bench: add script verification benchmark for P2TR script-path spends #35038

1. 
   theStack commented at 3:20 pm on April 9, 2026: contributor

   Similarly as #34472 already did for key-path spends, this PR adds a benchmark for P2TR script-path spends. So far we don’t have a benchmark on master yet that exercises the verification of taproot commitments (VerifyTaprootCommitment).

   Note that the tapscript leaf intentionally includes a single OP_CHECKSIG as it likely reflects the real world best. Spending tapscript leafs without any signature checks don’t make much sense (they could be trivially tampered with and thus stolen by miners), and doing more than one signature check seems the exception rather than the rule.

   The primary motivation for this PR is to evaluate how potential secp256k1 changes in pubkey tweaking (e.g. #1843) may impact script verification performance.

2. bench: add script verification benchmark for P2TR script-path spends

   To reflect the likely most common real-world scenario, a single
   OP_CHECKSIG is contained in the Tapscript leaf.

   ba3cf01781
3. DrahtBot added the label Tests on Apr 9, 2026
4. 
   DrahtBot commented at 3:20 pm on April 9, 2026: contributor

   The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

   Reviews

   See the guideline for information on the review process.

   Type	Reviewers
   ACK	davidgumberg

   If your review is incorrectly listed, please copy-paste <!–meta-tag:bot-skip–> into the comment that the bot should ignore.

5. in src/bench/verify_script.cpp:86 in ba3cf01781

   81+        assert(false);
   82+    }()};
   83+    assert(txSpend.vin[0].scriptWitness.stack.size() == expected_wit_count);
   84+
   85     // Benchmark.
   86     bench.run([&] {
   

   ---

   ---

   

   davidgumberg commented at 1:41 am on April 10, 2026:

   nit: while touching this area:

   0    bench.unit("script").run([&] {
   

6. 
   davidgumberg commented at 1:45 am on April 10, 2026: contributor

   crACK https://github.com/bitcoin/bitcoin/pull/35038/commits/ba3cf017819177d6517292c200db7991c4371746

   0./build/bin/bench_bitcoin --filter="VerifyScript.*"
   

   ns/script	script/s	err%	ins/script	cyc/script	IPC	bra/script	miss%	total	benchmark
   20,238.11	49,411.73	0.2%	467,600.47	86,770.69	5.389	4,548.08	0.0%	0.01	VerifyScriptP2TR_KeyPath
   36,073.54	27,721.15	0.1%	846,768.85	154,685.88	5.474	9,090.15	0.2%	0.01	VerifyScriptP2TR_ScriptPath
   20,260.06	49,358.19	0.3%	459,586.47	86,832.23	5.293	5,805.08	0.1%	0.01	VerifyScriptP2WPKH

   ---

   Feel free to disregard as out-of-scope, but maybe while touching this file anyways would be nice to use new setup api introduced in #34208 for the nested if script benchmark:

    0-    bench.run([&] {
    1-        auto stack_copy = stack;
    2+    bench.unit("script").epochIterations(1).setup([&] {
    3+        stack = {};
    4+    }).run([&] {
    5         ScriptError error;
    6-        bool ret = EvalScript(stack_copy, script, 0, BaseSignatureChecker(), SigVersion::BASE, &error);
    7+        bool ret = EvalScript(stack, script, 0, BaseSignatureChecker(), SigVersion::BASE, &error);
    8         assert(ret);
    9     });
   10 }
   

   on master:

   ns/op	op/s	err%	ins/op	cyc/op	IPC	bra/op	miss%	total	benchmark
   19,890.59	50,275.02	1.3%	673,992.65	85,238.74	7.907	143,404.89	0.0%	0.01	VerifyNestedIfScript

   vs this diff:

   ns/script	script/s	err%	ins/script	cyc/script	IPC	bra/script	miss%	total	benchmark
   12,089.00	82,719.83	0.4%	375,703.00	51,987.00	7.227	69,249.00	0.2%	0.00	VerifyNestedIfScript

Contributors
theStack DrahtBot davidgumberg

Labels
Tests