Add optimized support for array<byte>

11 files changed, 105 insertions, 39 deletions

diff --git a/searchlib/src/tests/features/prod_features.cpp b/searchlib/src/tests/features/prod_features.cpp
index 51a98bec3d9..626a470cb5c 100644
--- a/searchlib/src/tests/features/prod_features.cpp
+++ b/searchlib/src/tests/features/prod_features.cpp
@@ -1,6 +1,4 @@
 // Copyright 2017 Yahoo Holdings. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
-#include <vespa/log/log.h>
-LOG_SETUP("prod_features_test");
 
 #include "prod_features.h"
 #include <vespa/searchlib/attribute/attributeguard.h>
@@ -50,6 +48,9 @@ LOG_SETUP("prod_features_test");
 #include <vespa/vespalib/util/string_hash.h>
 #include <cmath>
 
+#include <vespa/log/log.h>
+LOG_SETUP("prod_features_test");
+
 using namespace search::features;
 using namespace search::fef;
 using namespace search::fef::test;
@@ -1065,6 +1066,25 @@ verifyCorrectDotProductExecutor(BlueprintFactory & factory, vespalib::stringref
     EXPECT_EQUAL(1u, deps.output.size());
 }
 
+template<typename T>
+void verifyArrayParser()
+{
+    std::vector<vespalib::string> v = {"(0:2,7:-3,1:-3)", "{0:2,7:-3,1:-3}", "[2 -3 0 0 0 0 0 -3]"};
+    for(const vespalib::string & s : v) {
+        std::vector<T> out;
+        ArrayParser::parse(s, out);
+        EXPECT_EQUAL(8u, out.size());
+        EXPECT_EQUAL(2,  out[0]);
+        EXPECT_EQUAL(-3, out[1]);
+        EXPECT_EQUAL(0,  out[2]);
+        EXPECT_EQUAL(0,  out[3]);
+        EXPECT_EQUAL(0,  out[4]);
+        EXPECT_EQUAL(0,  out[5]);
+        EXPECT_EQUAL(0,  out[6]);
+        EXPECT_EQUAL(-3, out[7]);
+    }
+}
+
 }
 
 void
@@ -1159,22 +1179,12 @@ Test::testDotProduct()
             EXPECT_EQUAL(out.getVector()[1].second, -3.5);
         }
     }
-    { // Array parser
-        std::vector<vespalib::string> v = {"(0:2,7:-3,1:-3)", "{0:2,7:-3,1:-3}", "[2 -3 0 0 0 0 0 -3]"};
-        for(const vespalib::string & s : v) {
-            std::vector<int32_t> out;
-            ArrayParser::parse(s, out);
-            EXPECT_EQUAL(8u, out.size());
-            EXPECT_EQUAL(2,  out[0]);
-            EXPECT_EQUAL(-3, out[1]);
-            EXPECT_EQUAL(0,  out[2]);
-            EXPECT_EQUAL(0,  out[3]);
-            EXPECT_EQUAL(0,  out[4]);
-            EXPECT_EQUAL(0,  out[5]);
-            EXPECT_EQUAL(0,  out[6]);
-            EXPECT_EQUAL(-3, out[7]);
-        }
-    }
+    verifyArrayParser<int8_t>();
+    verifyArrayParser<int16_t>();
+    verifyArrayParser<int32_t>();
+    verifyArrayParser<int64_t>();
+    verifyArrayParser<float>();
+    verifyArrayParser<double>();
     {
         vespalib::string s = "[[1:3]]";
         std::vector<int32_t> out;
@@ -1202,13 +1212,12 @@ Test::testDotProduct()
             assertDotProduct(0,   "(f:5,g:5)",             1, "wsextstr");
             assertDotProduct(550, "(a:1,b:2,c:3,d:4,e:5)", 1, "wsextstr");
         }
-        { // integer attribute
-            assertDotProduct(0,  "()",                    1, "wsint");
-            assertDotProduct(0,  "(6:5,7:5)",             1, "wsint");
-            assertDotProduct(55, "(1:1,2:2,3:3,4:4,5:5)", 1, "wsint");
+        for (const char * name : {"wsbyte", "wsint"}) {
+            assertDotProduct(0,  "()",                    1, name);
+            assertDotProduct(0,  "(6:5,7:5)",             1, name);
+            assertDotProduct(55, "(1:1,2:2,3:3,4:4,5:5)", 1, name);
         }
-        std::vector<const char *> attributes = {"arrint", "arrfloat", "arrint_fast", "arrfloat_fast"};
-        for (const char * name : attributes) {
+        for (const char * name : {"arrbyte", "arrint", "arrfloat", "arrint_fast", "arrfloat_fast"}) {
             assertDotProduct(0,  "()",                    1, name);
             assertDotProduct(0,  "(6:5,7:5)",             1, name);
             assertDotProduct(55,  "(0:1,1:2,2:3,3:4,4:5)", 1, name);
@@ -1260,6 +1269,8 @@ Test::setupForDotProductTest(FtFeatureTest & ft)
         bool fastSearch;
     };
     std::vector<Config> cfgList = { {"wsint", AVBT::INT32, AVCT::WSET, false},
+                                    {"wsbyte", AVBT::INT8, AVCT::WSET, false},
+                                    {"arrbyte", AVBT::INT8, AVCT::ARRAY, false},
                                     {"arrint", AVBT::INT32, AVCT::ARRAY, false},
                                     {"arrfloat", AVBT::FLOAT, AVCT::ARRAY, false},
                                     {"arrint_fast", AVBT::INT32, AVCT::ARRAY, true},
diff --git a/searchlib/src/vespa/searchlib/features/array_parser.cpp b/searchlib/src/vespa/searchlib/features/array_parser.cpp
index 7f6a5da7d4c..edfaed015e8 100644
--- a/searchlib/src/vespa/searchlib/features/array_parser.cpp
+++ b/searchlib/src/vespa/searchlib/features/array_parser.cpp
@@ -1,12 +1,17 @@
 // Copyright 2017 Yahoo Holdings. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
 
-#include "array_parser.h"
+#include "array_parser.hpp"
 
 #include <vespa/log/log.h>
 LOG_SETUP(".features.array_parser");
 
-namespace search {
-namespace features {
+namespace search::features {
+
+void
+ArrayParser::parse(const vespalib::string &input, std::vector<int8_t> &output)
+{
+    parse<std::vector<int8_t>, int16_t>(input, output);
+}
 
 void
 ArrayParser::logWarning(const vespalib::string &msg)
@@ -14,5 +19,4 @@ ArrayParser::logWarning(const vespalib::string &msg)
     LOG(warning, "%s", msg.c_str());
 }
 
-} // namespace features
-} // namespace search
+}
diff --git a/searchlib/src/vespa/searchlib/features/array_parser.h b/searchlib/src/vespa/searchlib/features/array_parser.h
index 53247eabe2c..8bf6e9ca365 100644
--- a/searchlib/src/vespa/searchlib/features/array_parser.h
+++ b/searchlib/src/vespa/searchlib/features/array_parser.h
@@ -3,6 +3,7 @@
 #pragma once
 
 #include <vespa/vespalib/stllike/string.h>
+#include <vector>
 
 namespace search::features {
 
@@ -37,9 +38,11 @@ public:
         uint32_t _index;
     };
 
-    template <typename OutputType>
+    template <typename OutputType, typename T = typename OutputType::value_type>
     static void parse(const vespalib::string &input, OutputType &output);
 
+    static void parse(const vespalib::string &input, std::vector<int8_t> &output);
+
     template <typename OutputType>
     static void parsePartial(const vespalib::string &input, OutputType &output);
 };
diff --git a/searchlib/src/vespa/searchlib/features/array_parser.hpp b/searchlib/src/vespa/searchlib/features/array_parser.hpp
index 92abec3aab9..089cb038797 100644
--- a/searchlib/src/vespa/searchlib/features/array_parser.hpp
+++ b/searchlib/src/vespa/searchlib/features/array_parser.hpp
@@ -11,11 +11,11 @@
 
 namespace search::features {
 
-template <typename OutputType>
+template <typename OutputType, typename T>
 void
 ArrayParser::parse(const vespalib::string &input, OutputType &output)
 {
-    typedef std::vector<ValueAndIndex<typename OutputType::value_type>> SparseVector;
+    typedef std::vector<ValueAndIndex<T>> SparseVector;
     SparseVector sparse;
     parsePartial(input, sparse);
     std::sort(sparse.begin(), sparse.end());
@@ -47,7 +47,7 @@ ArrayParser::parsePartial(const vespalib::string &input, OutputType &output)
                 try {
                     is >> key >> colon >> value;
                     if ((colon == ':') && is.eof()) {
-                        output.push_back(ValueAndIndexType(value, key));
+                        output.emplace_back(value, key);
                     } else {
                         logWarning(vespalib::make_string(
                                 "Could not parse item '%s' in query vector '%s', skipping. "
@@ -73,7 +73,7 @@ ArrayParser::parsePartial(const vespalib::string &input, OutputType &output)
             while (!is.eof()) {
                 try {
                     is >> value;
-                    output.push_back(ValueAndIndexType(value, index++));
+                    output.emplace_back(value, index++);
                 } catch (vespalib::IllegalArgumentException & e) {
                     logWarning(vespalib::make_string(
                             "Could not parse item[%ld] = '%s' in query vector '%s', skipping. "
diff --git a/searchlib/src/vespa/searchlib/features/dotproductfeature.cpp b/searchlib/src/vespa/searchlib/features/dotproductfeature.cpp
index 021320699ec..97216ef5437 100644
--- a/searchlib/src/vespa/searchlib/features/dotproductfeature.cpp
+++ b/searchlib/src/vespa/searchlib/features/dotproductfeature.cpp
@@ -398,11 +398,11 @@ DotProductBlueprint::createInstance() const
 
 namespace {
 
-template <typename T>
+template <typename T, typename AsT = T>
 void
 parseVectors(const Property& prop, std::vector<T>& values, std::vector<uint32_t>& indexes)
 {
-    typedef std::vector<ArrayParser::ValueAndIndex<T>> SparseV;
+    typedef std::vector<ArrayParser::ValueAndIndex<AsT>> SparseV;
     SparseV sparse;
     ArrayParser::parsePartial(prop.get(), sparse);
     if ( ! sparse.empty()) {
@@ -423,6 +423,12 @@ parseVectors(const Property& prop, std::vector<T>& values, std::vector<uint32_t>
     }
 }
 
+template <>
+void
+parseVectors<int8_t, int8_t>(const Property& prop, std::vector<int8_t>& values, std::vector<uint32_t>& indexes) {
+    parseVectors<int8_t, int16_t>(prop, values, indexes);
+}
+
 }
 
 namespace dotproduct {
diff --git a/searchlib/src/vespa/searchlib/features/utils.cpp b/searchlib/src/vespa/searchlib/features/utils.cpp
index 86558e89ec9..3f68e69ff25 100644
--- a/searchlib/src/vespa/searchlib/features/utils.cpp
+++ b/searchlib/src/vespa/searchlib/features/utils.cpp
@@ -4,10 +4,25 @@
 
 namespace search::features::util {
 
-template double strToNum<double>(vespalib::stringref str);
+template <>
+uint8_t
+strToNum<uint8_t>(vespalib::stringref str) {
+    return strToNum<uint16_t>(str);
+}
+
+template <>
+int8_t
+strToNum<int8_t>(vespalib::stringref str) {
+    return strToNum<int16_t>(str);
+}
+
+template double   strToNum<double>(vespalib::stringref str);
+template float    strToNum<float>(vespalib::stringref str);
+template uint16_t strToNum<uint16_t>(vespalib::stringref str);
 template uint32_t strToNum<uint32_t>(vespalib::stringref str);
 template uint64_t strToNum<uint64_t>(vespalib::stringref str);
-template int32_t strToNum<int32_t>(vespalib::stringref str);
-template int64_t strToNum<int64_t>(vespalib::stringref str);
+template int16_t  strToNum<int16_t>(vespalib::stringref str);
+template int32_t  strToNum<int32_t>(vespalib::stringref str);
+template int64_t  strToNum<int64_t>(vespalib::stringref str);
 
 }
diff --git a/vespalib/src/vespa/vespalib/hwaccelrated/generic.cpp b/vespalib/src/vespa/vespalib/hwaccelrated/generic.cpp
index 1bf7ea1c44c..d70071525c6 100644
--- a/vespalib/src/vespa/vespalib/hwaccelrated/generic.cpp
+++ b/vespalib/src/vespa/vespalib/hwaccelrated/generic.cpp
@@ -71,6 +71,17 @@ GenericAccelrator::dotProduct(const double * a, const double * b, size_t sz) con
 }
 
 int64_t
+GenericAccelrator::dotProduct(const int8_t * a, const int8_t * b, size_t sz) const
+{
+    return multiplyAdd<int64_t, int8_t, 4>(a, b, sz);
+}
+
+int64_t
+GenericAccelrator::dotProduct(const int16_t * a, const int16_t * b, size_t sz) const
+{
+    return multiplyAdd<int64_t, int16_t, 4>(a, b, sz);
+}
+int64_t
 GenericAccelrator::dotProduct(const int32_t * a, const int32_t * b, size_t sz) const
 {
     return multiplyAdd<int64_t, int32_t, 4>(a, b, sz);
diff --git a/vespalib/src/vespa/vespalib/hwaccelrated/generic.h b/vespalib/src/vespa/vespalib/hwaccelrated/generic.h
index 0115d5c55c7..f9aab3ae845 100644
--- a/vespalib/src/vespa/vespalib/hwaccelrated/generic.h
+++ b/vespalib/src/vespa/vespalib/hwaccelrated/generic.h
@@ -14,6 +14,8 @@ class GenericAccelrator : public IAccelrated
 public:
     float dotProduct(const float * a, const float * b, size_t sz) const override;
     double dotProduct(const double * a, const double * b, size_t sz) const override;
+    int64_t dotProduct(const int8_t * a, const int8_t * b, size_t sz) const override;
+    int64_t dotProduct(const int16_t * a, const int16_t * b, size_t sz) const override;
     int64_t dotProduct(const int32_t * a, const int32_t * b, size_t sz) const override;
     long long dotProduct(const int64_t * a, const int64_t * b, size_t sz) const override;
     void orBit(void * a, const void * b, size_t bytes) const override;
diff --git a/vespalib/src/vespa/vespalib/hwaccelrated/iaccelrated.h b/vespalib/src/vespa/vespalib/hwaccelrated/iaccelrated.h
index 7c2f2547240..aae60279d06 100644
--- a/vespalib/src/vespa/vespalib/hwaccelrated/iaccelrated.h
+++ b/vespalib/src/vespa/vespalib/hwaccelrated/iaccelrated.h
@@ -18,6 +18,8 @@ public:
     typedef std::unique_ptr<IAccelrated> UP;
     virtual float dotProduct(const float * a, const float * b, size_t sz) const = 0;
     virtual double dotProduct(const double * a, const double * b, size_t sz) const = 0;
+    virtual int64_t dotProduct(const int8_t * a, const int8_t * b, size_t sz) const = 0;
+    virtual int64_t dotProduct(const int16_t * a, const int16_t * b, size_t sz) const = 0;
     virtual int64_t dotProduct(const int32_t * a, const int32_t * b, size_t sz) const = 0;
     virtual long long dotProduct(const int64_t * a, const int64_t * b, size_t sz) const = 0;
     virtual void orBit(void * a, const void * b, size_t bytes) const = 0;
diff --git a/vespalib/src/vespa/vespalib/stllike/asciistream.cpp b/vespalib/src/vespa/vespalib/stllike/asciistream.cpp
index c141d35e80e..30a963c374c 100644
--- a/vespalib/src/vespa/vespalib/stllike/asciistream.cpp
+++ b/vespalib/src/vespa/vespalib/stllike/asciistream.cpp
@@ -235,6 +235,17 @@ asciistream & asciistream::operator >> (char & v)
     return *this;
 }
 
+asciistream & asciistream::operator >> (signed char & v)
+{
+    for (;(_rPos < length()) && std::isspace(_rbuf[_rPos]); _rPos++);
+    if (_rPos < length()) {
+        v = _rbuf[_rPos++];
+    } else {
+        throwUnderflow(_rPos);
+    }
+    return *this;
+}
+
 asciistream & asciistream::operator >> (unsigned char & v)
 {
     for (;(_rPos < length()) && std::isspace(_rbuf[_rPos]); _rPos++);
diff --git a/vespalib/src/vespa/vespalib/stllike/asciistream.h b/vespalib/src/vespa/vespalib/stllike/asciistream.h
index 88c3e1f7fc8..2f11d902283 100644
--- a/vespalib/src/vespa/vespalib/stllike/asciistream.h
+++ b/vespalib/src/vespa/vespalib/stllike/asciistream.h
@@ -64,6 +64,7 @@ public:
     asciistream & operator >> (FloatModifier v)           { _floatModifier = v; return *this; }
     asciistream & operator >> (bool & v);
     asciistream & operator >> (char & v);
+    asciistream & operator >> (signed char & v);
     asciistream & operator >> (unsigned char & v);
     asciistream & operator >> (std::string & v);
     asciistream & operator >> (string & v);