kyuz0 Strix Halo 8060S Ministral-3-14B-Instruct-2512-BF16 ROCm 7.2.3 64k tg32

kyuz0 amd-strix-halo-toolboxes llama-bench sweep (results.json generated 2026-05-18, Fedora 43 kernel 6.19.12-200, llama.cpp build 1a68ec937 / b9193). ROCm 7.2.3, -fa 1 -ngl 99 --mmap 0. Long-context cell: context depth 65536 tokens, generation of 32 tokens. ONE request, batch size 1, so the number is per-request, not aggregate across concurrent requests. Mean of the repeated runs; std 0.0 tok/s. Model file 25.16 GiB, 13.51 B params, quant BF16.

Not in the earlier kyuz0 filing, which covered only the default-context pp512/tg128 cells. Same results.json, the longctx32768 and longctx65536 rows.