From 36a4a54b1c9204d72e405f5bbfa2a1b595ccba04 Mon Sep 17 00:00:00 2001 From: tv Date: Fri, 6 Nov 2026 20:22:03 +0100 Subject: flood dissemination --- docs/development.md | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) (limited to 'docs/development.md') diff --git a/docs/development.md b/docs/development.md index 9296e8b..ed95bf0 100644 --- a/docs/development.md +++ b/docs/development.md @@ -30,10 +30,11 @@ unless `BENCH_WRITES` says otherwise (more, to see what grows with them); each scenario run several times, one run after another, with `BENCH_REPEAT=3`, and a summary of each workload's mean delay and messages per write, run by run (runs differ: one can't tell a small change from chance); with -`SIM_DISSEMINATION=plumtree`, every simulated node passes changes on by PlumTree -(`--dissemination`), the benchmark's views of the trees still being the writers' -trees'; with `SIM_TOPOLOGY=hyparview`, every simulated node chooses its neighbours by -HyParView (`--topology`; with PlumTree, as writers' trees need the shared graph); all at once with +`SIM_DISSEMINATION=plumtree` (or `flood`), every simulated node passes changes on by +PlumTree (or by flooding: `--dissemination`), the benchmark's views of the trees still +being the writers' trees'; with `SIM_TOPOLOGY=hyparview`, every simulated node chooses +its neighbours by HyParView (`--topology`; with PlumTree unless `SIM_DISSEMINATION` +says otherwise, as `--dissemination` defaults to then); all at once with `BENCH_PARALLEL=1`, faster on many cores, though sharing them makes delays come out somewhat higher; and in `bench_bridge`, large values from a writer on a fast transport through a bridge to a node on a slow one: what that node gets as sync messages and as -- cgit v1.3.1