|Sparse graphs? email@example.com (V.C. SREEDHAR) (1994-08-10)|
|Re: Sparse graphs? firstname.lastname@example.org (1994-08-17)|
|From:||"V.C. SREEDHAR" <email@example.com>|
|Date:||Wed, 10 Aug 1994 13:10:16 GMT|
How many of you are seriously using Sparse Evaluation Graphs
for data flow analysis? (Please ignore SSA form
and the related DFG, VDG , I know the story here;-) )
Only two persons that I know of have mentioned that they
use SEGs (to bit-vectors) in their work
Of course, one is the original guys (Ron Cytron and the gang)
And the other is Preston Briggs (Preston, please correct
me if I am wrong ;-) ). Preston mentions in his thesis that
he ("preffers") SEGs to bit-vectors for live-variable
anaysis (for reasons discussed in his thesis).
I know that SEGs are more
general than bit-vectors (also termed as slot-wise analysis).
Dhamdhere, Rosen and Zadeck in their PLDI'92 paper
show how to perform sparse evaluation using bit-vectors.
Are their any quantitative analysis that shows that one
is better than the other?
Thanks a lot
PS: For those you who are not aware of SEGs, please read the paper
by Choi et. al. in POPL'91.
Return to the
Search the comp.compilers archives again.