Closed Bug 501186 Opened 16 years ago Closed 12 years ago

Profile object creation performance

Categories

(Core :: JavaScript Engine, defect)

defect
Not set
normal

Tracking

()

RESOLVED FIXED

People

(Reporter: dmandelin, Assigned: gal)

References

Details

Attachments

(2 files)

Attached file Test case
Results of attached test case: run time (ms) empty loop run time (ms) cycles/obj alloc js -j (TM) 1307 38 280 jsc (webkit) 497 32 86 v8 397 104 65 We are 3.3-4.3x slower at allocating new empty objects. Let's find out why and fix it.
Assignee: general → gal
Attached patch patchSplinter Review
I don't see any speedup from this. Maybe I am not measuring right though.
Gregor, can you measure with this patch? If that moves the needle even a little, I can remove the other division too. I just want to make sure we are not chasing a ghost here.
I dont have the 1:1 mapping from source code to the disassembled instructions. There is still one division left where we spend now 50% of the time in this function due to shark. Probably its the GC_FREELIST_INDEX macro.
Depends on: 501515
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: