Re: RFR: 8262096: Vector API fails to work due to VectorShape initialization exception [v4]
On Fri, 26 Feb 2021 15:48:18 GMT, Vladimir Ivanov wrote:
> > I'd like to keep DoubleVector.SPECIES_PREFERRED.length() <=
> > VectorSupport.getMaxLaneCount(double.class) for Java programmers since the
> > VectorSupport_GetMaxLaneCount is used to implement a Java API.
>
> It doesn't make much sense to me. `VectorSupport` is an internal API for
> `jdk.incubator.vector` to consume.
> It's `jdk.incubator.vector` job to interpret the result and adapt accordingly.
Okay, I'm fine to fix it in jdk/incubator/vector/VectorShape.java if we don't
keep something like that.
For the updated fix, the {max/preferred} shape will be initialized as
shape-64-bit if hotspot doesn't support vectorization.
Testing:
- jdk/incubator/vector with MaxVectorSize=default/8/4 on Linux/x64
- jdk/incubator/vector without C2 on Linux/x64
Any comments?
Thanks.
-
PR: https://git.openjdk.java.net/jdk/pull/2722
Re: RFR: 8262096: Vector API fails to work due to VectorShape initialization exception [v4]
On Fri, 26 Feb 2021 15:37:08 GMT, Jie Fu wrote: > I'd like to keep DoubleVector.SPECIES_PREFERRED.length() <= > VectorSupport.getMaxLaneCount(double.class) for Java programmers since the > VectorSupport_GetMaxLaneCount is used to implement a Java API. It doesn't make much sense to me. `VectorSupport` is an internal API for `jdk.incubator.vector` to consume. It's `jdk.incubator.vector` job to interpret the result and adapt accordingly. - PR: https://git.openjdk.java.net/jdk/pull/2722
Re: RFR: 8262096: Vector API fails to work due to VectorShape initialization exception [v4]
On Fri, 26 Feb 2021 13:55:15 GMT, Vladimir Ivanov wrote: > IMO the fix should be in > `src/jdk.incubator.vector/share/classes/jdk/incubator/vector/VectorShape.java`. > > JVM does the right job when it signals vector support is absent (by returning > `-1`). > > `jdk.incubator.vector` implementation should take that into account and > choose a preferred shape for pure Java execution mode. Hi @iwanowww , Thanks for your review. >From the view of C2 compiler, you are right. But the Java programmer may be confused if we got something like DoubleVector.SPECIES_PREFERRED.length() > VectorSupport.getMaxLaneCount(double.class). I'd like to keep DoubleVector.SPECIES_PREFERRED.length() <= VectorSupport.getMaxLaneCount(double.class) for Java programmers since the VectorSupport_GetMaxLaneCount is used to implement a Java API. What do you think? Thanks. - PR: https://git.openjdk.java.net/jdk/pull/2722
Re: RFR: 8262096: Vector API fails to work due to VectorShape initialization exception [v4]
On Fri, 26 Feb 2021 02:38:00 GMT, Jie Fu wrote:
>> Hi all,
>>
>> Vector API fails to work when:
>> - case 1: MaxVectorSize is set to <=8, or
>> - case 2: C2 is disabled
>>
>> The reason is that {max/preferred} VectorShape initialization fails in both
>> cases.
>> And the root cause is that VectorSupport_GetMaxLaneCount [1] returns
>> unreasonable values (0 for case 1 and -1 for case 2).
>>
>> Vector API should not depend on C2 to run.
>> It should work even there is no JIT compiler since it's a Java-level api.
>> So let's fix it.
>>
>> Testing:
>> - jdk/incubator/vector with -XX:MaxVectorSize=default/8 on Linux/x64
>>
>> Thanks.
>> Best regards,
>> Jie
>>
>> [1]
>> https://github.com/openjdk/jdk/blob/master/src/hotspot/share/prims/vectorSupport.cpp#L364
>
> Jie Fu has updated the pull request incrementally with one additional commit
> since the last revision:
>
> Remove -XX:TieredStopAtLevel=3
IMO the fix should be in
`src/jdk.incubator.vector/share/classes/jdk/incubator/vector/VectorShape.java`.
JVM does the right job when it signals vector support is absent (by returning
`-1`).
`jdk.incubator.vector` implementation should take that into account and choose
a preferred shape for pure Java execution mode.
-
Changes requested by vlivanov (Reviewer).
PR: https://git.openjdk.java.net/jdk/pull/2722
Re: RFR: 8262096: Vector API fails to work due to VectorShape initialization exception [v4]
On Fri, 26 Feb 2021 02:38:00 GMT, Jie Fu wrote:
>> Hi all,
>>
>> Vector API fails to work when:
>> - case 1: MaxVectorSize is set to <=8, or
>> - case 2: C2 is disabled
>>
>> The reason is that {max/preferred} VectorShape initialization fails in both
>> cases.
>> And the root cause is that VectorSupport_GetMaxLaneCount [1] returns
>> unreasonable values (0 for case 1 and -1 for case 2).
>>
>> Vector API should not depend on C2 to run.
>> It should work even there is no JIT compiler since it's a Java-level api.
>> So let's fix it.
>>
>> Testing:
>> - jdk/incubator/vector with -XX:MaxVectorSize=default/8 on Linux/x64
>>
>> Thanks.
>> Best regards,
>> Jie
>>
>> [1]
>> https://github.com/openjdk/jdk/blob/master/src/hotspot/share/prims/vectorSupport.cpp#L364
>
> Jie Fu has updated the pull request incrementally with one additional commit
> since the last revision:
>
> Remove -XX:TieredStopAtLevel=3
Marked as reviewed by psandoz (Reviewer).
-
PR: https://git.openjdk.java.net/jdk/pull/2722
Re: RFR: 8262096: Vector API fails to work due to VectorShape initialization exception [v4]
> Hi all,
>
> Vector API fails to work when:
> - case 1: MaxVectorSize is set to <=8, or
> - case 2: C2 is disabled
>
> The reason is that {max/preferred} VectorShape initialization fails in both
> cases.
> And the root cause is that VectorSupport_GetMaxLaneCount [1] returns
> unreasonable values (0 for case 1 and -1 for case 2).
>
> Vector API should not depend on C2 to run.
> It should work even there is no JIT compiler since it's a Java-level api.
> So let's fix it.
>
> Testing:
> - jdk/incubator/vector with -XX:MaxVectorSize=default/8 on Linux/x64
>
> Thanks.
> Best regards,
> Jie
>
> [1]
> https://github.com/openjdk/jdk/blob/master/src/hotspot/share/prims/vectorSupport.cpp#L364
Jie Fu has updated the pull request incrementally with one additional commit
since the last revision:
Remove -XX:TieredStopAtLevel=3
-
Changes:
- all: https://git.openjdk.java.net/jdk/pull/2722/files
- new: https://git.openjdk.java.net/jdk/pull/2722/files/aa475b0a..bbe6150c
Webrevs:
- full: https://webrevs.openjdk.java.net/?repo=jdk=2722=03
- incr: https://webrevs.openjdk.java.net/?repo=jdk=2722=02-03
Stats: 1 line in 1 file changed: 0 ins; 1 del; 0 mod
Patch: https://git.openjdk.java.net/jdk/pull/2722.diff
Fetch: git fetch https://git.openjdk.java.net/jdk pull/2722/head:pull/2722
PR: https://git.openjdk.java.net/jdk/pull/2722
Re: RFR: 8262096: Vector API fails to work due to VectorShape initialization exception [v4]
On Fri, 26 Feb 2021 02:28:55 GMT, Paul Sandoz wrote: > In that case I think we can remove the execution with > `-XX:TieredStopAtLevel=x`. Fixed. Thanks. - PR: https://git.openjdk.java.net/jdk/pull/2722
