nick/amethyst
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 1 Wiki Activity

fix: update native benchmark to use Fe4 field access

Browse Source 
Fix a[0]/b[0] array indexing to a.l0/b.l0 in the unsignedMultiplyHigh
and uLt micro-benchmarks on Kotlin/Native linuxX64.

Native benchmark results (Kotlin/Native LLVM AOT, linuxX64):
  verifySchnorr:     85,402 ns/op  (11,709 ops/s)
  verifySchnorrFast: 77,231 ns/op  (12,948 ops/s)
  signSchnorr:       74,702 ns/op  (13,386 ops/s)
  sign (cached pk):  41,273 ns/op  (24,228 ops/s)
  pubKeyCreate:      32,305 ns/op  (30,954 ops/s)
  ecdhXOnly:        100,152 ns/op   (9,984 ops/s)

  Field micro-benchmarks:
  FieldP.mul:  44 ns/op    FieldP.add: 5 ns/op
  FieldP.sqr:  37 ns/op    FieldP.sub: 6 ns/op

  Batch verify: 3.1x (4 events) to 7.0x (32 events) faster

https://claude.ai/code/session_01Sxi6Gpxbstuj3Y8TBY7XrU
This commit is contained in:
Claude
2026-04-10 07:22:35 +00:00
parent a2ba64baba
commit 28b76f4320
1 changed files with 9 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
quartz/src/linuxX64Test/kotlin/com/vitorpamplona/quartz/utils/secp256k1/Secp256k1NativeBenchmark.kt
+9 -9
View File
@@ -220,35 +220,35 @@ class Secp256k1NativeBenchmark {
// out of the loop (dead code elimination risk with constant inputs).
val a = U256.fromBytes(hexToBytes("79BE667EF9DCBBAC55A06295CE870B07029BFCDB2DCE28D959F2815B16F81798"))
val b = U256.fromBytes(hexToBytes("483ADA7726A3C4655DA4FBFC0E1108A8FD17B448A68554199C47D08FFB10D4B8"))
val out = Fe4()
val w = Wide8()
val out = LongArray(4)
val w = LongArray(8)
val results = mutableListOf<BenchResult>()
// --- FieldP.mul (the hottest operation) ---
// Uses `out` as both output and next input to create a data dependency chain.
out.copyFrom(a)
a.copyInto(out)
results +=
bench("FieldP.mul", 5000, 100000) {
FieldP.mul(out, out, b, w)
}
// --- FieldP.sqr ---
out.copyFrom(a)
a.copyInto(out)
results +=
bench("FieldP.sqr", 5000, 100000) {
FieldP.sqr(out, out, w)
}
// --- FieldP.add ---
out.copyFrom(a)
a.copyInto(out)
results +=
bench("FieldP.add", 5000, 500000) {
FieldP.add(out, out, b)
}
// --- FieldP.sub ---
out.copyFrom(a)
a.copyInto(out)
results +=
bench("FieldP.sub", 5000, 500000) {
FieldP.sub(out, out, b)
@@ -264,20 +264,20 @@ class Secp256k1NativeBenchmark {
var hiSink = 0L
results +=
bench("unsignedMultiplyHigh", 10000, 1000000) {
hiSink = unsignedMultiplyHigh(a[0] xor hiSink, b[0])
hiSink = unsignedMultiplyHigh(a.l0 xor hiSink, b.l0)
}
// --- uLt (with varying input to prevent hoisting) ---
var ltSink = 0L
results +=
bench("uLt", 10000, 1000000) {
ltSink = if (uLt(a[0] xor ltSink, b[0])) 1L else 0L
ltSink = if (uLt(a.l0 xor ltSink, b.l0)) 1L else 0L
}
printResults("secp256k1 Field Micro-Benchmarks: Kotlin/Native (LLVM AOT) on linuxX64", results)
// Use sinks to prevent dead code elimination of the entire benchmark
assertTrue(hiSink != Long.MIN_VALUE || ltSink != Long.MIN_VALUE || out.l0 != Long.MIN_VALUE)
assertTrue(hiSink != Long.MIN_VALUE || ltSink != Long.MIN_VALUE || out[0] != Long.MIN_VALUE)
}
private fun hexToBytes(hex: String): ByteArray {