nvCOMP GZIP integration (#16770)

nvCOMP GZIP integration. Opt-in for now. Authors: - Vukasin Milovanovic (https://github.com/vuule) Approvers: - Mark Harris (https://github.com/harrism) - Nghia Truong (https://github.com/ttnghia) - Muhammad Haseeb (https://github.com/mhaseeb123) URL: #16770
rapidsai · Sep 11, 2024 · 750adca · 750adca
1 parent 4cdb1bf
commit 750adca
Show file tree

Hide file tree

Showing 4 changed files with 29 additions and 7 deletions.
diff --git a/cpp/include/cudf/io/nvcomp_adapter.hpp b/cpp/include/cudf/io/nvcomp_adapter.hpp
@@ -24,7 +24,7 @@
 namespace CUDF_EXPORT cudf {
 namespace io::nvcomp {
 
-enum class compression_type { SNAPPY, ZSTD, DEFLATE, LZ4 };
+enum class compression_type { SNAPPY, ZSTD, DEFLATE, LZ4, GZIP };
 
 /**
  * @brief Set of parameters that impact whether nvCOMP features are enabled.

diff --git a/cpp/src/io/comp/nvcomp_adapter.cpp b/cpp/src/io/comp/nvcomp_adapter.cpp
@@ -23,6 +23,7 @@
 #include <cudf/utilities/error.hpp>
 
 #include <nvcomp/deflate.h>
+#include <nvcomp/gzip.h>
 #include <nvcomp/lz4.h>
 #include <nvcomp/snappy.h>
 #include <nvcomp/zstd.h>
@@ -44,6 +45,8 @@ auto batched_decompress_get_temp_size_ex(compression_type compression, Args&&...
       return nvcompBatchedLZ4DecompressGetTempSizeEx(std::forward<Args>(args)...);
     case compression_type::DEFLATE:
       return nvcompBatchedDeflateDecompressGetTempSizeEx(std::forward<Args>(args)...);
+    case compression_type::GZIP:
+      return nvcompBatchedGzipDecompressGetTempSizeEx(std::forward<Args>(args)...);
     default: CUDF_FAIL("Unsupported compression type");
   }
 }
@@ -73,6 +76,8 @@ auto batched_decompress_async(compression_type compression, Args&&... args)
     case compression_type::DEFLATE:
       return nvcompBatchedDeflateDecompressAsync(std::forward<Args>(args)...);
     case compression_type::LZ4: return nvcompBatchedLZ4DecompressAsync(std::forward<Args>(args)...);
+    case compression_type::GZIP:
+      return nvcompBatchedGzipDecompressAsync(std::forward<Args>(args)...);
     default: CUDF_FAIL("Unsupported compression type");
   }
 }
@@ -84,6 +89,7 @@ std::string compression_type_name(compression_type compression)
     case compression_type::ZSTD: return "Zstandard";
     case compression_type::DEFLATE: return "Deflate";
     case compression_type::LZ4: return "LZ4";
+    case compression_type::GZIP: return "GZIP";
   }
   return "compression_type(" + std::to_string(static_cast<int>(compression)) + ")";
 }
@@ -359,8 +365,8 @@ std::optional<std::string> is_compression_disabled_impl(compression_type compres
         return "nvCOMP use is disabled through the `LIBCUDF_NVCOMP_POLICY` environment variable.";
       }
       return std::nullopt;
+    default: return "Unsupported compression type";
   }
-  return "Unsupported compression type";
 }
 
 std::optional<std::string> is_compression_disabled(compression_type compression,
@@ -396,7 +402,8 @@ std::optional<std::string> is_decompression_disabled_impl(compression_type compr
                                                           feature_status_parameters params)
 {
   switch (compression) {
-    case compression_type::DEFLATE: {
+    case compression_type::DEFLATE:
+    case compression_type::GZIP: {
       if (not params.are_all_integrations_enabled) {
         return "DEFLATE decompression is experimental, you can enable it through "
                "`LIBCUDF_NVCOMP_POLICY` environment variable.";
@@ -447,6 +454,7 @@ std::optional<std::string> is_decompression_disabled(compression_type compressio
 size_t required_alignment(compression_type compression)
 {
   switch (compression) {
+    case compression_type::GZIP:
     case compression_type::DEFLATE: return nvcompDeflateRequiredAlignment;
     case compression_type::SNAPPY: return nvcompSnappyRequiredAlignment;
     case compression_type::ZSTD: return nvcompZstdRequiredAlignment;
@@ -462,7 +470,7 @@ std::optional<size_t> compress_max_allowed_chunk_size(compression_type compressi
     case compression_type::SNAPPY: return nvcompSnappyCompressionMaxAllowedChunkSize;
     case compression_type::ZSTD: return nvcompZstdCompressionMaxAllowedChunkSize;
     case compression_type::LZ4: return nvcompLZ4CompressionMaxAllowedChunkSize;
-    default: return std::nullopt;
+    default: CUDF_FAIL("Unsupported compression type");
   }
 }
 

diff --git a/cpp/src/io/parquet/reader_impl_chunking.cu b/cpp/src/io/parquet/reader_impl_chunking.cu
@@ -865,8 +865,18 @@ std::vector<row_range> compute_page_splits_by_row(device_span<cumulative_page_in
 
     switch (codec.compression_type) {
       case GZIP:
-        gpuinflate(
-          d_comp_in_view, d_comp_out_view, d_comp_res_view, gzip_header_included::YES, stream);
+        if (cudf::io::nvcomp_integration::is_all_enabled()) {
+          nvcomp::batched_decompress(nvcomp::compression_type::GZIP,
+                                     d_comp_in_view,
+                                     d_comp_out_view,
+                                     d_comp_res_view,
+                                     codec.max_decompressed_size,
+                                     codec.total_decomp_size,
+                                     stream);
+        } else {
+          gpuinflate(
+            d_comp_in_view, d_comp_out_view, d_comp_res_view, gzip_header_included::YES, stream);
+        }
         break;
       case SNAPPY:
         if (cudf::io::nvcomp_integration::is_stable_enabled()) {

diff --git a/docs/cudf/source/user_guide/io/io.md b/docs/cudf/source/user_guide/io/io.md
@@ -75,7 +75,6 @@ IO format.
 
 </div>
 
-
 **Notes:**
 
 - \[¹\] - Not all orientations are GPU-accelerated.
@@ -177,4 +176,9 @@ If no value is set, behavior will be the same as the "STABLE" option.
     +-----------------------+--------+--------+--------------+--------------+---------+--------+--------------+--------------+--------+
     | DEFLATE               | ❌     | ❌     | ❌           | ❌           | ❌      | ❌     | Experimental | Experimental | ❌     |
     +-----------------------+--------+--------+--------------+--------------+---------+--------+--------------+--------------+--------+
+    | LZ4                   | ❌     | ❌     | Stable       | Stable       | ❌      | ❌     | Stable       | Stable       | ❌     |
+    +-----------------------+--------+--------+--------------+--------------+---------+--------+--------------+--------------+--------+
+    | GZIP                  | ❌     | ❌     | Experimental | Experimental | ❌      | ❌     | ❌           | ❌           | ❌     |
+    +-----------------------+--------+--------+--------------+--------------+---------+--------+--------------+--------------+--------+
+
 ```