Each result now has the GPU's memory over the baseline after loading, the
first image, each warm-up pass and each timed pass (with the pass's peak), in
the output and as trace= in the results file, so growth across passes (a leak)
shows apart from a working set. -ocr-workers sets OCR's GPU worker threads, and
-ort-log info prints onnxruntime's arena extensions: what found the cause of
0.1.24's memory fix.
bin/gpu-bench times every GPU model (OCR mobile and server, the label
image encoder, segmentation; optionally the CPU and the label text
encoder) over a fixed 30-image corpus in tools/gpu_bench/corpus. Every
corpus file is checked against corpus.txt's SHA-256 before anything
runs; each model's files and models.config entry are fingerprinted;
the GPU, driver, power limit, CUDA runtime, cuDNN and onnxruntime are
recorded, and NVML samples memory, load, power and clocks during the
timed passes. Results go to bin/gpu-bench-results as text files.
The GPU Bench button opens a window that runs gpu-bench in its own
process (no console) and lays results files from several machines side
by side against a baseline, marking a different corpus or model.
