weaselab/conflict-set
2
0
Fork 0
Code Issues 4 Packages Releases 13 Activity

Commit to 32-bit versions

Browse Source
This commit is contained in:
Andrew Noyes
2024-07-02 10:30:22 -07:00
parent 01488880ef
commit 68ab9a9f08
2 changed files with 2 additions and 51 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
ConflictSet.cpp
View File

@@ -78,11 +78,6 @@ constexpr void removeKey(struct Node *) {}
// ==================== BEGIN IMPLEMENTATION ====================
#ifndef INTERNAL_VERSION_32_BIT
#define INTERNAL_VERSION_32_BIT 0
#endif
#if INTERNAL_VERSION_32_BIT
struct InternalVersionT {
constexpr InternalVersionT() = default;
constexpr explicit InternalVersionT(int64_t value) : value(value) {}
@@ -99,20 +94,6 @@ private:
uint32_t value;
};
thread_local InternalVersionT InternalVersionT::zero;
#else
struct InternalVersionT {
constexpr InternalVersionT() = default;
constexpr explicit InternalVersionT(int64_t value) : value(value) {}
constexpr int64_t toInt64() const { return value; } // GCOVR_EXCL_LINE
constexpr auto operator<=>(const InternalVersionT &rhs) const = default;
constexpr bool operator==(const InternalVersionT &) const = default;
static const InternalVersionT zero;
private:
int64_t value;
};
const InternalVersionT InternalVersionT::zero{0};
#endif
struct Entry {
InternalVersionT pointVersion;
@@ -527,13 +508,8 @@ std::string getSearchPath(Node *n);
// Each node with an entry present gets a budget of kBytesPerKey. Node0 always
// has an entry present.
// Induction hypothesis is that each node's surplus is >= kMinNodeSurplus
#if INTERNAL_VERSION_32_BIT
constexpr int kBytesPerKey = 112;
constexpr int kMinNodeSurplus = 80;
#else
constexpr int kBytesPerKey = 144;
constexpr int kMinNodeSurplus = 104;
#endif
constexpr int kMinChildrenNode3 = 2;
constexpr int kMinChildrenNode16 = 4;
constexpr int kMinChildrenNode48 = 17;
@@ -1833,7 +1809,6 @@ bool scan16(const InternalVersionT *vs, const uint8_t *is, int begin, int end,
uint64_t mask = vget_lane_u64(
vreinterpret_u64_u8(vshrn_n_u16(vreinterpretq_u16_u8(results), 4)), 0);
#if INTERNAL_VERSION_32_BIT
uint32x4_t w4[4];
memcpy(w4, vs, sizeof(w4));
uint32_t rv;
@@ -1853,12 +1828,6 @@ bool scan16(const InternalVersionT *vs, const uint8_t *is, int begin, int end,
uint64_t compared = vget_lane_u64(
vreinterpret_u64_u8(vshrn_n_u16(vreinterpretq_u16_u8(combined), 4)), 0);
#else
uint64_t compared = 0;
for (int i = 0; i < 16; ++i) {
compared |= uint64_t(vs[i] > readVersion) << (i << 2);
}
#endif
return !(compared & mask);
@@ -1871,17 +1840,11 @@ bool scan16(const InternalVersionT *vs, const uint8_t *is, int begin, int end,
indices, _mm_max_epu8(indices, _mm_set1_epi8(end - begin))));
uint32_t compared = 0;
#if INTERNAL_VERSION_32_BIT
if constexpr (kAVX512) {
compared = compare16_32bit_avx512(vs, readVersion);
} else {
compared = compare16_32bit(vs, readVersion);
}
#else
for (int i = 0; i < 16; ++i) {
compared |= (vs[i] > readVersion) << i;
}
#endif
return !(compared & mask);
#else
@@ -1914,7 +1877,7 @@ scan16(const InternalVersionT *vs, int begin, int end,
assert(0 <= end && end <= 16);
assert(begin <= end);
#if INTERNAL_VERSION_32_BIT && defined(HAS_ARM_NEON)
#if defined(HAS_ARM_NEON)
uint32x4_t w4[4];
memcpy(w4, vs, sizeof(w4));
uint32_t rv;
@@ -1938,7 +1901,7 @@ scan16(const InternalVersionT *vs, int begin, int end,
conflict &= end == 16 ? -1 : (uint64_t(1) << (end << 2)) - 1;
conflict >>= begin << 2;
return !conflict;
#elif INTERNAL_VERSION_32_BIT && defined(HAS_AVX)
#elif defined(HAS_AVX)
uint32_t conflict;
if constexpr (kAVX512) {
conflict = compare16_32bit_avx512(vs, readVersion);
@@ -2886,9 +2849,7 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
void addWrites(const WriteRange *writes, int count, int64_t writeVersion) {
assert(writeVersion >= newestVersionFullPrecision);
newestVersionFullPrecision = writeVersion;
#if INTERNAL_VERSION_32_BIT
InternalVersionT::zero = oldestVersion;
#endif
for (int i = 0; i < count; ++i) {
const auto &w = writes[i];
auto begin = std::span<const uint8_t>(w.begin.p, w.begin.len);
@@ -2910,9 +2871,7 @@ struct __attribute__((visibility("hidden"))) ConflictSet::Impl {
assert(o >= oldestVersionFullPrecision);
this->oldestVersionFullPrecision = o;
this->oldestVersion = oldestVersion;
#if INTERNAL_VERSION_32_BIT
InternalVersionT::zero = oldestVersion;
#endif
#ifdef NDEBUG
// This is here for performance reasons, since we want to amortize the cost
// of storing the search path as a string. In tests, we want to exercise the