mirror of
https://github.com/bitcoin/bitcoin.git
synced 2026-02-11 16:11:29 +00:00
5878f35446
248b6a27c351690d3596711cc36b8102977adeab optimization: peel align-head and unroll body to 64 bytes (Lőrinc) e7114fc6dc3488c2584d42779ff2b102e4d1db99 optimization: migrate fixed-size obfuscation from `std::vector<std::byte>` to `uint64_t` (Lőrinc) 478d40afc6faaca47b5cf94bb461692d03347599 refactor: encapsulate `vector`/`array` keys into `Obfuscation` (Lőrinc) 377aab8e5a8da2ea20383b4dde59094cc42d3407 refactor: move `util::Xor` to `Obfuscation().Xor` (Lőrinc) fa5d296e3beb312e2bc39532a12bcdf187c6da91 refactor: prepare mempool_persist for obfuscation key change (Lőrinc) 6bbf2d9311b47a8a15c17d9fe11828ee623d98e0 refactor: prepare `DBWrapper` for obfuscation key change (Lőrinc) 0b8bec8aa6260c499c2663ab7a1c905da0d312c3 scripted-diff: unify xor-vs-obfuscation nomenclature (Lőrinc) 972697976c027b5199150a98e886c199b7ffc335 bench: make ObfuscationBench more representative (Lőrinc) 618a30e326e9bcfd72e0e2645ce49f8b2a88714d test: compare util::Xor with randomized inputs against simple impl (Lőrinc) a5141cd39ecbd3a7bbae5bf2f755cc5aa7c41da3 test: make sure dbwrapper obfuscation key is never obfuscated (Lőrinc) 54ab0bd64c3657f2f9be83ba587a0811e0dd8ca1 refactor: commit to 8 byte obfuscation keys (Lőrinc) 7aa557a37b73df264afffdc7c00fba47a339aee0 random: add fixed-size `std::array` generation (Lőrinc) Pull request description: This change is part of [[IBD] - Tracking PR for speeding up Initial Block Download](https://github.com/bitcoin/bitcoin/pull/32043) ### Summary Current block obfuscations are done byte-by-byte, this PR batches them to 64 bit primitives to speed up obfuscating bigger memory batches. This is especially relevant now that https://github.com/bitcoin/bitcoin/pull/31551 was merged, having bigger obfuscatable chunks. Since this obfuscation is optional, the speedup measured here depends on whether it's a [random value](https://github.com/bitcoin/bitcoin/pull/31144#issuecomment-2523295114) or [completely turned off](https://github.com/bitcoin/bitcoin/pull/31144#issuecomment-2519764142) (i.e. XOR-ing with 0). ### Changes in testing, benchmarking and implementation * Added new tests comparing randomized inputs against a trivial implementation and performing roundtrip checks with random chunks. * Migrated `std::vector<std::byte>(8)` keys to plain `uint64_t`; * Process unaligned bytes separately and unroll body to 64 bytes. ### Assembly Memory alignment is enforced by a small peel-loop (`std::memcpy` is optimized out on tested platform), with an `std::assume_aligned<8>` check, see the Godbolt listing at https://godbolt.org/z/59EMv7h6Y for details <details> <summary>Details</summary> Target & Compiler | Stride (per hot-loop iter) | Main operation(s) in loop | Effective XORs / iter -- | -- | -- | -- Clang x86-64 (trunk) | 64 bytes | 4 × movdqu → pxor → store | 8 × 64-bit GCC x86-64 (trunk) | 64 bytes | 4 × movdqu/pxor sequence, enabled by 8-way unroll | 8 × 64-bit GCC RV32 (trunk) | 8 bytes | copy 8 B to temp → 2 × 32-bit XOR → copy back | 1 × 64-bit (as 2 × 32-bit) GCC s390x (big-endian 14.2) | 64 bytes | 8 × XC (mem-mem 8-B XOR) with key cached on stack | 8 × 64-bit </details> ### Endianness The only endianness issue was with bit rotation, intended to realign the key if obfuscation halted before full key consumption. Elsewhere, memory is read, processed, and written back in the same endianness, preserving byte order. Since CI lacks a big-endian machine, testing was done locally via Docker. <details> <summary>Details</summary> ```bash brew install podman pigz softwareupdate --install-rosetta podman machine init podman machine start docker run --platform linux/s390x -it ubuntu:latest /bin/bash apt update && apt install -y git build-essential cmake ccache pkg-config libevent-dev libboost-dev libssl-dev libsqlite3-dev python3 && \ cd /mnt && git clone --depth=1 https://github.com/bitcoin/bitcoin.git && cd bitcoin && git remote add l0rinc https://github.com/l0rinc/bitcoin.git && git fetch --all && git checkout l0rinc/optimize-xor && \ cmake -B build && cmake --build build --target test_bitcoin -j$(nproc) && \ ./build/bin/test_bitcoin --run_test=streams_tests ``` </details> ### Measurements (micro benchmarks and full IBDs) > cmake -B build -DBUILD_BENCH=ON -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=gcc/clang -DCMAKE_CXX_COMPILER=g++/clang++ && \ cmake --build build -j$(nproc) && \ build/bin/bench_bitcoin -filter='ObfuscationBench' -min-time=5000 <details> <summary>GNU 14.2.0</summary> > Before: | ns/byte | byte/s | err% | ins/byte | cyc/byte | IPC | bra/byte | miss% | total | benchmark |--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|--------:|----------:|:---------- | 0.84 | 1,184,138,235.64 | 0.0% | 9.01 | 3.03 | 2.971 | 1.00 | 0.1% | 5.50 | `ObfuscationBench` > After (first optimizing commit): | ns/byte | byte/s | err% | ins/byte | cyc/byte | IPC | bra/byte | miss% | total | benchmark |--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|--------:|----------:|:---------- | 0.04 | 28,365,698,819.44 | 0.0% | 0.34 | 0.13 | 2.714 | 0.07 | 0.0% | 5.33 | `ObfuscationBench` > and (second optimizing commit): | ns/byte | byte/s | err% | ins/byte | cyc/byte | IPC | bra/byte | miss% | total | benchmark |--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|--------:|----------:|:---------- | 0.03 | 32,464,658,919.11 | 0.0% | 0.50 | 0.11 | 4.474 | 0.08 | 0.0% | 5.29 | `ObfuscationBench` </details> <details> <summary>Clang 20.1.7</summary> > Before: | ns/byte | byte/s | err% | ins/byte | cyc/byte | IPC | bra/byte | miss% | total | benchmark |--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|--------:|----------:|:---------- | 0.89 | 1,124,087,330.23 | 0.1% | 6.52 | 3.20 | 2.041 | 0.50 | 0.2% | 5.50 | `ObfuscationBench` > After (first optimizing commit): | ns/byte | byte/s | err% | ins/byte | cyc/byte | IPC | bra/byte | miss% | total | benchmark |--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|--------:|----------:|:---------- | 0.08 | 13,012,464,203.00 | 0.0% | 0.65 | 0.28 | 2.338 | 0.13 | 0.8% | 5.50 | `ObfuscationBench` > and (second optimizing commit): | ns/byte | byte/s | err% | ins/byte | cyc/byte | IPC | bra/byte | miss% | total | benchmark |--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|--------:|----------:|:---------- | 0.02 | 41,231,547,045.17 | 0.0% | 0.30 | 0.09 | 3.463 | 0.02 | 0.0% | 5.47 | `ObfuscationBench` </details> i.e. 27.4x faster obfuscation with GCC, 36.7x faster with Clang For other benchmark speedups see https://corecheck.dev/bitcoin/bitcoin/pulls/31144 ------ Running an IBD until 888888 blocks reveals a 4% speedup. <details> <summary>Details</summary> SSD: ```bash COMMITS="8324a00bd4a6a5291c841f2d01162d8a014ddb02 5ddfd31b4158a89b0007cfb2be970c03d9278525"; \ STOP_HEIGHT=888888; DBCACHE=1000; \ 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/ibd-${COMMITS// /-}-$STOP_HEIGHT-$DBCACHE-$CC.json" \ --parameter-list COMMIT ${COMMITS// /,} \ --prepare "killall bitcoind; rm -rf $DATA_DIR/*; git checkout {COMMIT}; git clean -fxd; git reset --hard; \ cmake -B build -DCMAKE_BUILD_TYPE=Release -DENABLE_WALLET=OFF && \ cmake --build build -j$(nproc) --target bitcoind && \ ./build/bin/bitcoind -datadir=$DATA_DIR -stopatheight=1 -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 -printtoconsole=0" ``` > 8324a00bd4 test: Compare util::Xor with randomized inputs against simple impl > 5ddfd31b41 optimization: Xor 64 bits together instead of byte-by-byte ```python Benchmark 1: COMPILER=gcc ./build/bin/bitcoind -datadir=/mnt/my_storage/BitcoinData -stopatheight=888888 -dbcache=1000 -blocksonly -printtoconsole=0 (COMMIT = 8324a00bd4a6a5291c841f2d01162d8a014ddb02) Time (abs ≡): 25033.413 s [User: 33953.984 s, System: 2613.604 s] Benchmark 2: COMPILER=gcc ./build/bin/bitcoind -datadir=/mnt/my_storage/BitcoinData -stopatheight=888888 -dbcache=1000 -blocksonly -printtoconsole=0 (COMMIT = 5ddfd31b4158a89b0007cfb2be970c03d9278525) Time (abs ≡): 24110.710 s [User: 33389.536 s, System: 2660.292 s] Relative speed comparison 1.04 COMPILER=gcc ./build/bin/bitcoind -datadir=/mnt/my_storage/BitcoinData -stopatheight=888888 -dbcache=1000 -blocksonly -printtoconsole=0 (COMMIT = 8324a00bd4a6a5291c841f2d01162d8a014ddb02) 1.00 COMPILER=gcc ./build/bin/bitcoind -datadir=/mnt/my_storage/BitcoinData -stopatheight=888888 -dbcache=1000 -blocksonly -printtoconsole=0 (COMMIT = 5ddfd31b4158a89b0007cfb2be970c03d9278525) ``` > HDD: ```bash COMMITS="71eb6eaa740ad0b28737e90e59b89a8e951d90d9 46854038e7984b599d25640de26d4680e62caba7"; \ STOP_HEIGHT=888888; DBCACHE=4500; \ 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' $c || exit 1; done) && \ hyperfine \ --sort 'command' \ --runs 2 \ --export-json "$BASE_DIR/ibd-${COMMITS// /-}-$STOP_HEIGHT-$DBCACHE-$CC.json" \ --parameter-list COMMIT ${COMMITS// /,} \ --prepare "killall bitcoind; rm -rf $DATA_DIR/*; git checkout {COMMIT}; git clean -fxd; git reset --hard; \ cmake -B build -DCMAKE_BUILD_TYPE=Release -DENABLE_WALLET=OFF && cmake --build build -j$(nproc) --target bitcoind && \ ./build/bin/bitcoind -datadir=$DATA_DIR -stopatheight=1 -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 -printtoconsole=0" ``` > 71eb6eaa74 test: compare util::Xor with randomized inputs against simple impl > 46854038e7 optimization: migrate fixed-size obfuscation from `std::vector<std::byte>` to `uint64_t` ```python Benchmark 1: COMPILER=gcc ./build/bin/bitcoind -datadir=/mnt/my_storage/BitcoinData -stopatheight=888888 -dbcache=4500 -blocksonly -printtoconsole=0 (COMMIT = 71eb6eaa740ad0b28737e90e59b89a8e951d90d9) Time (mean ± σ): 37676.293 s ± 83.100 s [User: 36900.535 s, System: 2220.382 s] Range (min … max): 37617.533 s … 37735.053 s 2 runs Benchmark 2: COMPILER=gcc ./build/bin/bitcoind -datadir=/mnt/my_storage/BitcoinData -stopatheight=888888 -dbcache=4500 -blocksonly -printtoconsole=0 (COMMIT = 46854038e7984b599d25640de26d4680e62caba7) Time (mean ± σ): 36181.287 s ± 195.248 s [User: 34962.822 s, System: 1988.614 s] Range (min … max): 36043.226 s … 36319.349 s 2 runs Relative speed comparison 1.04 ± 0.01 COMPILER=gcc ./build/bin/bitcoind -datadir=/mnt/my_storage/BitcoinData -stopatheight=888888 -dbcache=4500 -blocksonly -printtoconsole=0 (COMMIT = 71eb6eaa740ad0b28737e90e59b89a8e951d90d9) 1.00 COMPILER=gcc ./build/bin/bitcoind -datadir=/mnt/my_storage/BitcoinData -stopatheight=888888 -dbcache=4500 -blocksonly -printtoconsole=0 (COMMIT = 46854038e7984b599d25640de26d4680e62caba7) ``` </details> ACKs for top commit: achow101: ACK 248b6a27c351690d3596711cc36b8102977adeab maflcko: review ACK 248b6a27c351690d3596711cc36b8102977adeab 🎻 ryanofsky: Code review ACK 248b6a27c351690d3596711cc36b8102977adeab. Looks good! Thanks for adapting this and considering all the suggestions. I did leave more comments below but non are important and this looks good as-is Tree-SHA512: ef541cd8a1f1dc504613c4eaa708202e32ae5ac86f9c875e03bcdd6357121f6af0860ef83d513c473efa5445b701e59439d416effae1085a559716b0fd45ecd6