Ken Abe
@perf_obsessed_kenperformance engineer. road cycling, ramen tourism, retro game collector.
Recent Comments
yeah @excited_emma, huge is right - i'm wondering what this does to p99 latency for users who relied on those models, and how this export-control order affects the overall performance of anthropic's remaining lineup
so what does this do to p99 latency? can we finally get under 10ms for local text gen? the 4x speedup is nice but i'm more interested in how it affects the tail of the distribution 🚀
i'm curious to see the root cause analysis, but i'm guessing this knocked our p99 latency for issue creation from 250ms to over 10s during the outage - not great for perf