feat(test): introduce functional benchmarks (BenchFramework) and introduce mempool, ZMQ, RPC and REST benchmarks

Makefile.am

@@ -252,6 +252,7 @@ dist_noinst_SCRIPTS = autogen.sh
 EXTRA_DIST = $(DIST_SHARE) $(DIST_CONTRIB) $(WINDOWS_PACKAGING) $(OSX_PACKAGING) $(BIN_CHECKS)
 
 EXTRA_DIST += \
+	test/bench \
     test/functional \
     test/fuzz
 
@@ -319,5 +320,5 @@ clean-docs:
 
 clean-local: clean-docs
 	rm -rf coverage_percent.txt test_dash.coverage/ total.coverage/ fuzz.coverage/ test/tmp/ cache/ $(OSX_APP)
-	rm -rf test/functional/__pycache__ test/functional/test_framework/__pycache__ test/cache share/rpcauth/__pycache__
+	rm -rf test/bench/__pycache__ test/functional/__pycache__ test/functional/test_framework/__pycache__ test/cache share/rpcauth/__pycache__
 	rm -rf dist/

ci/lint/04_install.sh

@@ -34,11 +34,15 @@ if [ -z "${SKIP_PYTHON_INSTALL}" ]; then
 fi
 
 # NOTE: BUMP ALSO contrib/containers/ci/ci-slim.Dockerfile
+${CI_RETRY_EXE} pip3 install aiohttp==3.13.3
 ${CI_RETRY_EXE} pip3 install codespell==2.2.1
 ${CI_RETRY_EXE} pip3 install flake8==5.0.4
+${CI_RETRY_EXE} pip3 install jinja2==3.1.6
 ${CI_RETRY_EXE} pip3 install lief==0.13.2
+${CI_RETRY_EXE} pip3 install multiprocess==0.70.19
 ${CI_RETRY_EXE} pip3 install mypy==0.981
 ${CI_RETRY_EXE} pip3 install pyzmq==24.0.1
+${CI_RETRY_EXE} pip3 install tabulate==0.10.0
 ${CI_RETRY_EXE} pip3 install vulture==2.6
 
 SHELLCHECK_VERSION=v0.8.0

configure.ac

@@ -2062,6 +2062,7 @@ AC_CONFIG_LINKS([src/.bear-tidy-config:src/.bear-tidy-config])
 AC_CONFIG_LINKS([src/.clang-tidy:src/.clang-tidy])
 AC_CONFIG_LINKS([src/ipc/.clang-tidy:src/ipc/.clang-tidy])
 AC_CONFIG_LINKS([src/test/.clang-tidy:src/test/.clang-tidy])
+AC_CONFIG_LINKS([test/bench/bench_runner.py:test/bench/bench_runner.py])
 AC_CONFIG_LINKS([test/functional/test_runner.py:test/functional/test_runner.py])
 AC_CONFIG_LINKS([test/fuzz/test_runner.py:test/fuzz/test_runner.py])
 AC_CONFIG_LINKS([test/util/test_runner.py:test/util/test_runner.py])

contrib/containers/ci/ci-slim.Dockerfile

@@ -77,13 +77,15 @@ ENV UV_SYSTEM_PYTHON=1
 # Install Python packages
 # NOTE: if versions are changed, update ci/lint/04_install.sh
 RUN uv pip install --system --break-system-packages \
+    aiohttp==3.13.3 \
     codespell==2.2.1 \
     flake8==5.0.4 \
-    jinja2 \
+    jinja2==3.1.6 \
     lief==0.13.2 \
-    multiprocess \
+    multiprocess==0.70.19 \
     mypy==0.981 \
     pyzmq==24.0.1 \
+    tabulate==0.10.0 \
     vulture==2.6
 
 # Install packages relied on by tests

src/.bear-tidy-config

@@ -1,4 +1,5 @@
 {
+  "schema": "4.0",
   "output": {
     "content": {
       "include_only_existing_source": true,

test/bench/bench_framework.py

@@ -0,0 +1,133 @@
+#!/usr/bin/env python3
+# Copyright (c) 2026 The Dash Core developers
+# Distributed under the MIT software license, see the accompanying
+# file COPYING or http://www.opensource.org/licenses/mit-license.php.
+
+import os
+import sys
+import time
+from typing import Dict, List, Optional
+
+# Allow imports from the functional test framework.
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..', 'functional'))
+
+from bench_results import (  # noqa: E402
+    BenchResult,
+    results_to_markdown,
+    save_results,
+)
+from test_framework.test_framework import BitcoinTestFramework  # noqa: E402
+
+
+class BenchFramework(BitcoinTestFramework):
+
+    def set_test_params(self) -> None:
+        """Initialise benchmark state, then delegate to ``set_bench_params``."""
+        self.warmup_iterations: int = 0
+        self.bench_iterations: int = 1
+        self.bench_name: str = type(self).__name__
+        self.results_file: Optional[str] = None
+        # Raw latency samples keyed by measurement name.
+        self._samples: Dict[str, List[float]] = {}
+        self._timer_start: Optional[float] = None
+        self.set_bench_params()
+
+    def setup_nodes(self) -> None:
+        """Merge --daemon-args into extra_args just before nodes start."""
+        raw = getattr(self.options, "daemon_args", None) or ""
+        daemon_args = raw.split() if raw else []
+        if daemon_args and self.extra_args is not None:
+            for node_args in self.extra_args:
+                node_args.extend(daemon_args)
+        super().setup_nodes()
+
+    def run_test(self) -> None:
+        """Execute warmup, timed iterations, then report."""
+        self.results_file = getattr(self.options, "results_file", None)
+        if self.warmup_iterations > 0:
+            self.log.info(
+                "Warming up (%d iteration%s)...",
+                self.warmup_iterations,
+                "s" if self.warmup_iterations != 1 else "",
+            )
+            for i in range(self.warmup_iterations):
+                self.log.debug("  warmup %d/%d", i + 1, self.warmup_iterations)
+                self.run_bench()
+                self._samples.clear()
+
+        self.log.info(
+            "Running benchmark (%d iteration%s)...",
+            self.bench_iterations,
+            "s" if self.bench_iterations != 1 else "",
+        )
+        for i in range(self.bench_iterations):
+            self.log.debug("  iteration %d/%d", i + 1, self.bench_iterations)
+            self.run_bench()
+
+        self._report_results()
+
+    def add_options(self, parser) -> None:  # type: ignore[override]
+        """Adds bench-specific args. Subclasses should call super first."""
+        parser.add_argument(
+            "--daemon-args",
+            dest="daemon_args",
+            default=None,
+            help="Extra daemon arguments as a single string "
+                 "(e.g. --daemon-args=\"-rpcworkqueue=1024 -rpcthreads=8\")",
+        )
+        parser.add_argument(
+            "--results-file",
+            dest="results_file",
+            default=None,
+            help="Save results to a JSON file",
+        )
+
+    def set_bench_params(self) -> None:
+        """Benchmarks must override this to set ``num_nodes``, etc."""
+        raise NotImplementedError
+
+    def run_bench(self) -> None:
+        """Benchmarks must override this to define the workload."""
+        raise NotImplementedError
+
+    def start_timer(self) -> None:
+        """Mark the beginning of a timed section."""
+        if self._timer_start is not None:
+            self.log.warning("start_timer() called twice without stop_timer()")
+        self._timer_start = time.perf_counter()
+
+    def stop_timer(self, name: str) -> float:
+        """Record elapsed time (ms) since the last ``start_timer()`` call, returns in ms."""
+        if self._timer_start is None:
+            raise RuntimeError("stop_timer() called without start_timer()")
+        elapsed_ms = (time.perf_counter() - self._timer_start) * 1000.0
+        self._timer_start = None
+        self._samples.setdefault(name, []).append(elapsed_ms)
+        return elapsed_ms
+
+    def record_sample(self, name: str, value_ms: float) -> None:
+        """Directly record a latency sample (ms) without using the timer."""
+        self._samples.setdefault(name, []).append(value_ms)
+
+    def _build_results(self) -> List[BenchResult]:
+        """Convert raw samples into a list of ``BenchResult`` objects."""
+        return [
+            BenchResult.from_samples(name, samples)
+            for name, samples in self._samples.items()
+            if samples
+        ]
+
+    def _report_results(self) -> None:
+        """Print a summary of all recorded measurements."""
+        results = self._build_results()
+        md = results_to_markdown(results, title=self.bench_name)
+        print(md)
+
+        if self.results_file:
+            save_results(results, self.results_file, label=self.bench_name)
+            self.log.info("Results saved to %s", self.results_file)
+
+    @property
+    def samples(self) -> Dict[str, List[float]]:
+        """Access the raw sample data."""
+        return self._samples