Controlled branch outcomes¶

Generated workload catalogue · Execution contract

Purpose¶

Provide known taken/not-taken streams for predictor-policy comparisons.

Setup¶

Property	Value
Workload ID	`branch-pattern`
Selection	`--workload branch-pattern`
Generator	`buildRV32IBenchmark() in RV32IBenchmarks.cpp`
End condition	EBREAK after the loop and result store

A 32-word outcome ring. --pattern selects taken, not-taken, alternating, bursts, or irregular; the seed makes irregular data deterministic.

Use the shared address map and parameter bounds. N includes any training iterations; n means size.

Implementation¶

Load one outcome, branch if it is one, and increment x10 only on zero. Advance the ring index modulo 32. A separate countdown branch controls the outer loop.

if (outcomes[k % 32] == 0) ++sum;

This is a behavioral summary of the generated assembly, not the host program being timed.

Expected result¶

Result equals the number of zeros among the first N ring entries. Taken gives 0; not-taken gives N; alternating begins with zero. Bursts uses eight zeros, eight ones, repeated: N=12 gives 8.

The runner checks the generator's expected result in addition to retirement-by-retirement oracle validation. Output-array assertions, where described, check more than a checksum.

Measurement and running¶

The irregular ring is deterministic, not a nondeterministic predictor test. Its RNG state follows common input generation, so keep both size and seed fixed. Body-window/training options separate cold entry from an already exercised predictor.

build/rv32i_experiment --core advanced --workload branch-pattern \
  --iterations 2 --size 16 --seed 1 --pattern bursts \
  --fidelity behavioral --period 4000 \
  --allocation-limit-mib 512 --max-cycles 4096

The explicit period is for a bounded functional check, not a hardware-speed result. For comparisons, use the serial campaign driver, identical case inputs, and a certified per-target clock.

Source and tests¶

src/analysis/RV32IBenchmarks.cpp: emitted instructions and independent expected results.
tests/cpp/RV32IBenchmarkTest.cpp: generator results, parameter bounds, and edge cases.
tests/checks/RV32IBenchmarkCLITest.py and tests/checks/BenchmarkCampaignCLITest.py: shared execution/campaign checks.