dungba88 commented on PR #14226: URL: https://github.com/apache/lucene/pull/14226#issuecomment-2707997958
I tried the idea of stopping at second pass with the original k, but the benchmark results look weird for all algorithms, as if it doesn't matter at all. This is much different from the last runs. ``` recall latency_(ms) fanout quantized greediness index_s index_docs/s force_merge_s num_segments index_size_(MB) vec_disk_(MB) vec_RAM_(MB) algorithm stdev 0.959 150.588 0 no -1.00 117.12 4269.31 0.00 7 1490.07 1464.844 1464.844 prorata 3 0.959 152.532 0 no -1.00 116.56 4289.64 0.00 7 1489.97 1464.844 1464.844 prorata 6 0.956 160.835 0 no -1.00 117.71 4247.84 0.00 7 1489.95 1464.844 1464.844 prorata 9 0.960 157.832 0 no -1.00 116.22 4302.11 0.00 7 1490.00 1464.844 1464.844 prorata 16 0.960 155.564 0 no -1.00 117.92 4240.05 0.00 7 1490.03 1464.844 1464.844 optimistic 3 0.961 152.263 0 no -1.00 118.04 4235.96 0.00 7 1490.08 1464.844 1464.844 optimistic 6 0.962 158.384 0 no -1.00 118.79 4209.25 0.00 7 1490.06 1464.844 1464.844 optimistic 9 0.959 162.569 0 no -1.00 117.15 4268.14 0.00 7 1489.99 1464.844 1464.844 optimistic 16 0.956 152.338 0 no -1.00 119.60 4180.57 0.00 7 1489.98 1464.844 1464.844 simplified 3 0.960 161.406 0 no -1.00 117.71 4247.69 0.00 7 1490.12 1464.844 1464.844 simplified 6 0.960 151.540 0 no -1.00 117.11 4269.42 0.00 7 1490.02 1464.844 1464.844 simplified 9 0.959 159.400 0 no -1.00 116.79 4281.34 0.00 7 1490.10 1464.844 1464.844 simplified 16 ``` Will try to look deeper. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org