Hello folks,<div><br></div><div>This is the oh-so-controversial cut-over. ;] I'm continuing to convert other places which aren't quite so performance sensitive or (in most cases) use significantly slower functions, but I wanted to get confirmation on this one.</div>
<div><br></div><div>I've performed as careful analysis of the performance impact of this change as I can. I've run the lex-only mode with before and after clang binaries over very large source files. The runs were under the Linux 'perf' tool which does hardware perf-event based instrumenting of the execution, and I had it perform 40 runs of each command. I've run these with a forced maximum kernel scheduling priority and affinity pinned to a single CPU. This reduces the context switches to one (the CC1 invocation), results in zero CPU migrations, and ensures that all measurements occur on the same core of the same chip with a warm cache. The timings are (consequentially) quite stable.</div>
<div><br></div><div>Here are the numbers for single-source GCC:</div><div><div><font face="'courier new', monospace"> Performance counter stats for './bin/old_clang -fsyntax-only -Xclang -Eonly ../tools/clang/INPUTS/gcc.c -w' (40 runs):</font></div>
<div><font face="'courier new', monospace"><br></font></div><div><font face="'courier new', monospace"> 653.285653 task-clock # 0.998 CPUs utilized ( +- 0.19% )</font></div>
<div><font face="'courier new', monospace"> 1 context-switches # 0.000 M/sec ( +- 0.00% )</font></div><div><font face="'courier new', monospace"> 0 CPU-migrations # 0.000 M/sec ( +- 42.37% )</font></div>
<div><font face="'courier new', monospace"> 21,909 page-faults # 0.034 M/sec ( +- 0.00% )</font></div><div><font face="'courier new', monospace"> 1,656,298,378 cycles # 2.535 GHz ( +- 0.18% )</font></div>
<div><font face="'courier new', monospace"> 584,610,147 stalled-cycles-frontend # 35.30% frontend cycles idle ( +- 0.52% )</font></div><div><font face="'courier new', monospace"> 368,002,172 stalled-cycles-backend # 22.22% backend cycles idle ( +- 0.77% )</font></div>
<div><font face="'courier new', monospace"> 2,240,704,537 instructions # 1.35 insns per cycle</font></div><div><font face="'courier new', monospace"> # 0.26 stalled cycles per insn ( +- 0.02% )</font></div>
<div><font face="'courier new', monospace"> 494,832,319 branches # 757.452 M/sec ( +- 0.02% )</font></div><div><font face="'courier new', monospace"> 17,343,711 branch-misses # 3.50% of all branches ( +- 0.05% )</font></div>
<div><font face="'courier new', monospace"><br></font></div><div><font face="'courier new', monospace"> 0.654568555 seconds time elapsed ( +- 0.19% )</font></div>
<div><font face="'courier new', monospace"> </font></div><div><font face="'courier new', monospace"> </font></div><div><font face="'courier new', monospace"> Performance counter stats for './bin/new_clang -fsyntax-only -Xclang -Eonly ../tools/clang/INPUTS/gcc.c -w' (40 runs):</font></div>
<div><font face="'courier new', monospace"><br></font></div><div><font face="'courier new', monospace"> 652.817626 task-clock # 0.998 CPUs utilized ( +- 0.16% )</font></div>
<div><font face="'courier new', monospace"> 1 context-switches # 0.000 M/sec ( +- 0.00% )</font></div><div><font face="'courier new', monospace"> 0 CPU-migrations # 0.000 M/sec ( +- 69.80% )</font></div>
<div><font face="'courier new', monospace"> 21,913 page-faults # 0.034 M/sec ( +- 0.00% )</font></div><div><font face="'courier new', monospace"> 1,655,504,961 cycles # 2.536 GHz ( +- 0.16% )</font></div>
<div><font face="'courier new', monospace"> 579,918,671 stalled-cycles-frontend # 35.03% frontend cycles idle ( +- 0.47% )</font></div><div><font face="'courier new', monospace"> 356,098,374 stalled-cycles-backend # 21.51% backend cycles idle ( +- 0.72% )</font></div>
<div><font face="'courier new', monospace"> 2,240,570,707 instructions # 1.35 insns per cycle</font></div><div><div><div><font face="'courier new', monospace"> # 0.26 stalled cycles per insn ( +- 0.02% )</font></div>
<div><font face="'courier new', monospace"> 486,263,162 branches # 744.868 M/sec ( +- 0.02% )</font></div><div><font face="'courier new', monospace"> 16,554,052 branch-misses # 3.40% of all branches ( +- 0.05% )</font></div>
<div><font face="'courier new', monospace"><br></font></div><div><font face="'courier new', monospace"> 0.654093465 seconds time elapsed ( +- 0.16% )</font></div>
</div></div><div><br></div><div><br></div><div>Here are the numbers for a single-source file with every source file in Lex, Parse, and Sema from Clang included into it:</div><div><font face="'courier new', monospace"> Performance counter stats for './bin/old_clang -x c++ -fsyntax-only -Xclang -Eonly -I../include -Iinclude -I../tools/clang/include -Itools/clang/include ../tools/clang/INPUTS/all-clang.cpp' (40 runs):</font></div>
<div><font face="'courier new', monospace"><br></font></div><div><font face="'courier new', monospace"> 249.007880 task-clock # 0.998 CPUs utilized ( +- 0.27% )</font></div>
<div><font face="'courier new', monospace"> 1 context-switches # 0.000 M/sec ( +- 0.00% )</font></div><div><font face="'courier new', monospace"> 0 CPU-migrations # 0.000 M/sec ( +- 48.04% )</font></div>
<div><font face="'courier new', monospace"> 9,013 page-faults # 0.036 M/sec ( +- 0.00% )</font></div><div><font face="'courier new', monospace"> 630,166,134 cycles # 2.531 GHz ( +- 0.26% )</font></div>
<div><font face="'courier new', monospace"> 243,962,215 stalled-cycles-frontend # 38.71% frontend cycles idle ( +- 0.70% )</font></div><div><font face="'courier new', monospace"> 152,285,530 stalled-cycles-backend # 24.17% backend cycles idle ( +- 0.90% )</font></div>
<div><font face="'courier new', monospace"> 776,325,373 instructions # 1.23 insns per cycle </font></div><div><font face="'courier new', monospace"> # 0.31 stalled cycles per insn ( +- 0.03% )</font></div>
<div><font face="'courier new', monospace"> 180,715,948 branches # 725.744 M/sec ( +- 0.02% )</font></div><div><font face="'courier new', monospace"> 6,543,840 branch-misses # 3.62% of all branches ( +- 0.03% )</font></div>
<div><font face="'courier new', monospace"><br></font></div><div><font face="'courier new', monospace"> 0.249614202 seconds time elapsed ( +- 0.27% )</font></div>
<div><font face="'courier new', monospace"><br></font></div><div><font face="'courier new', monospace"><br></font></div><div><font face="'courier new', monospace"> Performance counter stats for './bin/new_clang -x c++ -fsyntax-only -Xclang -Eonly -I../include -Iinclude -I../tools/clang/include -Itools/clang/include ../tools/clang/INPUTS/all-clang.cpp' (40 runs):</font></div>
<div><font face="'courier new', monospace"><br></font></div><div><font face="'courier new', monospace"> 247.344347 task-clock # 0.998 CPUs utilized ( +- 0.24% )</font></div>
<div><font face="'courier new', monospace"> 1 context-switches # 0.000 M/sec ( +- 0.00% )</font></div><div><font face="'courier new', monospace"> 0 CPU-migrations # 0.000 M/sec ( +- 0.00% )</font></div>
<div><font face="'courier new', monospace"> 9,018 page-faults # 0.036 M/sec ( +- 0.00% )</font></div><div><font face="'courier new', monospace"> 626,293,160 cycles # 2.532 GHz ( +- 0.23% )</font></div>
<div><font face="'courier new', monospace"> 240,474,650 stalled-cycles-frontend # 38.40% frontend cycles idle ( +- 0.60% )</font></div><div><font face="'courier new', monospace"> 147,531,931 stalled-cycles-backend # 23.56% backend cycles idle ( +- 0.84% )</font></div>
<div><font face="'courier new', monospace"> 767,828,567 instructions # 1.23 insns per cycle </font></div><div><font face="'courier new', monospace"> # 0.31 stalled cycles per insn ( +- 0.03% )</font></div>
<div><font face="'courier new', monospace"> 173,028,132 branches # 699.544 M/sec ( +- 0.02% )</font></div><div><font face="'courier new', monospace"> 6,031,206 branch-misses # 3.49% of all branches ( +- 0.04% )</font></div>
<div><font face="'courier new', monospace"><br></font></div><div><font face="'courier new', monospace"> 0.247948747 seconds time elapsed ( +- 0.24% )</font></div>
<div><br></div></div><div><br></div><div>The measured changes in performance are all down into the noise floor, despite everything I did to lower that. From what I can tell, this change has no observable performance impact, and reduces the number of hashing functions in use. Good to commit?</div>