Predictor table aliasing¶

Generated workload catalogue · Execution contract

Purpose¶

Place opposite-outcome branches at a controllable PC spacing.

Setup¶

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

--branch-spacing is a power of two from 4 through 256, default 16, measured in instruction words.

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

Implementation¶

The first BEQ is always taken and skips an addition of 99. Padding instructions increment x9. The second BNE is always not taken and allows x10+=1. Finally include x9 in the checksum.

// S is the static distance in instruction words.
padding_count += S - 2;
++sum;

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

Expected result¶

Result=N*(S-1) modulo 2^32. With S=16 and N=2, result=30. The two branch PCs are 4*S bytes apart.

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¶

Spacing may map the two branches to the same predictor entry, depending on policy/indexing and --entries. The outer-loop branch can also interfere; this is a concrete collision workload, not an isolated abstract table experiment.

build/rv32i_experiment --core advanced --workload branch-alias \
  --iterations 2 --size 16 --seed 1 --branch-spacing 16 \
  --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.