Skip to content

zstd enc: Remove unneeded masks #145

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Aug 3, 2019
Merged

zstd enc: Remove unneeded masks #145

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 12 additions & 24 deletions zstd/enc_dfast.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,16 +82,11 @@ func (e *doubleFastEncoder) Encode(blk *blockEnc, src []byte) {
stepSize++
}

// TEMPLATE

const kSearchStrength = 8

// nextEmit is where in src the next emitLiteral should start from.
nextEmit := s
cv := load6432(src, s)
// nextHash is the hash at s
nextHashS := hash5(cv, dFastShortTableBits)
nextHashL := hash8(cv, dFastLongTableBits)

// Relative offsets
offset1 := int32(blk.recentOffsets[0])
Expand Down Expand Up @@ -119,8 +114,8 @@ encodeLoop:
panic("offset0 was 0")
}

nextHashS = nextHashS & dFastShortTableMask
nextHashL = nextHashL & dFastLongTableMask
nextHashS := hash5(cv, dFastShortTableBits)
nextHashL := hash8(cv, dFastLongTableBits)
candidateL := e.longTable[nextHashL]
candidateS := e.table[nextHashS]

Expand Down Expand Up @@ -172,8 +167,6 @@ encodeLoop:
break encodeLoop
}
cv = load6432(src, s)
nextHashS = hash5(cv, dFastShortTableBits)
nextHashL = hash8(cv, dFastLongTableBits)
continue
}
const repOff2 = 1
Expand Down Expand Up @@ -221,8 +214,6 @@ encodeLoop:
break encodeLoop
}
cv = load6432(src, s)
nextHashS = hash5(cv, dFastShortTableBits)
nextHashL = hash8(cv, dFastLongTableBits)
// Swap offsets
offset1, offset2 = offset2, offset1
continue
Expand Down Expand Up @@ -296,8 +287,6 @@ encodeLoop:
break encodeLoop
}
cv = load6432(src, s)
nextHashS = hash5(cv, dFastShortTableBits)
nextHashL = hash8(cv, dFastLongTableBits)
}

// A 4-byte match has been found. Update recent offsets.
Expand Down Expand Up @@ -354,20 +343,18 @@ encodeLoop:
cv1 := load6432(src, index1)
te0 := tableEntry{offset: index0 + e.cur, val: uint32(cv0)}
te1 := tableEntry{offset: index1 + e.cur, val: uint32(cv1)}
e.longTable[hash8(cv0, dFastLongTableBits)&dFastLongTableMask] = te0
e.longTable[hash8(cv1, dFastLongTableBits)&dFastLongTableMask] = te1
e.longTable[hash8(cv0, dFastLongTableBits)] = te0
e.longTable[hash8(cv1, dFastLongTableBits)] = te1
cv0 >>= 8
cv1 >>= 8
te0.offset++
te1.offset++
te0.val = uint32(cv0)
te1.val = uint32(cv1)
e.table[hash5(cv0, dFastShortTableBits)&dFastShortTableMask] = te0
e.table[hash5(cv1, dFastShortTableBits)&dFastShortTableMask] = te1
e.table[hash5(cv0, dFastShortTableBits)] = te0
e.table[hash5(cv1, dFastShortTableBits)] = te1

cv = load6432(src, s)
nextHashS = hash5(cv1>>8, dFastShortTableBits)
nextHashL = hash8(cv, dFastLongTableBits)

if !canRepeat {
continue
Expand All @@ -381,14 +368,17 @@ encodeLoop:
break
}

// Store this, since we have it.
nextHashS := hash5(cv1>>8, dFastShortTableBits)
nextHashL := hash8(cv, dFastLongTableBits)

// We have at least 4 byte match.
// No need to check backwards. We come straight from a match
l := 4 + e.matchlen(s+4, o2+4, src)

// Store this, since we have it.
entry := tableEntry{offset: s + e.cur, val: uint32(cv)}
e.longTable[nextHashL&dFastLongTableMask] = entry
e.table[nextHashS&dFastShortTableMask] = entry
e.longTable[nextHashL] = entry
e.table[nextHashS] = entry
seq.matchLen = uint32(l) - zstdMinMatch
seq.litLen = 0

Expand All @@ -408,8 +398,6 @@ encodeLoop:
break encodeLoop
}
cv = load6432(src, s)
nextHashS = hash5(cv, dFastShortTableBits)
nextHashL = hash8(cv, dFastLongTableBits)
}
}

Expand Down
17 changes: 6 additions & 11 deletions zstd/enc_fast.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -124,8 +124,6 @@ func (e *fastEncoder) Encode(blk *blockEnc, src []byte) {
// nextEmit is where in src the next emitLiteral should start from.
nextEmit := s
cv := load6432(src, s)
// nextHash is the hash at s
nextHash := hash6(cv, hashLog)

// Relative offsets
offset1 := int32(blk.recentOffsets[0])
Expand Down Expand Up @@ -157,8 +155,8 @@ encodeLoop:
panic("offset0 was 0")
}

nextHash2 := hash6(cv>>8, hashLog) & tableMask
nextHash = nextHash & tableMask
nextHash := hash6(cv, hashLog)
nextHash2 := hash6(cv>>8, hashLog)
candidate := e.table[nextHash]
candidate2 := e.table[nextHash2]
repIndex := s - offset1 + 2
Expand Down Expand Up @@ -207,8 +205,6 @@ encodeLoop:
break encodeLoop
}
cv = load6432(src, s)
//nextHash = hashLen(cv, hashLog, mls)
nextHash = hash6(cv, hashLog)
continue
}
coffset0 := s - (candidate.offset - e.cur)
Expand Down Expand Up @@ -245,7 +241,6 @@ encodeLoop:
break encodeLoop
}
cv = load6432(src, s)
nextHash = hash6(cv, hashLog)
}
// A 4-byte match has been found. We'll later see if more than 4 bytes.
offset2 = offset1
Expand Down Expand Up @@ -292,15 +287,16 @@ encodeLoop:
break encodeLoop
}
cv = load6432(src, s)
nextHash = hash6(cv, hashLog)

// Check offset 2
if o2 := s - offset2; canRepeat && o2 > 0 && load3232(src, o2) == uint32(cv) {
if o2 := s - offset2; canRepeat && load3232(src, o2) == uint32(cv) {
// We have at least 4 byte match.
// No need to check backwards. We come straight from a match
l := 4 + e.matchlen(s+4, o2+4, src)

// Store this, since we have it.
e.table[nextHash&tableMask] = tableEntry{offset: s + e.cur, val: uint32(cv)}
nextHash := hash6(cv, hashLog)
e.table[nextHash] = tableEntry{offset: s + e.cur, val: uint32(cv)}
seq.matchLen = uint32(l) - zstdMinMatch
seq.litLen = 0
// Since litlen is always 0, this is offset 1.
Expand All @@ -319,7 +315,6 @@ encodeLoop:
}
// Prepare next loop.
cv = load6432(src, s)
nextHash = hash6(cv, hashLog)
}
}

Expand Down