RFR(S): 8202169: Reduce ctw_2 duration by parallelizing CtwRunner invocations

Vladimir Kozlov vladimir.kozlov at oracle.com
Wed Apr 25 03:15:15 UTC 2018


Good.

Thanks,
Vladimir

On 4/24/18 6:54 PM, Mikael Vidstedt wrote:
> 
> I decided to add some additional argument checking, please review:
> 
> Webrev: 
> http://cr.openjdk.java.net/~mikael/webrevs/8202169/webrev.01/open/webrev/
> Incremental webrev: 
> http://cr.openjdk.java.net/~mikael/webrevs/8202169/webrev.01.incr/open/webrev/
> 
> Cheers,
> Mikael
> 
>> On Apr 24, 2018, at 4:45 PM, Igor Ignatyev <igor.ignatyev at oracle.com 
>> <mailto:igor.ignatyev at oracle.com>> wrote:
>>
>> Hi Mikael,
>>
>> looks good to me.
>>
>> -- Igor
>>
>>> On Apr 23, 2018, at 7:07 PM, Vladimir Kozlov 
>>> <vladimir.kozlov at oracle.com <mailto:vladimir.kozlov at oracle.com>> wrote:
>>>
>>> Looks good.
>>>
>>> Ask Igor Ignatyev to look as author of code.
>>>
>>> Thanks,
>>> Vladimir
>>>
>>> On 4/23/18 6:58 PM, Mikael Vidstedt wrote:
>>>> Please review this change which parallelizes the 
>>>> CompileTheWorld/CtwRunner invocations to reduce the (wall clock) 
>>>> duration of running the ctw_2 jtreg test group.
>>>> bug: https://bugs.openjdk.java.net/browse/JDK-8202169
>>>> webrev: 
>>>> http://cr.openjdk.java.net/~mikael/webrevs/8202169/webrev.00/open/webrev/
>>>> The CompileTheWorld class already has support for specifying the 
>>>> start and end class (index), so this change just enables passing the 
>>>> relevant arguments in to the CtwRunner wrapper class. For 
>>>> flexibility I made it possible both to specify the start/stop values 
>>>> both as absolute index, or as percentage values. This change only 
>>>> uses the latter format.
>>>> Cheers,
>>>> Mikael
>>
> 


More information about the hotspot-compiler-dev mailing list