[Webkit-unassigned] [Bug 104638] Support op_switch_imm bytecode on DFG.

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=104638





--- Comment #5 from Roman Zhuykov <zhroma at ispras.ru>  2012-12-12 05:16:07 PST ---
(In reply to comment #2)
Certainly, many comparisons for large switch statements perform slower then jump table. ​I don't like the idea of additional phase which will only expand switch into blocks. This will at least perform slower then processing op_switch in DFGBytecodeParser.
Speaking about using Switch DFG node all the way through to the backend — it seems complicated to implement processing such a node in all other phases. I see a lot of code that assumes each basic block in control-flow graph have edges to one or two other blocks (jump or branch). Basic block with switch node may have up to 1000 outgoing edges. Is it possible to implement such support or there is any logic which prevents creating blocks with more than two exits?

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.