Benchmark                            Metric                    Change
  ===============================================================================
- Init.Data.List.Sublist async         branches                    1.8%  (97.5 σ)
- Init.Data.List.Sublist async         instructions                1.6%  (75.8 σ)
- Init.Prelude async                   branches                    3.6% (244.5 σ)
- Init.Prelude async                   instructions                3.3% (196.8 σ)
- Std.Data.Internal.List.Associative   branches                    1.6% (181.9 σ)
- Std.Data.Internal.List.Associative   instructions                1.4% (155.2 σ)
+ big_deceq_rec                        branch-misses              -1.5% (-20.6 σ)
+ libleanshared.so                     binary size                -1.9%
- stdlib                               process pre-definitions     1.4%  (22.9 σ)
- stdlib                               type checking               1.3%  (62.6 σ)
+ stdlib size                          bytes .ir                  -1.5%
+ stdlib size                          lines C                    -1.4%
+ stdlib size                          max dynamic symbols        -1.7%
+ tests/compiler                       sum binary sizes           -1.3%

  Benchmark                      Metric                       Change
  ============================================================================
- Init.Data.List.Sublist async   branches                       1.1%  (45.6 σ)
- Init.Prelude async             branches                       2.5%  (85.5 σ)
- Init.Prelude async             instructions                   2.3%  (72.3 σ)
- binarytrees                    maxrss                         2.1%  (76.0 σ)
+ libleanshared.so               binary size                   -1.9%
+ stdlib                         grind mark subsingleton       -2.0% (-31.4 σ)
- stdlib                         let-to-have transformation     1.5%  (67.6 σ)
+ stdlib size                    bytes .ir                     -1.5%
+ stdlib size                    lines C                       -1.4%
+ stdlib size                    max dynamic symbols           -1.7%
+ tests/compiler                 sum binary sizes              -1.3%

  Benchmark            Metric                       Change
  ===================================================================
- Init.Prelude async   branches                       1.4%   (61.1 σ)
- Init.Prelude async   instructions                   1.3%   (46.4 σ)
- channel.lean         boundedn_seq                   1.4%
+ libleanshared.so     binary size                   -1.9%
+ riscv-ast.lean       maxrss                        -1.6%  (-39.0 σ)
+ riscv-ast.lean       task-clock                    -1.7% (-341.4 σ)
+ simp_subexpr         task-clock                    -2.8%  (-35.7 σ)
+ simp_subexpr         wall-clock                    -2.9%  (-30.3 σ)
+ stdlib               compilation (IR)              -1.8%  (-25.1 σ)
- stdlib               let-to-have transformation     3.0%   (21.2 σ)
+ stdlib               tactic execution              -1.2%  (-22.9 σ)
+ stdlib size          bytes .ir                     -1.5%
+ stdlib size          lines C                       -1.3%
+ stdlib size          max dynamic symbols           -1.7%
+ tests/compiler       sum binary sizes              -1.3%
+ unionfind            task-clock                   -11.5%  (-20.1 σ)
+ unionfind            wall-clock                   -11.4%  (-20.0 σ)

  Benchmark                 Metric                      Change
  =======================================================================
- Init.Prelude async        branches                      1.3%  (168.4 σ)
- Init.Prelude async        instructions                  1.2%  (130.1 σ)
- binarytrees               maxrss                        2.1%   (34.7 σ)
- channel.lean              boundedn_seq                  1.4%
+ libleanshared.so          binary size                  -1.7%
+ omega_stress.lean async   branch-misses                -2.3%  (-20.8 σ)
+ riscv-ast.lean            branch-misses                -3.1%  (-92.5 σ)
+ stdlib                    grind typeclass inference    -2.3% (-194.8 σ)
+ stdlib                    longest rebuild path         -1.9%  (-31.8 σ)
+ stdlib size               bytes .ir                    -1.3%
+ stdlib size               lines C                      -1.2%
+ stdlib size               max dynamic symbols          -1.7%
+ tests/compiler            sum binary sizes             -1.2%