This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] Vectorizer cost model implementation
- From: Eric Christopher <echristo at apple dot com>
- To: "Jagasia, Harsha" <harsha dot jagasia at amd dot com>
- Cc: "Dorit Nuzman" <DORIT at il dot ibm dot com>, "GCC Patches" <gcc-patches at gcc dot gnu dot org>, "Meissner, Michael" <michael dot meissner at amd dot com>
- Date: Wed, 6 Jun 2007 18:30:03 -0700
- Subject: Re: [patch] Vectorizer cost model implementation
- References: <D5B24B5251882048AD03DDFA431BB790012F01F0@SAUSEXMB3.amd.com>
I have attached the cost model patch. The test cases have been reduced
to a smaller subset which I think should give us a fairly good
coverage
to start with. Currently I have only added the tests on x86 and
x86-64.
Want to add them in numerical order maybe?
And a few comments:
+ /* Cost model disabled. */
+ if (!flag_vect_cost_model)
+ {
+ if (vect_print_dump_info (REPORT_DETAILS))
+ fprintf (vect_dump, "cost model disabled.");
+ return 0;
+ }
Want to add this where vect_estimate_min_profitable_iters is called from
rather than inside? Lot of code to just return 0.
And in a similar vein:
STMT_VINFO_TYPE (stmt_info) = reduc_vec_info_type;
+ vect_model_reduction_cost (stmt_info, epilog_reduc_code,
ncopies);
Shouldn't we determine if we're going to compute costs before doing so?
Otherwise it looks OK to me (note I can't approve it).
-eric