RFR: 8369467: Rdtsc: Avoid initialize_elapsed_counter when UseFastUnorderedTimeStamps will be disabled

Thu Oct 9 10:06:03 UTC 2025

On Thu, 9 Oct 2025 09:25:22 GMT, Markus Grönlund <mgronlun at openjdk.org> wrote:

> Hi Axel,
> 
> The else branch for VM_Version::supports_tscinv_ext(), i.e., the do_time_measurements() loop, is unreliable and should be removed.
> 
> Rdtsc should only be used when INV_TSC is reported by the HW.
> 
> Can you remove that piece, please?

That would be even better. It was my suggestion / question in #27710:
> Another question is whether we should support non-invariant rdtsc (even if it needs to be explicitly enabled by a user). I wonder of its usefulness when it will not be accurate, adds a startup cost, and most kernels are fast enough with much more accurate timing information.

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

PR Comment: https://git.openjdk.org/jdk/pull/27713#issuecomment-3385117563