release bump

* fixing performance issue under Zen 4 processors

* trying something else

---------

Co-authored-by: Daniel Lemire <dlemire@lemire.me>

CMakeLists.txt

@@ -3,7 +3,7 @@ cmake_minimum_required(VERSION 3.14)
 project(
     simdjson
     # The version number is modified by tools/release.py
-    VERSION 3.12.1
+    VERSION 3.12.2
     DESCRIPTION "Parsing gigabytes of JSON per second"
     HOMEPAGE_URL "https://simdjson.org/"
     LANGUAGES CXX C

Doxyfile

@@ -38,7 +38,7 @@ PROJECT_NAME           = simdjson
 # could be handy for archiving the generated documentation or if some version
 # control system is used.
 
-PROJECT_NUMBER         = "3.12.1"
+PROJECT_NUMBER         = "3.12.2"
 
 # Using the PROJECT_BRIEF tag one can provide an optional one line description
 # for a project that appears at the top of each page and should give viewer a

README.md

@@ -1,5 +1,4 @@
 
-[![Ubuntu 20.04 CI](https://github.com/simdjson/simdjson/workflows/Ubuntu%2020.04%20CI%20(GCC%209)/badge.svg)](https://simdjson.org/plots.html)
 [![Fuzzing Status](https://oss-fuzz-build-logs.storage.googleapis.com/badges/simdjson.svg)](https://bugs.chromium.org/p/oss-fuzz/issues/list?sort=-opened&can=1&q=proj:simdjson)
 [![][license img]][license] [![][licensemit img]][licensemit]
 

include/simdjson/icelake/simd.h

@@ -148,14 +148,18 @@ namespace simd {
 
     // Copies to 'output" all bytes corresponding to a 0 in the mask (interpreted as a bitset).
     // Passing a 0 value for mask would be equivalent to writing out every byte to output.
-    // Only the first 32 - count_ones(mask) bytes of the result are significant but 32 bytes
+    // Only the first 64 - count_ones(mask) bytes of the result are significant but 64 bytes
     // get written.
     // Design consideration: it seems like a function with the
     // signature simd8<L> compress(uint32_t mask) would be
     // sensible, but the AVX ISA makes this kind of approach difficult.
     template<typename L>
     simdjson_inline void compress(uint64_t mask, L * output) const {
-      _mm512_mask_compressstoreu_epi8 (output,~mask,*this);
+      // we deliberately avoid _mm512_mask_compressstoreu_epi8 for portability
+      // (AMD Zen4 has terrible performance with it, it is effectively broken)
+      // _mm512_mask_compressstoreu_epi8 (output,~mask,*this);
+      __m512i compressed = _mm512_maskz_compress_epi8(~mask, *this);
+      _mm512_storeu_si512(output, compressed); // could use a mask
     }
 
     template<typename L>

include/simdjson/simdjson_version.h

@@ -4,7 +4,7 @@
 #define SIMDJSON_SIMDJSON_VERSION_H
 
 /** The version of simdjson being used (major.minor.revision) */
-#define SIMDJSON_VERSION "3.12.1"
+#define SIMDJSON_VERSION "3.12.2"
 
 namespace simdjson {
 enum {
@@ -19,7 +19,7 @@ enum {
   /**
    * The revision (major.minor.REVISION) of simdjson being used.
    */
-  SIMDJSON_VERSION_REVISION = 1
+  SIMDJSON_VERSION_REVISION = 2
 };
 } // namespace simdjson
 

singleheader/simdjson.cpp

@@ -1,4 +1,4 @@
-/* auto-generated on 2025-02-11 13:43:34 -0500. Do not edit! */
+/* auto-generated on 2025-02-14 16:11:36 -0500. Do not edit! */
 /* including simdjson.cpp:  */
 /* begin file simdjson.cpp */
 #define SIMDJSON_SRC_SIMDJSON_CPP
@@ -20813,14 +20813,18 @@ namespace simd {
 
     // Copies to 'output" all bytes corresponding to a 0 in the mask (interpreted as a bitset).
     // Passing a 0 value for mask would be equivalent to writing out every byte to output.
-    // Only the first 32 - count_ones(mask) bytes of the result are significant but 32 bytes
+    // Only the first 64 - count_ones(mask) bytes of the result are significant but 64 bytes
     // get written.
     // Design consideration: it seems like a function with the
     // signature simd8<L> compress(uint32_t mask) would be
     // sensible, but the AVX ISA makes this kind of approach difficult.
     template<typename L>
     simdjson_inline void compress(uint64_t mask, L * output) const {
-      _mm512_mask_compressstoreu_epi8 (output,~mask,*this);
+      // we deliberately avoid _mm512_mask_compressstoreu_epi8 for portability
+      // (AMD Zen4 has terrible performance with it, it is effectively broken)
+      // _mm512_mask_compressstoreu_epi8 (output,~mask,*this);
+      __m512i compressed = _mm512_maskz_compress_epi8(~mask, *this);
+      _mm512_storeu_si512(output, compressed); // could use a mask
     }
 
     template<typename L>
@@ -23443,14 +23447,18 @@ namespace simd {
 
     // Copies to 'output" all bytes corresponding to a 0 in the mask (interpreted as a bitset).
     // Passing a 0 value for mask would be equivalent to writing out every byte to output.
-    // Only the first 32 - count_ones(mask) bytes of the result are significant but 32 bytes
+    // Only the first 64 - count_ones(mask) bytes of the result are significant but 64 bytes
     // get written.
     // Design consideration: it seems like a function with the
     // signature simd8<L> compress(uint32_t mask) would be
     // sensible, but the AVX ISA makes this kind of approach difficult.
     template<typename L>
     simdjson_inline void compress(uint64_t mask, L * output) const {
-      _mm512_mask_compressstoreu_epi8 (output,~mask,*this);
+      // we deliberately avoid _mm512_mask_compressstoreu_epi8 for portability
+      // (AMD Zen4 has terrible performance with it, it is effectively broken)
+      // _mm512_mask_compressstoreu_epi8 (output,~mask,*this);
+      __m512i compressed = _mm512_maskz_compress_epi8(~mask, *this);
+      _mm512_storeu_si512(output, compressed); // could use a mask
     }
 
     template<typename L>

singleheader/simdjson.h

@@ -1,4 +1,4 @@
-/* auto-generated on 2025-02-11 13:43:34 -0500. Do not edit! */
+/* auto-generated on 2025-02-14 16:11:36 -0500. Do not edit! */
 /* including simdjson.h:  */
 /* begin file simdjson.h */
 #ifndef SIMDJSON_H
@@ -2437,7 +2437,7 @@ namespace std {
 #define SIMDJSON_SIMDJSON_VERSION_H
 
 /** The version of simdjson being used (major.minor.revision) */
-#define SIMDJSON_VERSION "3.12.1"
+#define SIMDJSON_VERSION "3.12.2"
 
 namespace simdjson {
 enum {
@@ -2452,7 +2452,7 @@ enum {
   /**
    * The revision (major.minor.REVISION) of simdjson being used.
    */
-  SIMDJSON_VERSION_REVISION = 1
+  SIMDJSON_VERSION_REVISION = 2
 };
 } // namespace simdjson
 
@@ -17948,14 +17948,18 @@ namespace simd {
 
     // Copies to 'output" all bytes corresponding to a 0 in the mask (interpreted as a bitset).
     // Passing a 0 value for mask would be equivalent to writing out every byte to output.
-    // Only the first 32 - count_ones(mask) bytes of the result are significant but 32 bytes
+    // Only the first 64 - count_ones(mask) bytes of the result are significant but 64 bytes
     // get written.
     // Design consideration: it seems like a function with the
     // signature simd8<L> compress(uint32_t mask) would be
     // sensible, but the AVX ISA makes this kind of approach difficult.
     template<typename L>
     simdjson_inline void compress(uint64_t mask, L * output) const {
-      _mm512_mask_compressstoreu_epi8 (output,~mask,*this);
+      // we deliberately avoid _mm512_mask_compressstoreu_epi8 for portability
+      // (AMD Zen4 has terrible performance with it, it is effectively broken)
+      // _mm512_mask_compressstoreu_epi8 (output,~mask,*this);
+      __m512i compressed = _mm512_maskz_compress_epi8(~mask, *this);
+      _mm512_storeu_si512(output, compressed); // could use a mask
     }
 
     template<typename L>
@@ -65401,14 +65405,18 @@ namespace simd {
 
     // Copies to 'output" all bytes corresponding to a 0 in the mask (interpreted as a bitset).
     // Passing a 0 value for mask would be equivalent to writing out every byte to output.
-    // Only the first 32 - count_ones(mask) bytes of the result are significant but 32 bytes
+    // Only the first 64 - count_ones(mask) bytes of the result are significant but 64 bytes
     // get written.
     // Design consideration: it seems like a function with the
     // signature simd8<L> compress(uint32_t mask) would be
     // sensible, but the AVX ISA makes this kind of approach difficult.
     template<typename L>
     simdjson_inline void compress(uint64_t mask, L * output) const {
-      _mm512_mask_compressstoreu_epi8 (output,~mask,*this);
+      // we deliberately avoid _mm512_mask_compressstoreu_epi8 for portability
+      // (AMD Zen4 has terrible performance with it, it is effectively broken)
+      // _mm512_mask_compressstoreu_epi8 (output,~mask,*this);
+      __m512i compressed = _mm512_maskz_compress_epi8(~mask, *this);
+      _mm512_storeu_si512(output, compressed); // could use a mask
     }
 
     template<typename L>