src/backup/chunker.rs: another try to speedup the chunker

This commit is contained in:
Dietmar Maurer 2019-02-21 18:56:41 +01:00
parent 532a69eb04
commit 61834c1cbf
1 changed files with 16 additions and 7 deletions

View File

@ -1,7 +1,10 @@
/// Note: doö not use 32 or 64, because that always computes hash 0
/// Note: window size 32 or 64, is faster because we can
/// speedup modulo operations, but always computes hash 0
/// for constant data streams .. 0,0,0,0,0,0
const CA_CHUNKER_WINDOW_SIZE: usize = 48;
/// so we use a modified the chunk boundary test too not
/// use hash value 0 to detect a boundary.
const CA_CHUNKER_WINDOW_SIZE: usize = 64;
/// Slinding window chunker (Buzhash)
///
@ -164,14 +167,16 @@ impl Chunker {
}
}
let mut idx = self.chunk_size % CA_CHUNKER_WINDOW_SIZE;
//let mut idx = self.chunk_size % CA_CHUNKER_WINDOW_SIZE;
let mut idx = self.chunk_size & 0x3f;
while pos < data_len {
// roll window
let enter = data[pos];
let leave = self.window[idx];
self.h = self.h.rotate_left(1) ^
BUZHASH_TABLE[leave as usize].rotate_left(CA_CHUNKER_WINDOW_SIZE as u32) ^
//BUZHASH_TABLE[leave as usize].rotate_left(CA_CHUNKER_WINDOW_SIZE as u32) ^
BUZHASH_TABLE[leave as usize] ^
BUZHASH_TABLE[enter as usize];
self.chunk_size += 1;
@ -186,7 +191,8 @@ impl Chunker {
return pos;
}
idx = self.chunk_size % CA_CHUNKER_WINDOW_SIZE;
//idx = self.chunk_size % CA_CHUNKER_WINDOW_SIZE;
idx = self.chunk_size & 0x3f;
//idx += 1; if idx >= CA_CHUNKER_WINDOW_SIZE { idx = 0 };
}
@ -201,9 +207,12 @@ impl Chunker {
if self.chunk_size < self.chunk_size_min { return false; }
(self.h & self.break_test_value) <= 2
//(self.h & 0x1ffff) <= 2 //THIS IS SLOW!!!
//(self.h & self.break_test_value) <= 2 // Bad on 0 streams
// simply add arbitrary value (10) before testing
((self.h+10) & self.break_test_value) <= 2
}
// This is the original implementation from casync