Guided, one-card-at-a-time batch checker for NVIDIA CMP 170HX cards (PCI IDs 10de:20c2 / 10de:2082) — verifies the PCIe capacitor mod, records PCIe bus health, backs up VBIOS, runs the cmpunlocker verification, and optionally stresses VRAM. Built for a 23-card run, with one folder + JSON report per card serial.

Reference harness (auto-cloned into harness/ by preflight):

170th-Street — community docs, incl. modifications/pcie-capacitor-mod.md (the pass criteria: LnkSta Width x16 after the mod, x4 when missing)
cmpunlocker — verify.sh is executed read-only per card (nothing is installed or patched by this tool)

Requirements

Linux x86-64, Python 3.8+
pciutils (lspci, setpci) — auto-installed if a package manager is present
Root (sudo) for: PCI bus rescan, link retrain, sysfs ROM read, dmesg, verify.sh
NVIDIA driver with nvidia-smi (nvidia-open 610.xx+, see cmpunlocker) for unlock verification
nvflash (not packaged — download from NVIDIA or pass --nvflash /path/to/nvflash) for VBIOS backup and serial auto-read
Optional: vkmark or vulkan-tools for the 5-minute VRAM stress test

Missing tools are reported by preflight; packages are auto-installed where possible, otherwise instructions are printed.

Usage

python3 check_cards.py               # full run, 23 cards, guided
python3 check_cards.py --preflight   # tool check / install / harness clone only
python3 check_cards.py --demo        # self-test with fake hardware (no GPU needed)
python3 check_cards.py --help        # all flags

Normal run flow, repeated per card:

Insert card N/23, then press Enter (s = skip, q = quit)
GPU auto-detected on the bus (PCI rescan retry offered if absent)
Serial auto-read from nvflash UUID; falls back to a manual label prompt
VBIOS backup → cards/<serial>/bios.rom (nvflash, sysfs ROM fallback)
PCIe checks: link speed/width, AER counters, full config-space dump
Link retrain test + full bus rescan test (device must survive both)
Kernel log scrape (dmesg/journalctl) for this BDF: Xid / AER / NVRM lines
nvidia-smi snapshot + harness/cmpunlocker/verify.sh (read-only)
Last step, optional: Run 5-min Vulkan VRAM stress test now? (y/N) — refused by default, --vulkan always|never to override
Result + report.json written; progress saved to progress.json (resumable)

Checks & scoring (lenient)

Result	Criteria
FAIL	no device detected, link down/untrained, device lost after rescan, link failed retrain, active uncorrectable AER errors
WARN	link width ≠ `x16` (capacitor mod missing/incomplete), Gen1 only (Gen2 unlock patch not active), correctable AER errors, kernel errors / Xid, VBIOS backup failed, cmpunlocker verify not `unlocked-ok`, errors during VRAM stress
PASS	everything clean

Output

cards/
└── <serial>/
    ├── bios.rom              # VBIOS backup (+ sha256 in report)
    ├── report.json           # full result: link, AER before/after, retrain,
    │                          #   rescan, dmesg, verify, vulkan, PASS/WARN/FAIL
    ├── lspci.txt             # lspci -vvv (before)
    ├── lspci_after.txt       # lspci -vvv (after retrain/rescan)
    ├── config_space.hex      # lspci -nnvvxxxx full hex dump
    ├── dmesg.log             # kernel lines related to this card
    └── verify_unlocker.log   # cmpunlocker verify.sh output
progress.json                 # which card indices are done (resume support)

Useful flags

Flag	Meaning
`--cards N` / `--start N`	total cards (default 23) / first index
`--nvflash PATH`	explicit nvflash binary
`--serial XYZ`	override serial for the next card
`--vulkan ask\|always\|never`	VRAM test policy (default: ask, last step)
`--skip-bios`	skip VBIOS backup
`--no-install` / `--no-clone`	don't auto-install packages / don't clone harness
`--restart`	ignore `progress.json` and start over
`--no-pause`	don't wait for the insert prompt

Safety

This tool only reads card state: verify.sh is run without installing anything, no modules are patched, no firmware is flashed. The only writes are local files, PCI link retrain/rescan sysfs triggers, and your own reports.