RFR: 8261671: X86 I2L conversion can be skipped for certain masked positive values [v5]

[Marcus G K Williams]

On Thu, 4 Mar 2021 19:52:37 GMT, Vladimir Kozlov <kvn at openjdk.org> wrote:

>> Marcus G K Williams has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   Clean up Expr.java & assembler_x86.hpp
>>   
>>   Remove unneeded bzhi in assembler_x86.hpp
>>   and unused longExpr's in Expr.java
>>   
>>   Signed-off-by: Marcus G K Williams <marcus.williams at intel.com>
>
> test/hotspot/jtreg/compiler/codegen/BMI2.java line 45:
> 
>> 43:             long z = BMITests.bzhiI2L(ix, i);
>> 44: 
>> 45:             for (int i2 = 0; i2 < ITERATIONS; i2++) {
> 
> I know that you follow BMI.java code example but we need to update these tests. Since you are writing new test you may do it right. See discussion for https://github.com/openjdk/jdk/pull/2811
> 
> And, please, don't use 1000,000 iterations. C2 compilation is triggered with 10,000. So 50,000 would be more than enough.

Thanks. I will have an update that lowers iterations. I've read through #2811 and will see what I can apply to BMI2.java tests.

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

PR: https://git.openjdk.java.net/jdk/pull/2590