{"event":"test_supervisor_started","pid":94012,"limit_bytes":4294967296,"start_seconds":30.0,"idle_seconds":60.0,"sample_ms":100}
{"event":"test_resource_sample","elapsed_ms":0,"usage":{"cpu_time_ns":146916,"resident_bytes":622592,"physical_bytes":213088,"peak_physical_bytes":213088,"disk_read_bytes":0,"disk_write_bytes":0}}

running 4 tests
ds4: Metal device Apple M5 Max, 128.00 GiB RAM
ds4: Metal 4 tensor API enabled for Tensor kernels
ds4: drift-patch flags hc_stable=on norm_unify=on kv_raw_f32=off rope_exp2_log2=off math_safe=off tensor_matmul=on
ds4: Qwen reference kernels: Powered by MTPLX — https://github.com/youssofal/mtplx

thread 'engine::metal::qwen_mtplx::eval::execution_tests::mtplx_eval_drains_failure_preserves_cause_and_retries_only_failed_nodes' (2307597) panicked at src/engine/metal/qwen_mtplx/eval.rs:242:37:
original primitive panic
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
test engine::metal::qwen_mtplx::eval::execution_tests::mtplx_eval_drains_failure_preserves_cause_and_retries_only_failed_nodes ... ok
ds4: Metal device Apple M5 Max, 128.00 GiB RAM
ds4: Metal 4 tensor API enabled for Tensor kernels
ds4: drift-patch flags hc_stable=on norm_unify=on kv_raw_f32=off rope_exp2_log2=off math_safe=off tensor_matmul=on
ds4: Qwen reference kernels: Powered by MTPLX — https://github.com/youssofal/mtplx
{"event":"test_resource_sample","elapsed_ms":110,"usage":{"cpu_time_ns":88764166,"resident_bytes":73383936,"physical_bytes":137184000,"peak_physical_bytes":167707392,"disk_read_bytes":589824,"disk_write_bytes":0}}
test engine::metal::qwen_mtplx::eval::execution_tests::mtplx_eval_releases_intermediates_and_preserves_descriptor_donation ... ok
ds4: Metal device Apple M5 Max, 128.00 GiB RAM
ds4: Metal 4 tensor API enabled for Tensor kernels
ds4: drift-patch flags hc_stable=on norm_unify=on kv_raw_f32=off rope_exp2_log2=off math_safe=off tensor_matmul=on
ds4: Qwen reference kernels: Powered by MTPLX — https://github.com/youssofal/mtplx
test engine::metal::qwen_mtplx::eval::execution_tests::mtplx_eval_retains_tracer_topology_without_input_donation ... ok
ds4: Metal device Apple M5 Max, 128.00 GiB RAM
ds4: Metal 4 tensor API enabled for Tensor kernels
ds4: drift-patch flags hc_stable=on norm_unify=on kv_raw_f32=off rope_exp2_log2=off math_safe=off tensor_matmul=on
ds4: Qwen reference kernels: Powered by MTPLX — https://github.com/youssofal/mtplx
test engine::metal::qwen_mtplx::eval::execution_tests::mtplx_eval_runs_cpu_gpu_siblings_and_async_continuations ... ok

test result: ok. 4 passed; 0 failed; 0 ignored; 0 measured; 513 filtered out; finished in 0.17s

{"event":"test_resource_summary","elapsed_ms":220,"peak":{"cpu_time_ns":88764166,"resident_bytes":73383936,"physical_bytes":137184000,"peak_physical_bytes":167707392,"disk_read_bytes":589824,"disk_write_bytes":0},"error":null}