RFR: Changes to simplify MU accounting

[Kelvin Nilsen]

On Wed, 10 May 2023 00:13:41 GMT, Y. Srinivas Ramakrishna <ysr at openjdk.org> wrote:

>> This represents a small part of the expand-old-on-demand branch.  By itself, this may result in some performance regression because we no longer use MU (mutator utilization) reports to guide generation sizing decisions.  On the other hand, the guidance provided by MU metrics does not seem to be a very accurate predictor of ideal generation sizes.
>
> src/hotspot/share/gc/shenandoah/shenandoahMmuTracker.hpp line 75:
> 
>> 73:   TruncatedSeq _mmu_average;
>> 74: 
>> 75:   static double process_time_seconds();
> 
> It seems like `process_time_seconds()` is no longer needed either.

Thanks.  removing from expand-old-on-demand branch.

> src/hotspot/share/gc/shenandoah/shenandoahMmuTracker.hpp line 87:
> 
>> 85:   void initialize();
>> 86: 
>> 87:   // At completion of each GC cycle (not including interrupted cycles), we invoke one of the following to record the
> 
> In the case of GC's that have been canceled, is the work quantum in service of the partial/canceled cycle just dropped on the floor? Should it intentionally be? Should it intentionally not be?

The quantum of GC utilization work that is spent on concurrent GC up to the point of interruption is totaled into the subsequent report for degenerated or full gc when that completes.  I'll make this more clear in the comment (on expand-old-on-demand branch).

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

PR Review Comment: https://git.openjdk.org/shenandoah/pull/274#discussion_r1202812896
PR Review Comment: https://git.openjdk.org/shenandoah/pull/274#discussion_r1202810730