, though, in particular because those things would already speculate the loads anyway, no?