Scriptundo

[sedited]

A refactor that moves the responsibility of managing coins from ConnectBlock to ConnectTip. Coins are now pre-fetched into the CBlockUndo in ConnectTip and then used in ConnectBlock. Down the road, this allows block validation without having a coins view present, which might be useful for non-assumevalid swiftsync, UTreeXO, and a sans-IO kernel library API.

[l0rinc]

@willcl-ark, is it still necessary to rebase on top of here 0c50e4c5846417224c96965fda3f7a9264721707?

[github-actions]

📊 Benchmark results for this run (14490747253) will be available at: https://bitcoin-dev-tools.github.io/benchcoin/results/pr-162/14490747253/index.html after the github pages "build and deployment" action has completed.
🚀 Speedups: mainnet-default-uninstrumented: 1.4%, mainnet-large-uninstrumented: 1.8%

[github-actions]

📊 Benchmark results for this run (14499155218) will be available at: https://bitcoin-dev-tools.github.io/benchcoin/results/pr-162/14499155218/index.html after the github pages "build and deployment" action has completed.
🚀 Speedups: mainnet-default-uninstrumented: -0.1%, mainnet-large-uninstrumented: -0.1%

[l0rinc]

quickly ran a reindex-chainstate IBD (single run, so may not be that stable), but it does indicate some measurable speedup - I'll bench it properly once it's pushed to where you pushed originally :p

COMMITS="817edfb21e4fa96289d69d4e1dbabb6b9ef9d8f5 168fa6702676e9c4a46f61d86807b63d69567e85"; \
STOP_HEIGHT=888888; DBCACHE=45000; \
CC=gcc; CXX=g++; \
BASE_DIR="/mnt/my_storage"; DATA_DIR="$BASE_DIR/BitcoinData"; LOG_DIR="$BASE_DIR/logs"; \
(for c in $COMMITS; do git fetch origin $c -q && git log -1 --pretty=format:'%h %s' $c || exit 1; done) && \
hyperfine \
  --sort 'command' \
  --runs 1 \
  --export-json "$BASE_DIR/rdx-${COMMITS// /-}-$STOP_HEIGHT-$DBCACHE-$CC.json" \
  --parameter-list COMMIT ${COMMITS// /,} \
  --prepare "killall bitcoind; rm -f $DATA_DIR/debug.log; git checkout {COMMIT}; git clean -fxd; git reset --hard; \
    cmake -B build -DCMAKE_BUILD_TYPE=Release -DENABLE_WALLET=OFF && cmake --build build -j$(nproc) --target bitcoind && \
    ./build/bin/bitcoind -datadir=$DATA_DIR -stopatheight=$STOP_HEIGHT -dbcache=5000 -printtoconsole=0; sleep 100" \
  --cleanup "cp $DATA_DIR/debug.log $LOG_DIR/debug-{COMMIT}-$(date +%s).log" \
  "COMPILER=$CC ./build/bin/bitcoind -datadir=$DATA_DIR -stopatheight=$STOP_HEIGHT -dbcache=$DBCACHE -blocksonly -reindex-chainstate -connect=0 -printtoconsole=0"

817edfb Merge bitcoin#32245: doc: Updates how to reproduce fuzz CI failure locally
168fa67 Use undo data in ConnectBlock

Benchmark 1: COMPILER=gcc ./build/bin/bitcoind -datadir=/mnt/my_storage/BitcoinData -stopatheight=888888 -dbcache=45000 -blocksonly -reindex-chainstate -connect=0 -printtoconsole=0 (COMMIT = 817edfb21e4fa96289d69d4e1dbabb6b9ef9d8f5)
  Time (abs ≡):        16777.027 s               [User: 20534.671 s, System: 677.014 s]
 
Benchmark 2: COMPILER=gcc ./build/bin/bitcoind -datadir=/mnt/my_storage/BitcoinData -stopatheight=888888 -dbcache=45000 -blocksonly -reindex-chainstate -connect=0 -printtoconsole=0 (COMMIT = 168fa6702676e9c4a46f61d86807b63d69567e85)
  Time (abs ≡):        16406.915 s               [User: 20119.624 s, System: 657.841 s]

Relative speed comparison
        1.02          COMPILER=gcc ./build/bin/bitcoind -datadir=/mnt/my_storage/BitcoinData -stopatheight=888888 -dbcache=45000 -blocksonly -reindex-chainstate -connect=0 -printtoconsole=0 (COMMIT = 817edfb21e4fa96289d69d4e1dbabb6b9ef9d8f5)
        1.00          COMPILER=gcc ./build/bin/bitcoind -datadir=/mnt/my_storage/BitcoinData -stopatheight=888888 -dbcache=45000 -blocksonly -reindex-chainstate -connect=0 -printtoconsole=0 (COMMIT = 168fa6702676e9c4a46f61d86807b63d69567e85)

i.e. 2.2% faster reindex-chainstate.

[github-actions]

📊 Benchmark results for this run (14548030742) will be available at: https://bitcoin-dev-tools.github.io/benchcoin/results/pr-162/14548030742/index.html after the github pages "build and deployment" action has completed.
🚀 Speedups: mainnet-default-uninstrumented: -1.1%, mainnet-large-uninstrumented: 0.1%