Core mechanism
The generator emits class-file instructions corresponding to tape nodes, loads inputs, evaluates primitives and stores outputs or adjoints. The JVM then applies its own optimisation and compilation pipeline.
Engine Internals & GPU Performance
Inspect the Tape Behind a Generated Kernel
TeaVM records the actual differentiable Java tape and displays its operations. This browser run does not emit a JVM class file; the Learn JVM example explains that separate cpu-jit step.
Market
This exact source runs in TeaVM. Form changes update its Java literals and reset manual edits.
import com.nablatensor.engine.ADouble;
import com.nablatensor.engine.AadRecorder;
import com.nablatensor.engine.AadTape;
import com.nablatensor.quant.EquityMarket;
public final class BytecodeKernelRiskStudio {
public static void main(String[] args) {
double maturity = 1;
int steps = 1;
EquityMarket market = EquityMarket.of()
.spot(100)
.strike(100)
.vol(.2)
.rate(.03)
.maturity(maturity)
.build();
AadTape tape = AadRecorder.record(rec -> {
ADouble spot = rec.input("spot", market.spot()); ADouble strike = rec.input("strike", market.strike()); ADouble vol = rec.input("vol",
market.vol()); ADouble rate = rec.input("rate", market.rate()); ADouble t = rec.input("maturity",
market.maturity()); ADouble dt = t.div(steps); ADouble drift = rate.sub(vol.mul(vol)
.mul(.5))
.mul(dt); ADouble diffusion = vol.mul(dt.sqrt()); ADouble terminal = spot; for (int i = 0; i < steps; i++) terminal = terminal.mul(drift.add(diffusion.mul(rec.randn()))
.exp()); rec.output(terminal.sub(strike)
.max(0)
.mul(rate.neg()
.mul(t)
.exp()));
});
int active = 0;
for (int i = 0; i < tape.size(); i++) {
if (tape.isActive(i)) active++;
String op = tape.op(i)
.name();
int a = tape.argA(i), b = tape.argB(i);
System.out.println("NODE|" + i + "|" + op + "|" + a + "|" + b + "|" + tape.isActive(i));
}
System.out.println("RESULT|" + tape.size() + "|" + active + "|" + steps + "|" + maturity);
}
}
TeaVM compiles and runs the Java source above in this browser.
Implementation guide
Specialising a tape into JVM bytecode
JIT-oriented replay can generate a small class specialised to the recorded graph, avoiding interpretation overhead on repeated evaluations.
Practical workflow
Inspect generated size and instruction mix, separate first-call warm-up from steady-state timing, keep a scalar reference for parity, and profile representative tapes rather than toy expressions.
Scope and review point
Generated bytecode is an implementation technique, not a guaranteed speedup. JVM version, workload shape and warm-up behaviour affect the result.