search for: gawlitza

...n a 2.4GHz machine. We obtain non-trivial bit size reductions for the small benchmarks, having results that match those found by previous, non-conservative works, such as Stephenson’s et al’s [2]. Moreover, our implementation is based on a very modern algorithm, by Su and Wagner [1], augmented with Gawlitza’s technique to handle cycles [3]. We believe that this is the fastest implementation of such an analysis. Finally, the third item is also no longer a problem. Presently LLVM offers the Lazy Value Info interface that reports when variables are constants. The current LVI implementation also provides...

...large benchmarks? If small benchmarks yield good results and large benchmarks yield poor results, then your analysis probably needs more work to be useful for real-world programs. > Moreover, our implementation is based on a very modern algorithm, by > Su and Wagner [1], augmented with Gawlitza’s technique to handle > cycles [3]. We believe that this is the fastest implementation of such > an analysis. > > Finally, the third item is also no longer a problem. Presently LLVM > offers the Lazy Value Info interface that reports when variables are > constants. The current...

On Wed, Mar 23, 2011 at 6:24 PM, John Criswell <criswell at illinois.edu> wrote: > > > On 3/23/11 8:06 AM, Douglas do Couto Teixeira wrote: > > > > the execution of a program. Thus, for each integer variable, a range > > analysis determines its lower and upper limits. A very simple range analysis > > > would, for instance, map each variable to the limits