Comment by titzer
10 months ago
Virgil doesn't use frame pointers. If you don't have dynamic stack allocation, the frame of a given function has a fixed size can be found with a simple (binary-search) table lookup. Virgil's technique uses an additional page-indexed range that further restricts the lookup to be a few comparisons on average (O(log(# retpoints per page)). It combines the unwind info with stackmaps for GC. It takes very little space.
The main driver is in (https://github.com/titzer/virgil/blob/master/rt/native/Nativ... the rest of the code in the directory implements the decoding of metadata.
I think frame pointers only make sense if frames are dynamically-sized (i.e. have stack allocation of data). Otherwise it seems weird to me that a dynamic mechanism is used when a static mechanism would suffice; mostly because no one agreed on an ABI for the metadata encoding, or an unwind routine.
I believe the 1-2% measurement number. That's in the same ballpark as pervasive checks for array bounds checks. It's weird that the odd debugging and profiling task gets special pleading for a 1% cost but adding a layer of security gets the finger. Very bizarre priorities.
You can add bounds checks to c, but that costs a hell of a lot more than 1-2%. C++ has them off by default for std::vector because c++ is designed by and for the utterly insane. Other than that, I can't off the top of my head think of a language that doesn't have them.
The bounds safety C compiler extension research by Apple has measured the runtime impact of adding bounds checking to C and it is not a lot more than 1-2% in almost all cases. Even in microbenchmarks its often around 5%. The impact on media encoding and decoding was around 1-2% and the overall power use on the device did not change.
https://www.youtube.com/watch?v=RK9bfrsMdAM https://llvm.org/devmtg/2023-05/slides/TechnicalTalks-May11/...
It's a myth that bounds checking has extraordinary performance costs and cannot be enabled without slowing everything to a halt. Maybe this was the case 10 years ago or 20 years ago or something, but not today.
> C++ has them off by default for std::vector because c++ is designed by and for the utterly insane.
And for those who value performance and don't want to pay the cost of "a lot more than 1-2%" ;p
The data I've seen for turning on bounds checks in std::vector shows overhead considerably lower than 1-2%.
std::vector falls into the category of things that are easy to bounds check, st the cost, even under today's primitive compilers, is low. It's direct pointer accesses—which are common in c but not in c++ or most other languages—that are hard to and therefore cost more to bounds check.
1 reply →
std::regexp, std::map, fronzen ABI.... apparently the value for performance is relative at WG21.