Rusty Russell
|
1764d6c907
|
grind_htlc_tx_fee: benchmark.
Takes 15 seconds on my laptop to do the worst-case grind:
$ onchaind/test/run-grind_feerate 250001
250001 iterations in 15893 msec = 63574 nsec each
It's not worth optimizing as it's 75% in libsecp:
29.65% run-grind_feera run-grind_feerate [.] secp256k1_fe_mul_inner
23.51% run-grind_feera run-grind_feerate [.] secp256k1_fe_sqr_inner
11.04% run-grind_feera run-grind_feerate [.] secp256k1_gej_double_var.part.6.constprop.34
9.56% run-grind_feera run-grind_feerate [.] secp256k1_scalar_reduce_512
5.70% run-grind_feera run-grind_feerate [.] Round
Even forcing a compile with -O3 -flto, it's only 13883 msec = 55534 nsec each.
Fixes: #291
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
|
2018-04-04 02:31:41 +00:00 |
|