RFR(XS) 8224558: x86 Fix replicateB encoding
vladimir.x.ivanov at oracle.com
Wed May 22 18:37:09 UTC 2019
> On 22 May 2019, at 21:19, Deshpande, Vivek R <vivek.r.deshpande at intel.com> wrote:
> Hi Vladimir
> They both(Repl4B_mem vs loadB+Repl4B ) would be same in this case. May be we can remove the memory variants as the current memory variants are not correct.
> I can prepare a patch according to that.
Yes, I prefer redundant instructions to go away.
The only difference I noticed is how byte is loaded (movzbl in fixed Repl4B_mem vs movsbl in loadB), but I assume it doesn’t matter, right?
> -----Original Message-----
> From: Vladimir Ivanov [mailto:vladimir.x.ivanov at oracle.com]
> Sent: Wednesday, May 22, 2019 11:04 AM
> To: Deshpande, Vivek R <vivek.r.deshpande at intel.com>; 'hotspot-compiler-dev at openjdk.java.net compiler' <hotspot-compiler-dev at openjdk.java.net>
> Subject: Re: RFR(XS) 8224558: x86 Fix replicateB encoding
> What's the benefit of keeping those memory variants now and not simply rely on a pair of load + replicate instructions to match (ReplicateB (LoadB mem)) (e.g., Repl4B_mem vs loadB+Repl4B)?
> Best regards,
> Vladimir Ivanov
More information about the hotspot-compiler-dev