Re: Lack of capability to represent arbitrary alias dependent information

2017-06-12 Thread Richard Biener
On Mon, Jun 12, 2017 at 2:00 PM, Bin.Cheng wrote: > On Mon, Jun 12, 2017 at 11:46 AM, Bin.Cheng wrote: >> On Mon, Jun 12, 2017 at 10:15 AM, Richard Biener >> wrote: >>> On Mon, Jun 12, 2017 at 11:02 AM, Bin.Cheng wrote: HI, GCC adds runtime alias checks for data references in passes l

Re: Lack of capability to represent arbitrary alias dependent information

2017-06-12 Thread Bin.Cheng
On Mon, Jun 12, 2017 at 11:46 AM, Bin.Cheng wrote: > On Mon, Jun 12, 2017 at 10:15 AM, Richard Biener > wrote: >> On Mon, Jun 12, 2017 at 11:02 AM, Bin.Cheng wrote: >>> HI, >>> GCC adds runtime alias checks for data references in passes like >>> vectorizer, it would be very useful to pass along

Re: Lack of capability to represent arbitrary alias dependent information

2017-06-12 Thread Bin.Cheng
On Mon, Jun 12, 2017 at 10:15 AM, Richard Biener wrote: > On Mon, Jun 12, 2017 at 11:02 AM, Bin.Cheng wrote: >> HI, >> GCC adds runtime alias checks for data references in passes like >> vectorizer, it would be very useful to pass along the runtime alias >> dependent information to later passes.

Re: Lack of capability to represent arbitrary alias dependent information

2017-06-12 Thread Richard Biener
On Mon, Jun 12, 2017 at 11:02 AM, Bin.Cheng wrote: > HI, > GCC adds runtime alias checks for data references in passes like > vectorizer, it would be very useful to pass along the runtime alias > dependent information to later passes. Given below expample: > > int foo (int *a, int *b, int *c, int

Lack of capability to represent arbitrary alias dependent information

2017-06-12 Thread Bin.Cheng
HI, GCC adds runtime alias checks for data references in passes like vectorizer, it would be very useful to pass along the runtime alias dependent information to later passes. Given below expample: int foo (int *a, int *b, int *c, int *d, int *e, int len, int v) { int k, x; for (k = 0; k < le