Feature #10329: [PATCH 2/1] optimize: foo == "literal string"

• Status: Rejected
• Assignee: -
• Target version: 2.2.0

[normalperson (Eric Wong)]

Follow-up-to: [Feature #10326] (optimize: recv << "literal string")

This speeds up some string comparisons greatly, vm2_streq1 results:

[["loop_whileloop2",
[[0.21688890410587192,
0.21697380719706416,
0.21685226494446397,
0.21750432904809713],
[0.21682232478633523,
0.2153299949131906,
0.21504606027156115,
0.21595032326877117]]],
["vm2_streq1",
[[0.8923470871523023,
0.9007423971779644,
0.8952413029037416,
0.9235154730267823],
[0.42798056406900287,
0.3970582918263972,
0.4244600120000541,
0.4243976562283933]]]]

Elapsed time: 7.020550268 (sec)

benchmark results:
minimum results in each 4 measurements.
Execution time (sec)
name a b
loop_whileloop2 0.217 0.215
vm2_streq1* 0.675 0.182

Speedup ratio: compare with the result of `a' (greater is better)
name b
loop_whileloop2 1.008
vm2_streq1* 3.711

I have not figured out the reverse (yoda condition), yet:

"literal string" == foo

---

benchmark/bm_vm2_streq1.rb | 6 ++++++
compile.c | 23 +++++++++++++++++++++++
insns.def | 21 +++++++++++++++++++++
test/ruby/test_string.rb | 29 +++++++++++++++++++++++++++++
4 files changed, 79 insertions(+)
create mode 100644 benchmark/bm_vm2_streq1.rb

[normalperson (Eric Wong)]

• Related to Feature #10326: [PATCH] optimize: recv << "literal string" added

[normalperson (Eric Wong)]

• Status changed from Open to Rejected

obsoleted by Feature #10423