Streaming copy and checksum¶

Generated workload catalogue · Execution contract

Purpose¶

Produce regular sequential loads and stores while retaining an independently checkable output image.

Setup¶

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

n seeded words at input 0x1000; output at 0x2000.

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

Implementation¶

At each outer iteration reset both pointers, copy each word, and add it to the checksum.

for (size_t i = 0; i < n; ++i) { output[i] = input[i]; sum += input[i]; }

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

Expected result¶

Every output word equals the corresponding input word, and result=N*sum(input) modulo 2^32.

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¶

With zero-wait FF memory this measures instruction scheduling and port activity, not a cache miss rate. Reuse the same image for a future memory hierarchy, while recording that new target configuration.

build/rv32i_experiment --core advanced --workload stream \
  --iterations 2 --size 16 --seed 1  \
  --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.