-
Notifications
You must be signed in to change notification settings - Fork 3
Performance of LLVM 14 vs LLVM 13
Laurynas Biveinis edited this page Sep 2, 2022
·
3 revisions
Filtered for unodb::db
:
-
micro_benchmark_key_prefix
: 3% slowdown (unpredictable_prepend_key_prefix
) to 0% speedup (unpredictable_cut_key_prefix
) -
micro_benchmark_n4
: 10% slowdown (n4_full_scan/100
) to 0% speedup (minimal_n4_sequential_insert/64
) -
micro_benchmark_n16
: 7% slowdown (minimal_n16_tree_full_scan/512
) to 1% speedup (n16_sequential_add/10
) -
micro_benchmark_n48
: 9% slowdown (shrink_n256_to_n48_randomly/64) to 5% speedup (
full_n48_tree_full_scan/128`) -
micro_benchmark_n256
: 7% slowdown (grow_n48_to_n256_sequentially/64
) to 12% speedup (grow_n48_to_n256_randomly/64
)
Filtered for unodb::olc_db
:
-
micro_benchmark_key_prefix
: 2% slowdown (unpredictable_get_shared_length
) to 1% slowdown (unpredictable_prepend_key_prefix
) -
micro_benchmark_n4
: 3% slowdown (n4_full_scan/512
) to 4% speedup (full_n4_sequential_delete/65534
) -
micro_benchmark_n16
: 5% slowdown (grow_n4_to_n16_sequentially/20
) to 0% speedup (minimal_n16_tree_random_gets/64
) -
micro_benchmark_n48
: 4% slowdown (minimal_n48_tree_full_scan/8
) to 9% speedup (grow_n16_to_n48_randomly/8`) -
micro_benchmark_n256
: 5% slowdown (grow_n48_to_n256_sequentially/64
) to 2% speedup (full_n256_tree_random_delete/192
)