On Fri, 15 Mar 2024 13:58:05 GMT, Hamlin Li <m...@openjdk.org> wrote:

>> Hi,
>> Can you help to review this patch?
>> Thanks
>> 
>> This is a continuation of work based on [1] by @XiaohongGong, most work was 
>> done in that pr. In this new pr, just rebased the code in [1], then added 
>> some minor changes (renaming of calling method, add libsleef as extra lib in 
>> CI cross-build on aarch64 in github workflow); I aslo tested the combination 
>> of following scenarios:
>> * at build time
>>   * with/without sleef
>>   * with/without sve support 
>> * at runtime
>>   * with/without sleef
>>   * with/without sve support 
>> 
>> [1] https://github.com/openjdk/jdk/pull/16234
>> 
>> ## Regression Test
>> * test/jdk/jdk/incubator/vector/
>> * test/hotspot/jtreg/compiler/vectorapi/
>> 
>> ## Performance Test
>> Previously, @XiaohongGong has shared the data: 
>> https://github.com/openjdk/jdk/pull/16234#issuecomment-1767727028
>
> Hamlin Li has updated the pull request incrementally with one additional 
> commit since the last revision:
> 
>   fix jni includes

> Some updates: I had a look at the source code for libsleef, and frankly, to 
> integrate building of libsleef into the OpenJDK build is going to be a major 
> PITA. :-( They have a cmake based build system, which generates a ton of 
> native build tools, which preprocess source code and generates output 
> artifacts (like sleef.h).
> 
> I guess we _could_ do it, but I am terrified about trying to get that to 
> work, especially for cross-compilation.
> 
> So after seeing this, I think the better solution is to require a 
> pre-compiled libsleef present in the system. (What I just called the "worse 
> solution" an hour ago...). I guess we can make the requirement of libsleef 
> the default value, but allow the user to override this to skip libsleef. At 
> least then it is an active choice to disable libsleef for your build.
> 
> Or, we could do something like how we used to do for freetype: have a 
> --with-libsleef-src that points to a checked out version of libsleef, and let 
> configure build it. In this case we could even build it as a static archive, 
> so there'd be no need for libvectormath.so to have any external dependencies.
> 
> In any case, we are going to have to consider carefully how to proceed.

No worry, seems works for me, I just had a very draft version tested locally.
we just need to get the final inline versions of sleef generated by sleef cmake 
install.

A question about the licensing: how long does it take to finish the legal 
process of the sleef licence?

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

PR Comment: https://git.openjdk.org/jdk/pull/18294#issuecomment-2015341870

Reply via email to