RFR: 8317453: NMT: Performance benchmarks are needed to measure speed and memory [v4]

[Thomas Stuefe]

On Fri, 17 Jan 2025 16:27:40 GMT, Gerard Ziemski <gziemski at openjdk.org> wrote:

> I'm really not sure what this is actually trying to benchmark - who will be collecting this data and analysing it and what will they do with their results?

+1. Gerard, could provide a description of what this benchmark does, without having to analyze the patch? Neither text nor jbs text seem to explain that. How does this replay mechanism work, and why no simpler alternative?

Basic first question, what exactly is replayed? the recordings, or the recordings + allocations themselves? If the former, why not just measure NMT overhead by comparing NMT on with off?

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

PR Comment: https://git.openjdk.org/jdk/pull/23115#issuecomment-2599767212