perf: lower the default allocation for bignums #4102
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
ggreif
commented
Jul 5, 2023
ggreif
commented
Jul 5, 2023
rts/motoko-rts/src/types.rs
Outdated
| @@ -921,6 +921,10 @@ pub(crate) unsafe fn block_size(address: usize) -> Words<u32> { | |||
|
|
|||
| TAG_BIGINT => { | |||
| let bigint = address as *mut BigInt; | |||
| if (*bigint).mp_int.used != (*bigint).mp_int.alloc { | |||
| println!(100, "used:{}, alloc: {}", (*bigint).mp_int.used, (*bigint).mp_int.alloc); | |||
There was a problem hiding this comment.
Building up the arr (wins back 1-2 words)
debug.print: used:3, alloc: 5
debug.print: used:5, alloc: 7
debug.print: used:9, alloc: 11
debug.print: used:18, alloc: 19
debug.print: used:36, alloc: 37
debug.print: used:72, alloc: 73
debug.print: used:143, alloc: 145
debug.print: used:286, alloc: 287
debug.print: used:572, alloc: 573
Deserialising from Candid blob (wins back 1 word consistently)
debug.print: used:3, alloc: 4
debug.print: used:5, alloc: 6
debug.print: used:9, alloc: 10
debug.print: used:18, alloc: 19
debug.print: used:36, alloc: 37
debug.print: used:72, alloc: 73
debug.print: used:143, alloc: 144
debug.print: used:286, alloc: 287
debug.print: used:572, alloc: 573
ggreif
commented
Jul 5, 2023
ggreif
commented
Jul 5, 2023
rts/motoko-rts/src/types.rs
Outdated
| if (*bigint).mp_int.used != (*bigint).mp_int.alloc { | ||
| println!(100, "used:{}, alloc: {}", (*bigint).mp_int.used, (*bigint).mp_int.alloc); | ||
| } | ||
| (*bigint).mp_int.alloc = (*bigint).mp_int.used; |
There was a problem hiding this comment.
stats are in https://github.com/dfinity/motoko/pull/4102/files/e15afc367f58f61db4756f6f41e513890d7c8163#r1253257503 and also past commits...
ggreif
force-pushed
the
gabor/bignum-space
branch
from
7185dc5 to
d4a6d4a
Compare
this addresses #4100, but more improvements are to come
ggreif
force-pushed
the
gabor/bignum-space
branch
from
d4a6d4a to
50420e0
Compare
But probably better to redesign the logic and make BigInt shortening independent of `word_size()`
ggreif
commented
Jul 6, 2023
rts/motoko-rts/src/bigint.rs
Outdated
| @@ -51,6 +51,7 @@ unsafe fn mp_alloc<M: Memory>(mem: &mut M, size: Bytes<u32>) -> *mut u8 { | |||
| let size = size.as_usize(); | |||
| debug_assert_eq!((size % core::mem::size_of::<mp_digit>()), 0); | |||
| (*blob).mp_int.alloc = (size / core::mem::size_of::<mp_digit>()) as i32; | |||
| (*blob).mp_int.used = (*blob).mp_int.alloc; // Otherwise, `block_size() in `allocation_barrier()` may already shorten the object. | |||
There was a problem hiding this comment.
The problem with this might be that this changes the value of the bignum (mantissa).
luc-blaeser
approved these changes
Jul 12, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
this addresses the most pressing aspect of #4100, but more improvements are to come
Possible flanking measures:
HPstill points at the end of the recently allocated bignum after all arithmetic is done, and the bignum's capacity is bigger than its payload, we can cut back theHP(this probably only makes sense from specific callsites)