Remove upb_dstate and specialize upb_decode_fixed for perf improvement.

The compiler wasn't keeping upb_dstate in memory
anyway (which was the original goal).  This
simplifies the decoder.  upb_decode_fixed
was intended to minimize the number of branches,
but since it was calling out to memcpy as a
function, this turned out to be a pessimization.

Performance is encouraging:

  plain32.parsestream_googlemessage1.upb_table: 254 -> 242 (-4.72)
  plain32.parsestream_googlemessage2.upb_table: 357 -> 400 (12.04)
  plain32.parsetostruct_googlemessage1.upb_table_byref: 143 -> 144 (0.70)
  plain32.parsetostruct_googlemessage1.upb_table_byval: 122 -> 118 (-3.28)
  plain32.parsetostruct_googlemessage2.upb_table_byref: 189 -> 200 (5.82)
  plain32.parsetostruct_googlemessage2.upb_table_byval: 198 -> 200 (1.01)
  omitfp32.parsestream_googlemessage1.upb_table: 267 -> 265 (-0.75)
  omitfp32.parsestream_googlemessage2.upb_table: 377 -> 465 (23.34)
  omitfp32.parsetostruct_googlemessage1.upb_table_byref: 140 -> 151 (7.86)
  omitfp32.parsetostruct_googlemessage1.upb_table_byval: 131 -> 131 (0.00)
  omitfp32.parsetostruct_googlemessage2.upb_table_byref: 204 -> 214 (4.90)
  omitfp32.parsetostruct_googlemessage2.upb_table_byval: 200 -> 206 (3.00)
  plain.parsestream_googlemessage1.upb_table: 313 -> 317 (1.28)
  plain.parsestream_googlemessage2.upb_table: 476 -> 541 (13.66)
  plain.parsetostruct_googlemessage1.upb_table_byref: 189 -> 189 (0.00)
  plain.parsetostruct_googlemessage1.upb_table_byval: 165 -> 165 (0.00)
  plain.parsetostruct_googlemessage2.upb_table_byref: 263 -> 270 (2.66)
  plain.parsetostruct_googlemessage2.upb_table_byval: 248 -> 255 (2.82)
  omitfp.parsestream_googlemessage1.upb_table: 306 -> 305 (-0.33)
  omitfp.parsestream_googlemessage2.upb_table: 471 -> 531 (12.74)
  omitfp.parsetostruct_googlemessage1.upb_table_byref: 189 -> 190 (0.53)
  omitfp.parsetostruct_googlemessage1.upb_table_byval: 166 -> 172 (3.61)
  omitfp.parsetostruct_googlemessage2.upb_table_byref: 258 -> 270 (4.65)
  omitfp.parsetostruct_googlemessage2.upb_table_byval: 248 -> 265 (6.85)

1 files changed, 11 insertions, 12 deletions

diff --git a/perf-tests.sh b/perf-tests.sh
index 3833572..ca7e4ff 100755
--- a/perf-tests.sh
+++ b/perf-tests.sh
@@ -11,25 +11,24 @@ fi
 
 rm -f perf-tests.out
 
-make clean
-echo "-O3 -DNDEBUG -msse3 -DUPB_THREAD_UNSAFE" > perf-cppflags
-make $MAKETARGET
-make benchmark | sed -e 's/^/plain./g' | tee -a perf-tests.out
-
-make clean
-echo "-O3 -DNDEBUG -fomit-frame-pointer -msse3 -DUPB_THREAD_UNSAFE" > perf-cppflags
-make $MAKETARGET
-make benchmark | sed -e 's/^/omitfp./g' | tee -a perf-tests.out
-
 if [ x`uname -m` == xx86_64 ]; then
   make clean
-  echo "-O3 -DNDEBUG -msse3 -m32 -DUPB_THREAD_UNSAFE" > perf-cppflags
+  echo "-O3 -DNDEBUG -msse3 -m32" > perf-cppflags
   make upb_benchmarks
   make benchmark | sed -e 's/^/plain32./g' | tee -a perf-tests.out
 
   make clean
-  echo "-O3 -DNDEBUG -fomit-frame-pointer -msse3 -m32 -DUPB_THREAD_UNSAFE" > perf-cppflags
+  echo "-O3 -DNDEBUG -fomit-frame-pointer -msse3 -m32" > perf-cppflags
   make upb_benchmarks
   make benchmark | sed -e 's/^/omitfp32./g' | tee -a perf-tests.out
 fi
 
+make clean
+echo "-O3 -DNDEBUG -msse3" > perf-cppflags
+make $MAKETARGET
+make benchmark | sed -e 's/^/plain./g' | tee -a perf-tests.out
+
+make clean
+echo "-O3 -DNDEBUG -fomit-frame-pointer -msse3" > perf-cppflags
+make $MAKETARGET
+make benchmark | sed -e 's/^/omitfp./g' | tee -a perf-tests.out