feat: non exponential codegen for reset-reuse#12665
Conversation
|
!bench |
|
Benchmark results for d851a15 against d20b6ec are in! @hargoniX
Large changes (5✅, 38🟥) Too many entries to display here. View the full report on radar instead. Medium changes (7✅, 84🟥) Too many entries to display here. View the full report on radar instead. Small changes (20✅, 2113🟥) Too many entries to display here. View the full report on radar instead. |
|
!bench |
|
Benchmark results for 1657843 against d20b6ec are in! @hargoniX
Large changes (3✅)
Medium changes (3✅, 1🟥)
Small changes (51✅, 922🟥) Too many entries to display here. View the full report on radar instead. |
|
Mathlib CI status (docs):
|
|
Reference manual CI status:
|
|
!bench |
|
Benchmark results for 1414e72 against d20b6ec are in! @hargoniX
Large changes (3✅)
Medium changes (3✅, 1🟥)
Small changes (52✅, 919🟥) Too many entries to display here. View the full report on radar instead. |
|
!bench |
|
Benchmark results for 49e4a0b against d20b6ec are in! @hargoniX
Large changes (11✅, 16🟥) Too many entries to display here. View the full report on radar instead. Medium changes (5✅, 18🟥) Too many entries to display here. View the full report on radar instead. Small changes (54✅, 944🟥) Too many entries to display here. View the full report on radar instead. |
49e4a0b to
d598d0a
Compare
|
!bench |
|
Benchmark results for d598d0a against cd7f55b are in! @hargoniX
Large changes (11✅, 15🟥) Too many entries to display here. View the full report on radar instead. Medium changes (5✅, 13🟥)
Small changes (72✅, 629🟥) Too many entries to display here. View the full report on radar instead. |
|
!bench |
|
Benchmark results for 618677f against cd7f55b are in! @hargoniX
Large changes (11✅, 15🟥) Too many entries to display here. View the full report on radar instead. Medium changes (5✅, 15🟥)
Small changes (71✅, 721🟥) Too many entries to display here. View the full report on radar instead. |
|
!bench |
|
Benchmark results for 0a23f5b against cd7f55b are in! @hargoniX
Large changes (5✅, 25🟥) Too many entries to display here. View the full report on radar instead. Medium changes (4✅, 27🟥) Too many entries to display here. View the full report on radar instead. Small changes (26✅, 1455🟥) Too many entries to display here. View the full report on radar instead. |
|
!bench |
|
Benchmark results for 9bb5aa4 against cd7f55b are in! @hargoniX
Large changes (11✅, 15🟥) Too many entries to display here. View the full report on radar instead. Medium changes (5✅, 16🟥) Too many entries to display here. View the full report on radar instead. Small changes (71✅, 719🟥) Too many entries to display here. View the full report on radar instead. |
|
!bench |
|
Benchmark results for e6ae02b against cd7f55b are in! @hargoniX
Large changes (11✅, 15🟥) Too many entries to display here. View the full report on radar instead. Medium changes (5✅, 15🟥)
Small changes (74✅, 699🟥) Too many entries to display here. View the full report on radar instead. |
|
!bench |
|
Benchmark results for 53d0803 against cd7f55b are in! @hargoniX
Large changes (15✅, 8🟥) Too many entries to display here. View the full report on radar instead. Medium changes (10✅)
Small changes (374✅, 47🟥) Too many entries to display here. View the full report on radar instead. |
f23e947 to
186b394
Compare
|
!bench |
|
Benchmark results for 162e194 against e96d969 are in! @hargoniX
Large changes (15✅, 8🟥) Too many entries to display here. View the full report on radar instead. Medium changes (8✅, 1🟥)
Small changes (393✅, 6🟥) Too many entries to display here. View the full report on radar instead. |
162e194 to
b738255
Compare
This PR ports the expand reset/reuse pass from IR to LCNF. In addition it prevents exponential code generation unlike the old one. This results in a ~15% decrease in binary size and slight speedups across the board.
The change also removes the "is this reset actually used" syntactic approximation as the previous passes guarantee (at the moment) that all uses are in the continuation and will thus be caught by this.