[9] RFR(S): 8075324: Costs of memory operands in aarch64.ad are inconsistent (original) (raw)

Andrew Dinn adinn at redhat.com
Wed Mar 18 11:22:59 UTC 2015

On 18/03/15 10:46, Tobias Hartmann wrote:

thanks a lot for the detailed explanation! That makes sense to me.

I was confused by the fact that the generation of the additional add in the case of a scaled-indirect-with-offset, is done 'implicitly' in loadStore and therefore we have to account for that by setting a higher cost for the corresponding memory operand. Do you think we should still fix the cost of 'indOffI'? If so, here is the corresponding webrev: http://cr.openjdk.java.net/~thartmann/8075324/webrev.01/

Yes, that's the correct fix for the existing rules. I'll respond regarding your new proposed rules in the other thread.

regards,

Andrew Dinn

Senior Principal Software Engineer Red Hat UK Ltd Registered in UK and Wales under Company Registration No. 3798903 Directors: Michael Cunningham (USA), Matt Parson (USA), Charlie Peters (USA), Michael O'Neill (Ireland)

More information about the hotspot-compiler-dev mailing list