Comment #4 on issue 4014 by [email protected]: ES6 Map performance issues
https://code.google.com/p/v8/issues/detail?id=4014
True, I measure only around ~10% regression on populate-with-interned,
accesss-with-interned case. Might be flakiness.
Here are my numbers and test based on original jsPerf one
# i = interned keys
# f = flat string keys
# c = cons string keys
# m = mixed keys (every 5th is smi)
# a,b means populate map with keys of kind `a`, then read many times using
keys of kind `b`
1) Before JS implementation:
Run without feedback polution (all keys are strings):
i,i took 336 ms.
f,f took 348 ms.
c,i took 1100 ms.
c,c took 552 ms.
i,i took 353 ms.
f,f took 334 ms.
f,i took 709 ms.
i,f took 700 ms.
Run with feedback polution (few keys are smis):
m,m took 593 ms.
i,i took 363 ms.
f,f took 370 ms.
c,i took 1114 ms.
c,c took 569 ms.
i,i took 364 ms.
f,f took 368 ms.
f,i took 718 ms.
i,f took 719 ms
2) New Version:
Without type feedback pollution:
i,i took 369 ms.
f,f took 408 ms.
c,i took 1251 ms.
c,c took 585 ms.
i,i took 405 ms.
f,f took 397 ms.
f,i took 783 ms.
i,f took 789 ms.
With type feedback pollution:
m,m took 638 ms.
i,i took 481 ms.
f,f took 465 ms.
c,i took 1415 ms.
c,c took 695 ms.
i,i took 481 ms.
f,f took 468 ms.
f,i took 870 ms.
i,f took 906 ms.
Attachments:
g.js 1.7 KB
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
