From 82a49c88f917c73b56780398fee65665a6e2a25e Mon Sep 17 00:00:00 2001 From: "W. Felix Handte" Date: Mon, 13 Dec 2021 15:46:41 -0500 Subject: [PATCH] Increment Step by 1 not 2 I couldn't find a good way to spread `ip0` and `ip1` apart when we accelerate due to incompressible inputs. (The methods I tried slowed things down quite a bit.) Since we aren't splaying ip0 and ip1 apart (which would be like `0_1_2_3_`, as opposed to the `01__23__` we were actually doing), it's a big ambitious to increment `step` by 2. Instead, let's increment it by 1, which has the benefit sliiightly improving compression. Speed remains pretty much unchanged. --- lib/compress/zstd_fast.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/lib/compress/zstd_fast.c b/lib/compress/zstd_fast.c index 852f4286f..2bae2f70e 100644 --- a/lib/compress/zstd_fast.c +++ b/lib/compress/zstd_fast.c @@ -234,19 +234,19 @@ _start: /* Requires: ip0 */ hash0 = hash1; hash1 = ZSTD_hashPtr(ip2, hlog, mls); + /* advance to next positions */ + ip0 = ip1; + ip1 = ip2; + ip2 = ip0 + step; + ip3 = ip1 + step; + /* calculate step */ if (ip2 >= nextStep) { + step++; PREFETCH_L1(ip1 + 64); PREFETCH_L1(ip1 + 128); - step += 2; nextStep += kStepIncr; } - - /* advance to next positions */ - ip0 = ip1; - ip1 = ip2; - ip2 = ip0 + step; - ip3 = ip1 + step; } while (ip3 < ilimit); _cleanup: -- 2.47.3