[REVIEW] Add MD5 to existing hashing functionality

CHANGELOG.md

@@ -50,6 +50,7 @@
 - PR #5658 Add `filter_tokens` nvtext API
 - PR #5666 Add `filter_characters_of_type` strings API
 - PR #5673 Always build and test with per-thread default stream enabled in the GPU CI build
+- PR #5438 Add MD5 hash support
 - PR #5704 Initial `fixed_point` Column Support
 - PR #5716 Add `double_type_dispatcher` to libcudf
 - PR #5739 Add `nvtext::detokenize` API

cpp/CMakeLists.txt

@@ -423,6 +423,7 @@ add_library(cudf
             src/stream_compaction/drop_duplicates.cu
             src/datetime/datetime_ops.cu
             src/hash/hashing.cu
+            src/hash/hash_constants.cu
             src/partitioning/partitioning.cu
             src/quantiles/quantile.cu
             src/quantiles/quantiles.cu

cpp/include/cudf/detail/hashing.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2020, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -37,9 +37,21 @@ std::pair<std::unique_ptr<table>, std::vector<size_type>> hash_partition(
  * @param stream CUDA stream used for device memory operations and kernel launches.
  */
 std::unique_ptr<column> hash(table_view const& input,
+                             hash_id hash_function                     = hash_id::HASH_MURMUR3,
                              std::vector<uint32_t> const& initial_hash = {},
                              rmm::mr::device_memory_resource* mr = rmm::mr::get_default_resource(),
                              cudaStream_t stream                 = 0);
 
+std::unique_ptr<column> murmur_hash3_32(
+  table_view const& input,
+  std::vector<uint32_t> const& initial_hash = {},
+  rmm::mr::device_memory_resource* mr       = rmm::mr::get_default_resource(),
+  cudaStream_t stream                       = 0);
+
+std::unique_ptr<column> md5_hash(
+  table_view const& input,
+  rmm::mr::device_memory_resource* mr = rmm::mr::get_default_resource(),
+  cudaStream_t stream                 = 0);
+
 }  // namespace detail
 }  // namespace cudf

cpp/include/cudf/detail/utilities/hash_functions.cuh

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017, NVIDIA CORPORATION.
+ * Copyright (c) 2017-2020, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -16,10 +16,233 @@
 
 #pragma once
 
+#include <cudf/column/column_device_view.cuh>
 #include <cudf/strings/string_view.cuh>
+#include <hash/hash_constants.hpp>
+
+#include "cuda_runtime_api.h"
+#include "cudf/types.hpp"
 
 using hash_value_type = uint32_t;
 
+namespace cudf {
+namespace detail {
+/**
+ * Modified GPU implementation of
+ * https://johnnylee-sde.github.io/Fast-unsigned-integer-to-hex-string/ that is lowercase only and
+ * does not flip the endianness of the input.
+ */
+void CUDA_DEVICE_CALLABLE uint32ToLowercaseHexString(uint32_t num, char* destination)
+{
+  // Transform 0xABCD1234 => 0x0000ABCD00001234 => 0x0B0A0D0C02010403
+  uint64_t x = num;
+  x          = ((x & 0xFFFF0000) << 16) | ((x & 0xFFFF));
+  x          = ((x & 0xF0000000F) << 8) | ((x & 0xF0000000F0) >> 4) | ((x & 0xF0000000F00) << 16) |
+      ((x & 0xF0000000F000) << 4);
+
+  // Calculate a mask of ascii value offsets for bytes that contain alphabetical hex digits
+  uint64_t offsets = (((x + 0x0606060606060606) >> 4) & 0x0101010101010101) * 0x27;
+
+  x |= 0x3030303030303030;
+  x += offsets;
+  thrust::copy_n(thrust::seq, reinterpret_cast<uint8_t*>(&x), 8, destination);
+}
+
+struct MD5Hash {
+  __device__ MD5Hash(md5_hash_constants_type const* hash_constants,
+                     md5_shift_constants_type const* shift_constants)
+    : d_hash_constants(hash_constants), d_shift_constants(shift_constants)
+  {
+  }
+
+  /**
+   * @brief Core MD5 algorithm implementation. Processes a single 512-bit chunk,
+   * updating the hash value so far. Does not zero out the buffer contents.
+   */
+  void __device__ hash_step(md5_intermediate_data* hash_state) const
+  {
+    uint32_t A = hash_state->hash_value[0];
+    uint32_t B = hash_state->hash_value[1];
+    uint32_t C = hash_state->hash_value[2];
+    uint32_t D = hash_state->hash_value[3];
+
+    for (unsigned int j = 0; j < 64; j++) {
+      uint32_t F;
+      uint32_t g;
+      switch (j / 16) {
+        case 0:
+          F = (B & C) | ((~B) & D);
+          g = j;
+          break;
+        case 1:
+          F = (D & B) | ((~D) & C);
+          g = (5 * j + 1) % 16;
+          break;
+        case 2:
+          F = B ^ C ^ D;
+          g = (3 * j + 5) % 16;
+          break;
+        case 3:
+          F = C ^ (B | (~D));
+          g = (7 * j) % 16;
+          break;
+      }
+
+      uint32_t buffer_element_as_int;
+      std::memcpy(&buffer_element_as_int, hash_state->buffer + g * 4, 4);
+      F = F + A + d_hash_constants[j] + buffer_element_as_int;
+      A = D;
+      D = C;
+      C = B;
+      B = B + __funnelshift_l(F, F, d_shift_constants[((j / 16) * 4) + (j % 4)]);
+    }
+
+    hash_state->hash_value[0] += A;
+    hash_state->hash_value[1] += B;
+    hash_state->hash_value[2] += C;
+    hash_state->hash_value[3] += D;
+
+    hash_state->buffer_length = 0;
+  }
+  /**
+   * @brief Core MD5 element processing function
+   */
+  template <typename TKey>
+  void __device__ process(TKey const& key, md5_intermediate_data* hash_state) const
+  {
+    uint32_t const len  = sizeof(TKey);
+    uint8_t const* data = reinterpret_cast<uint8_t const*>(&key);
+    hash_state->message_length += len;
+
+    if (hash_state->buffer_length + len < 64) {
+      thrust::copy_n(thrust::seq, data, len, hash_state->buffer + hash_state->buffer_length);
+      hash_state->buffer_length += len;
+    } else {
+      uint32_t copylen = 64 - hash_state->buffer_length;
+
+      thrust::copy_n(thrust::seq, data, copylen, hash_state->buffer + hash_state->buffer_length);
+      hash_step(hash_state);
+
+      while (len > 64 + copylen) {
+        thrust::copy_n(thrust::seq, data + copylen, 64, hash_state->buffer);
+        hash_step(hash_state);
+        copylen += 64;
+      }
+
+      thrust::copy_n(thrust::seq, data + copylen, len - copylen, hash_state->buffer);
+      hash_state->buffer_length = len - copylen;
+    }
+  }
+
+  void __device__ finalize(md5_intermediate_data* hash_state, char* result_location) const
+  {
+    auto const full_length = (static_cast<uint64_t>(hash_state->message_length)) << 3;
+    thrust::fill_n(thrust::seq, hash_state->buffer + hash_state->buffer_length, 1, 0x80);
+
+    if (hash_state->buffer_length < 56) {
+      thrust::fill_n(thrust::seq,
+                     hash_state->buffer + hash_state->buffer_length + 1,
+                     (55 - hash_state->buffer_length),
+                     0x00);
+    } else {
+      thrust::fill_n(thrust::seq,
+                     hash_state->buffer + hash_state->buffer_length + 1,
+                     (64 - hash_state->buffer_length),
+                     0x00);
+      hash_step(hash_state);
+
+      thrust::fill_n(thrust::seq, hash_state->buffer, 56, 0x00);
+    }
+
+    thrust::copy_n(
+      thrust::seq, reinterpret_cast<uint8_t const*>(&full_length), 8, hash_state->buffer + 56);
+    hash_step(hash_state);
+
+#pragma unroll
+    for (int i = 0; i < 4; ++i)
+      uint32ToLowercaseHexString(hash_state->hash_value[i], result_location + (8 * i));
+  }
+
+  template <typename T, typename std::enable_if_t<is_chrono<T>()>* = nullptr>
+  void __device__ operator()(column_device_view col,
+                             size_type row_index,
+                             md5_intermediate_data* hash_state) const
+  {
+    release_assert(false && "MD5 Unsupported chrono type column");
+  }
+
+  template <typename T, typename std::enable_if_t<!is_fixed_width<T>()>* = nullptr>
+  void __device__ operator()(column_device_view col,
+                             size_type row_index,
+                             md5_intermediate_data* hash_state) const
+  {
+    release_assert(false && "MD5 Unsupported non-fixed-width type column");
+  }
+
+  template <typename T, typename std::enable_if_t<is_floating_point<T>()>* = nullptr>
+  void __device__ operator()(column_device_view col,
+                             size_type row_index,
+                             md5_intermediate_data* hash_state) const
+  {
+    T const& key = col.element<T>(row_index);
+    if (isnan(key)) {
+      T nan = std::numeric_limits<T>::quiet_NaN();
+      process(nan, hash_state);
+    } else if (key == T{0.0}) {
+      process(T{0.0}, hash_state);
+    } else {
+      process(key, hash_state);
+    }
+  }
+
+  template <typename T,
+            typename std::enable_if_t<is_fixed_width<T>() && !is_floating_point<T>() &&
+                                      !is_chrono<T>()>* = nullptr>
+  void CUDA_DEVICE_CALLABLE operator()(column_device_view col,
+                                       size_type row_index,
+                                       md5_intermediate_data* hash_state) const
+  {
+    process(col.element<T>(row_index), hash_state);
+  }
+
+ private:
+  md5_hash_constants_type const* d_hash_constants;
+  md5_shift_constants_type const* d_shift_constants;
+};
+
+template <>
+void CUDA_DEVICE_CALLABLE MD5Hash::operator()<string_view>(column_device_view col,
+                                                           size_type row_index,
+                                                           md5_intermediate_data* hash_state) const
+{
+  string_view key     = col.element<string_view>(row_index);
+  uint32_t const len  = static_cast<uint32_t>(key.size_bytes());
+  uint8_t const* data = reinterpret_cast<uint8_t const*>(key.data());
+
+  hash_state->message_length += len;
+
+  if (hash_state->buffer_length + len < 64) {
+    thrust::copy_n(thrust::seq, data, len, hash_state->buffer + hash_state->buffer_length);
+    hash_state->buffer_length += len;
+  } else {
+    uint32_t copylen = 64 - hash_state->buffer_length;
+    thrust::copy_n(thrust::seq, data, copylen, hash_state->buffer + hash_state->buffer_length);
+    hash_step(hash_state);
+
+    while (len > 64 + copylen) {
+      thrust::copy_n(thrust::seq, data + copylen, 64, hash_state->buffer);
+      hash_step(hash_state);
+      copylen += 64;
+    }
+
+    thrust::copy_n(thrust::seq, data + copylen, len - copylen, hash_state->buffer);
+    hash_state->buffer_length = len - copylen;
+  }
+}
+
+}  // namespace detail
+}  // namespace cudf
+
 // MurmurHash3_32 implementation from
 // https://github.com/aappleby/smhasher/blob/master/src/MurmurHash3.cpp
 //-----------------------------------------------------------------------------

cpp/include/cudf/hashing.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019, NVIDIA CORPORATION.
+ * Copyright (c) 2019-2020, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -35,6 +35,7 @@ namespace cudf {
  * @returns A column where each row is the hash of a column from the input
  */
 std::unique_ptr<column> hash(table_view const& input,
+                             hash_id hash_function                     = hash_id::HASH_MURMUR3,
                              std::vector<uint32_t> const& initial_hash = {},
                              rmm::mr::device_memory_resource* mr = rmm::mr::get_default_resource());
 

cpp/include/cudf/types.hpp

@@ -279,5 +279,14 @@ inline bool operator==(data_type const& lhs, data_type const& rhs) { return lhs.
  */
 std::size_t size_of(data_type t);
 
+/**
+ *  @brief Identifies the hash function to be used
+ */
+enum class hash_id {
+  HASH_IDENTITY = 0,  ///< Identity hash function that simply returns the key to be hashed
+  HASH_MURMUR3,       ///< Murmur3 hash function
+  HASH_MD5            ///< MD5 hash function
+};
+
 /** @} */
 }  // namespace cudf