RFR: 8276905: Function frag_col has a deployment target which is incompatible with this OS [v2]

Jayathirth D V jdv at openjdk.java.net
Mon Nov 15 05:36:34 UTC 2021


On Fri, 12 Nov 2021 03:59:52 GMT, Phil Race <prr at openjdk.org> wrote:

>> make/modules/java.desktop/lib/Awt2dLibraries.gmk line 850:
>> 
>>> 848:       COMMAND := $(METAL) -c -std=osx-metal2.0 \
>>> 849:           -mmacosx-version-min=$(MACOSX_VERSION_MIN) \
>>> 850:           -o $(SHADERS_AIR) $(SHADERS_SRC), \
>> 
>> No .. we decided metal requires macos 10.14 as a minimum. In fact it won't run (by policy) on earlier releases so settting it to what the broader JDK defines as a minimum is not appropriate right now.
>> And I've no idea what restrictions we'd be placing on metal saying it can only use 10.12 features.
>> Nor do I know what the xcode 12.4 used to build JDK 17 defaults to and how much a change to either 10.12 or 10.14 might require in re-testing.
>> 
>> So - 
>> we should use 10.14 
>> what's the actual impact of that on a current build using xcode 12.4 - does it change anything we use ?
>
> bear in mind "impact" might mean metal can't use some new faster code, not just runtime errors

Hi Phil,

Thanks for the review. I also had doubts using 10.12. And yes as you mentioned we dont use Metal pipeline for versions < macOS10.14.

We dont use any Metal API which is specific to >macOS10.14(We had such usage when Lanai was under development but they were removed subsequently). So i dont see any performance impact of making xcrun to use 10.14 as minimum version.

Thanks,
Jay

-------------

PR: https://git.openjdk.java.net/jdk/pull/6346



More information about the client-libs-dev mailing list