This is an automated email from the ASF dual-hosted git repository.
yiguolei pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/doris.git
The following commit(s) were added to refs/heads/master by this push:
new 92e1f687e40 [chore](columns) remove update_hashes_with_value for
SipHash (#31224)
92e1f687e40 is described below
commit 92e1f687e40d64775cb13565d976f10961d3e4bf
Author: Uniqueyou <[email protected]>
AuthorDate: Wed Feb 21 23:56:50 2024 +0800
[chore](columns) remove update_hashes_with_value for SipHash (#31224)
---
be/src/vec/columns/column.h | 22 ------------------
be/src/vec/columns/column_array.cpp | 5 ----
be/src/vec/columns/column_array.h | 3 ---
be/src/vec/columns/column_complex.h | 11 ++-------
be/src/vec/columns/column_const.cpp | 16 -------------
be/src/vec/columns/column_const.h | 3 ---
be/src/vec/columns/column_decimal.cpp | 6 -----
be/src/vec/columns/column_decimal.h | 2 --
be/src/vec/columns/column_map.cpp | 5 ----
be/src/vec/columns/column_map.h | 3 ---
be/src/vec/columns/column_nullable.cpp | 19 ---------------
be/src/vec/columns/column_nullable.h | 2 --
be/src/vec/columns/column_string.h | 5 ----
be/src/vec/columns/column_struct.cpp | 5 ----
be/src/vec/columns/column_struct.h | 3 ---
be/src/vec/columns/column_vector.cpp | 6 -----
be/src/vec/columns/column_vector.h | 3 ---
be/test/vec/columns/column_hash_func_test.cpp | 33 ---------------------------
18 files changed, 2 insertions(+), 150 deletions(-)
diff --git a/be/src/vec/columns/column.h b/be/src/vec/columns/column.h
index 322456a8f77..bf869961544 100644
--- a/be/src/vec/columns/column.h
+++ b/be/src/vec/columns/column.h
@@ -46,19 +46,6 @@
class SipHash;
-#define SIP_HASHES_FUNCTION_COLUMN_IMPL() \
- auto s = hashes.size(); \
- DCHECK(s == size()); \
- if (null_data == nullptr) { \
- for (size_t i = 0; i < s; i++) { \
- update_hash_with_value(i, hashes[i]); \
- } \
- } else { \
- for (size_t i = 0; i < s; i++) { \
- if (null_data[i] == 0) update_hash_with_value(i, hashes[i]); \
- } \
- }
-
#define DO_CRC_HASHES_FUNCTION_COLUMN_IMPL()
\
if (null_data == nullptr) {
\
for (size_t i = 0; i < s; i++) {
\
@@ -362,15 +349,6 @@ public:
LOG(FATAL) << get_name() << " update_hash_with_value siphash not
supported";
}
- /// Update state of hash function with value of n elements to avoid the
virtual function call
- /// null_data to mark whether need to do hash compute, null_data == nullptr
- /// means all element need to do hash function, else only *null_data != 0
need to do hash func
- /// do xxHash here, faster than other hash method
- virtual void update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict null_data
= nullptr) const {
- LOG(FATAL) << get_name() << " update_hashes_with_value siphash not
supported";
- }
-
/// Update state of hash function with value of n elements to avoid the
virtual function call
/// null_data to mark whether need to do hash compute, null_data == nullptr
/// means all element need to do hash function, else only *null_data != 0
need to do hash func
diff --git a/be/src/vec/columns/column_array.cpp
b/be/src/vec/columns/column_array.cpp
index 6d2914d8053..9abf13b9f3f 100644
--- a/be/src/vec/columns/column_array.cpp
+++ b/be/src/vec/columns/column_array.cpp
@@ -291,11 +291,6 @@ void ColumnArray::update_hash_with_value(size_t n,
SipHash& hash) const {
for (size_t i = 0; i < array_size; ++i)
get_data().update_hash_with_value(offset + i, hash);
}
-void ColumnArray::update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict
null_data) const {
- SIP_HASHES_FUNCTION_COLUMN_IMPL();
-}
-
// for every array row calculate xxHash
void ColumnArray::update_xxHash_with_value(size_t start, size_t end, uint64_t&
hash,
const uint8_t* __restrict
null_data) const {
diff --git a/be/src/vec/columns/column_array.h
b/be/src/vec/columns/column_array.h
index 046bc22ac48..17408bfa633 100644
--- a/be/src/vec/columns/column_array.h
+++ b/be/src/vec/columns/column_array.h
@@ -142,9 +142,6 @@ public:
void update_crc_with_value(size_t start, size_t end, uint32_t& hash,
const uint8_t* __restrict null_data) const
override;
- void update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict null_data) const
override;
-
void update_hashes_with_value(uint64_t* __restrict hashes,
const uint8_t* __restrict null_data =
nullptr) const override;
diff --git a/be/src/vec/columns/column_complex.h
b/be/src/vec/columns/column_complex.h
index 8e9686cb2a5..70d32da0b9b 100644
--- a/be/src/vec/columns/column_complex.h
+++ b/be/src/vec/columns/column_complex.h
@@ -220,15 +220,8 @@ public:
// TODO add hash function
}
- virtual void update_hashes_with_value(
- std::vector<SipHash>& hashes,
- const uint8_t* __restrict null_data = nullptr) const override {
- // TODO add hash function
- }
-
- virtual void update_hashes_with_value(
- uint64_t* __restrict hashes,
- const uint8_t* __restrict null_data = nullptr) const override {
+ void update_hashes_with_value(uint64_t* __restrict hashes,
+ const uint8_t* __restrict null_data =
nullptr) const override {
// TODO add hash function
}
diff --git a/be/src/vec/columns/column_const.cpp
b/be/src/vec/columns/column_const.cpp
index f7efec1f72a..e06e53b4289 100644
--- a/be/src/vec/columns/column_const.cpp
+++ b/be/src/vec/columns/column_const.cpp
@@ -93,22 +93,6 @@ ColumnPtr ColumnConst::permute(const Permutation& perm,
size_t limit) const {
return ColumnConst::create(data, limit);
}
-void ColumnConst::update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict
null_data) const {
- DCHECK(null_data == nullptr);
- DCHECK(hashes.size() == size());
- auto real_data = data->get_data_at(0);
- if (real_data.data == nullptr) {
- for (auto& hash : hashes) {
- hash.update(0);
- }
- } else {
- for (auto& hash : hashes) {
- hash.update(real_data.data, real_data.size);
- }
- }
-}
-
void ColumnConst::update_crcs_with_value(uint32_t* __restrict hashes,
doris::PrimitiveType type,
uint32_t rows, uint32_t offset,
const uint8_t* __restrict null_data)
const {
diff --git a/be/src/vec/columns/column_const.h
b/be/src/vec/columns/column_const.h
index 5498fbf7c20..746cb00fd5d 100644
--- a/be/src/vec/columns/column_const.h
+++ b/be/src/vec/columns/column_const.h
@@ -177,9 +177,6 @@ public:
data->update_hash_with_value(0, hash);
}
- void update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict null_data) const
override;
-
// (TODO.Amory) here may not use column_const update hash, and
PrimitiveType is not used.
void update_crcs_with_value(uint32_t* __restrict hashes, PrimitiveType
type, uint32_t rows,
uint32_t offset = 0,
diff --git a/be/src/vec/columns/column_decimal.cpp
b/be/src/vec/columns/column_decimal.cpp
index 95b247fc668..be81a7ad4cb 100644
--- a/be/src/vec/columns/column_decimal.cpp
+++ b/be/src/vec/columns/column_decimal.cpp
@@ -131,12 +131,6 @@ void ColumnDecimal<T>::update_hash_with_value(size_t n,
SipHash& hash) const {
hash.update(data[n]);
}
-template <typename T>
-void ColumnDecimal<T>::update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict
null_data) const {
- SIP_HASHES_FUNCTION_COLUMN_IMPL();
-}
-
template <typename T>
void ColumnDecimal<T>::update_crc_with_value(size_t start, size_t end,
uint32_t& hash,
const uint8_t* __restrict
null_data) const {
diff --git a/be/src/vec/columns/column_decimal.h
b/be/src/vec/columns/column_decimal.h
index 49b58ebaa4f..920f6ad8438 100644
--- a/be/src/vec/columns/column_decimal.h
+++ b/be/src/vec/columns/column_decimal.h
@@ -176,8 +176,6 @@ public:
const uint8_t* null_map) override;
void update_hash_with_value(size_t n, SipHash& hash) const override;
- void update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict null_data) const
override;
void update_hashes_with_value(uint64_t* __restrict hashes,
const uint8_t* __restrict null_data) const
override;
void update_crcs_with_value(uint32_t* __restrict hashes, PrimitiveType
type, uint32_t rows,
diff --git a/be/src/vec/columns/column_map.cpp
b/be/src/vec/columns/column_map.cpp
index dcfcdb46d70..4a1df8a6e26 100644
--- a/be/src/vec/columns/column_map.cpp
+++ b/be/src/vec/columns/column_map.cpp
@@ -271,11 +271,6 @@ void ColumnMap::update_hash_with_value(size_t n, SipHash&
hash) const {
}
}
-void ColumnMap::update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict null_data)
const {
- SIP_HASHES_FUNCTION_COLUMN_IMPL();
-}
-
void ColumnMap::update_xxHash_with_value(size_t start, size_t end, uint64_t&
hash,
const uint8_t* __restrict null_data)
const {
auto& offsets = get_offsets();
diff --git a/be/src/vec/columns/column_map.h b/be/src/vec/columns/column_map.h
index 206660d6a06..0e6ad3c3d91 100644
--- a/be/src/vec/columns/column_map.h
+++ b/be/src/vec/columns/column_map.h
@@ -182,9 +182,6 @@ public:
void update_crc_with_value(size_t start, size_t end, uint32_t& hash,
const uint8_t* __restrict null_data) const
override;
- void update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict null_data) const
override;
-
void update_hashes_with_value(uint64_t* __restrict hashes,
const uint8_t* __restrict null_data =
nullptr) const override;
diff --git a/be/src/vec/columns/column_nullable.cpp
b/be/src/vec/columns/column_nullable.cpp
index 0cd671eb110..e20b87af826 100644
--- a/be/src/vec/columns/column_nullable.cpp
+++ b/be/src/vec/columns/column_nullable.cpp
@@ -93,25 +93,6 @@ void ColumnNullable::update_hash_with_value(size_t n,
SipHash& hash) const {
}
}
-void ColumnNullable::update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict
null_data) const {
- DCHECK(null_data == nullptr);
- auto s = hashes.size();
- DCHECK(s == size());
- const auto* __restrict real_null_data =
- assert_cast<const ColumnUInt8&>(*null_map).get_data().data();
- if (!has_null()) {
- nested_column->update_hashes_with_value(hashes, nullptr);
- } else {
- for (int i = 0; i < s; ++i) {
- if (real_null_data[i] != 0) {
- hashes[i].update(0);
- }
- }
- nested_column->update_hashes_with_value(hashes, real_null_data);
- }
-}
-
void ColumnNullable::update_crcs_with_value(uint32_t* __restrict hashes,
doris::PrimitiveType type,
uint32_t rows, uint32_t offset,
const uint8_t* __restrict
null_data) const {
diff --git a/be/src/vec/columns/column_nullable.h
b/be/src/vec/columns/column_nullable.h
index eca4c57fceb..de01907650e 100644
--- a/be/src/vec/columns/column_nullable.h
+++ b/be/src/vec/columns/column_nullable.h
@@ -219,8 +219,6 @@ public:
const uint8_t* __restrict null_data) const
override;
void update_hash_with_value(size_t n, SipHash& hash) const override;
- void update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict null_data) const
override;
void update_crcs_with_value(uint32_t* __restrict hash, PrimitiveType type,
uint32_t rows,
uint32_t offset,
const uint8_t* __restrict null_data) const
override;
diff --git a/be/src/vec/columns/column_string.h
b/be/src/vec/columns/column_string.h
index 5c50f5ed2f1..9bf43c9c627 100644
--- a/be/src/vec/columns/column_string.h
+++ b/be/src/vec/columns/column_string.h
@@ -451,11 +451,6 @@ public:
hash.update(reinterpret_cast<const char*>(&chars[offset]),
string_size);
}
- void update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict null_data) const
override {
- SIP_HASHES_FUNCTION_COLUMN_IMPL();
- }
-
void update_crcs_with_value(uint32_t* __restrict hashes, PrimitiveType
type, uint32_t rows,
uint32_t offset,
const uint8_t* __restrict null_data) const
override;
diff --git a/be/src/vec/columns/column_struct.cpp
b/be/src/vec/columns/column_struct.cpp
index f6ab9c9604f..290452a1c8b 100644
--- a/be/src/vec/columns/column_struct.cpp
+++ b/be/src/vec/columns/column_struct.cpp
@@ -206,11 +206,6 @@ void ColumnStruct::update_hash_with_value(size_t n,
SipHash& hash) const {
}
}
-void ColumnStruct::update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict
null_data) const {
- SIP_HASHES_FUNCTION_COLUMN_IMPL();
-}
-
void ColumnStruct::update_xxHash_with_value(size_t start, size_t end,
uint64_t& hash,
const uint8_t* __restrict
null_data) const {
for (const auto& column : columns) {
diff --git a/be/src/vec/columns/column_struct.h
b/be/src/vec/columns/column_struct.h
index 1b1daee4452..d91d3800648 100644
--- a/be/src/vec/columns/column_struct.h
+++ b/be/src/vec/columns/column_struct.h
@@ -113,9 +113,6 @@ public:
void update_crc_with_value(size_t start, size_t end, uint32_t& hash,
const uint8_t* __restrict null_data) const
override;
- void update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict null_data) const
override;
-
void update_hashes_with_value(uint64_t* __restrict hashes,
const uint8_t* __restrict null_data =
nullptr) const override;
diff --git a/be/src/vec/columns/column_vector.cpp
b/be/src/vec/columns/column_vector.cpp
index 71f55af0a79..acc6fc1d7f3 100644
--- a/be/src/vec/columns/column_vector.cpp
+++ b/be/src/vec/columns/column_vector.cpp
@@ -110,12 +110,6 @@ void ColumnVector<T>::update_hash_with_value(size_t n,
SipHash& hash) const {
hash.update(data[n]);
}
-template <typename T>
-void ColumnVector<T>::update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict
null_data) const {
- SIP_HASHES_FUNCTION_COLUMN_IMPL();
-}
-
template <typename T>
void ColumnVector<T>::update_hashes_with_value(uint64_t* __restrict hashes,
const uint8_t* __restrict
null_data) const {
diff --git a/be/src/vec/columns/column_vector.h
b/be/src/vec/columns/column_vector.h
index dbc7524eaac..ff9197df357 100644
--- a/be/src/vec/columns/column_vector.h
+++ b/be/src/vec/columns/column_vector.h
@@ -321,9 +321,6 @@ public:
}
void update_hash_with_value(size_t n, SipHash& hash) const override;
- void update_hashes_with_value(std::vector<SipHash>& hashes,
- const uint8_t* __restrict null_data) const
override;
-
void update_crcs_with_value(uint32_t* __restrict hashes, PrimitiveType
type, uint32_t rows,
uint32_t offset,
const uint8_t* __restrict null_data) const
override;
diff --git a/be/test/vec/columns/column_hash_func_test.cpp
b/be/test/vec/columns/column_hash_func_test.cpp
index a95c7071af5..0a9471a1ab3 100644
--- a/be/test/vec/columns/column_hash_func_test.cpp
+++ b/be/test/vec/columns/column_hash_func_test.cpp
@@ -62,22 +62,14 @@ DataTypes create_scala_data_types() {
TEST(HashFuncTest, ArrayTypeTest) {
DataTypes dataTypes = create_scala_data_types();
- std::vector<uint64_t> sip_hash_vals(1);
std::vector<uint64_t> xx_hash_vals(1);
std::vector<uint32_t> crc_hash_vals(1);
- auto* __restrict sip_hashes = sip_hash_vals.data();
auto* __restrict xx_hashes = xx_hash_vals.data();
auto* __restrict crc_hashes = crc_hash_vals.data();
for (auto d : dataTypes) {
DataTypePtr a = std::make_shared<DataTypeArray>(d);
ColumnPtr col_a = a->create_column_const_with_default_value(1);
- // sipHash
- std::vector<SipHash> siphashs(1);
- col_a->update_hashes_with_value(siphashs);
- EXPECT_NO_FATAL_FAILURE(col_a->update_hashes_with_value(siphashs));
- sip_hashes[0] = siphashs[0].get64();
- std::cout << sip_hashes[0] << std::endl;
// xxHash
EXPECT_NO_FATAL_FAILURE(col_a->update_hashes_with_value(xx_hashes));
std::cout << xx_hashes[0] << std::endl;
@@ -184,22 +176,11 @@ TEST(HashFuncTest, ArrayCornerCaseTest) {
EXPECT_EQ(array_mutable_col->size(), 3);
- std::vector<uint64_t> sip_hash_vals(3);
std::vector<uint64_t> xx_hash_vals(3);
std::vector<uint32_t> crc_hash_vals(3);
- auto* __restrict sip_hashes = sip_hash_vals.data();
auto* __restrict xx_hashes = xx_hash_vals.data();
auto* __restrict crc_hashes = crc_hash_vals.data();
- // sipHash
- std::vector<SipHash> siphashs(3);
- array_mutable_col->update_hashes_with_value(siphashs);
-
EXPECT_NO_FATAL_FAILURE(array_mutable_col->update_hashes_with_value(siphashs));
- sip_hashes[0] = siphashs[0].get64();
- sip_hashes[1] = siphashs[1].get64();
- sip_hashes[2] = siphashs[2].get64();
- EXPECT_EQ(sip_hashes[0], sip_hash_vals[1]);
- EXPECT_TRUE(sip_hash_vals[0] != sip_hash_vals[2]);
// xxHash
EXPECT_NO_FATAL_FAILURE(array_mutable_col->update_hashes_with_value(xx_hashes));
EXPECT_EQ(xx_hashes[0], xx_hashes[1]);
@@ -214,21 +195,14 @@ TEST(HashFuncTest, ArrayCornerCaseTest) {
TEST(HashFuncTest, MapTypeTest) {
DataTypes dataTypes = create_scala_data_types();
- std::vector<uint64_t> sip_hash_vals(1);
std::vector<uint64_t> xx_hash_vals(1);
std::vector<uint32_t> crc_hash_vals(1);
- auto* __restrict sip_hashes = sip_hash_vals.data();
auto* __restrict xx_hashes = xx_hash_vals.data();
auto* __restrict crc_hashes = crc_hash_vals.data();
// data_type_map
for (int i = 0; i < dataTypes.size() - 1; ++i) {
DataTypePtr a = std::make_shared<DataTypeMap>(dataTypes[i],
dataTypes[i + 1]);
ColumnPtr col_a = a->create_column_const_with_default_value(1);
- // sipHash
- std::vector<SipHash> siphashs(1);
-
EXPECT_NO_FATAL_FAILURE(unpack_if_const(col_a).first->update_hashes_with_value(siphashs));
- sip_hashes[0] = siphashs[0].get64();
- std::cout << sip_hashes[0] << std::endl;
// xxHash
EXPECT_NO_FATAL_FAILURE(unpack_if_const(col_a).first->update_hashes_with_value(xx_hashes));
std::cout << xx_hashes[0] << std::endl;
@@ -242,21 +216,14 @@ TEST(HashFuncTest, MapTypeTest) {
TEST(HashFuncTest, StructTypeTest) {
DataTypes dataTypes = create_scala_data_types();
- std::vector<uint64_t> sip_hash_vals(1);
std::vector<uint64_t> xx_hash_vals(1);
std::vector<uint32_t> crc_hash_vals(1);
- auto* __restrict sip_hashes = sip_hash_vals.data();
auto* __restrict xx_hashes = xx_hash_vals.data();
auto* __restrict crc_hashes = crc_hash_vals.data();
// data_type_struct
DataTypePtr a = std::make_shared<DataTypeStruct>(dataTypes);
ColumnPtr col_a = a->create_column_const_with_default_value(1);
- // sipHash
- std::vector<SipHash> siphashs(1);
-
EXPECT_NO_FATAL_FAILURE(unpack_if_const(col_a).first->update_hashes_with_value(siphashs));
- sip_hashes[0] = siphashs[0].get64();
- std::cout << sip_hashes[0] << std::endl;
// xxHash
EXPECT_NO_FATAL_FAILURE(unpack_if_const(col_a).first->update_hashes_with_value(xx_hashes));
std::cout << xx_hashes[0] << std::endl;
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]