RFR: 8289552: Make intrinsic conversions between bit representations of half precision values and floats [v8]

[Smita Kamath]

On Wed, 21 Sep 2022 17:49:51 GMT, Jatin Bhateja <jbhateja at openjdk.org> wrote:

>> Smita Kamath has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   Addressed review comments
>
> src/hotspot/share/opto/convertnode.cpp line 166:
> 
>> 164: //=============================================================================
>> 165: //------------------------------Value------------------------------------------
>> 166: const Type* ConvF2HFNode::Value(PhaseGVN* phase) const {
> 
> IR framework based test will compliment newly introduced IR nodes.

Will add IR framework based test in addition to the gtest.

> src/hotspot/share/opto/convertnode.hpp line 107:
> 
>> 105: class ConvF2HFNode : public Node {
>> 106:   public:
>> 107:   ConvF2HFNode( Node *in1 ) : Node(0,in1) {}
> 
> Additional space b/w , and in1

Hi Jatin, this spacing is similar to other class declarations in the file. So I will keep it consistent. Thanks.

> src/hotspot/share/opto/convertnode.hpp line 146:
> 
>> 144: class ConvHF2FNode : public Node {
>> 145:   public:
>> 146:   ConvHF2FNode( Node *in1 ) : Node(0,in1) {}
> 
> Space b/w , and in1

Same reasoning as above.

> src/hotspot/share/runtime/sharedRuntime.cpp line 452:
> 
>> 450: // Reference implementation at src/java.base/share/classes/java/lang/Float.java:floatToFloat16
>> 451: JRT_LEAF(jshort, SharedRuntime::f2hf(jfloat  x))
>> 452:   jint doppel = SharedRuntime::f2i(x);
> 
> Newly added constant value computation runtime routines can be validated by a gtest.

I have created a bug entry for the gtest. Will add it in a separate PR.

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

PR: https://git.openjdk.org/jdk/pull/9781