The riser-rig 3080
The moody Radeon
The loyal M2
$ envidia
SUSPECT NVIDIA GeForce RTX 3080 (link x1 gen 1, busy 99%, temp 78°C, fan 100%)
1. Has it been passed around? link drops, resets and re-negotiations
Disappeared on you 2 times (Xid 79)NVIDIA driver error, across every boot still in the journal: GPU has fallen off the bus.
Made the motherboard repeat itself 1,482 timesPCIe replay counter: packets that arrived corrupted and were re-sent. A handful is noise; a climbing number means a marginal link (riser, slot, seating, power).
Renegotiated the relationship 3 timesReplay rollovers: the same packet failed repeatedly and the link had to retrain.
Only giving you x1 of an x16 connectionx1 is what mining risers run at. The PCIe link trained below what the card and slot support.
Made 1 scene (Xid 31)GPU memory page fault (usually an application bug).
2. Does it have a secret past? wear, firmware and signs of a mining rig
Says its VBIOS is 94.02.42.00.A9Cannot be verified offline. A version not listed for your exact card suggests a reflash, which miners do to change memory timings and power limits.
3. Is it working behind your back? who is using it right now
t-rex (pid 4242) is a known crypto miner/tmp/.cache/t-rex — if you did not start this, kill it and find out how it got there.
Seeing 2 processes right now (2 actually busy)Busiest over 2s: t-rex (98%, 9700 MiB), Xorg (1%, 120 MiB).
VERDICT Caught in 4K. This is not a misunderstanding.
Cheated on you 1,490 times as of this check.
On the record: 1,702 times since 2026-10-02 (9 boots watched).
$ envidia
SUSPECT AMD Radeon RX 6800 XT (speed 16.0 GT/s PCIe, busy 6%)
1. Has it been passed around? link drops, resets and re-negotiations
Had to be reset by the driver 2 timesamdgpu logged a GPU reset after a hang, across every boot still in the journal.
Had 14 awkward moments with the motherboardPCIe correctable errors this boot (RxErr 3, BadTLP 9, BadDLLP 2). Recovered automatically; a climbing count means a marginal link.
Only giving you x8 of an x16 connectionWrong slot, lanes shared with an M.2 drive, a riser cable, or dirty or damaged contacts.
Stopped answering the driver 2 timesamdgpu ring timeouts, each usually followed by a reset.
2. Does it have a secret past? wear, firmware and signs of a mining rig
Says its VBIOS is 113-D4120500-101Cannot be verified offline. Compare against the stock versions for your exact card; a mismatch suggests a reflash.
3. Is it working behind your back? who is using it right now
Seeing 3 processes right now (2 actually busy)Busiest over 2s: gnome-shell (4%), firefox (2%). Being used is not cheating.
VERDICT It's complicated. There are things here that deserve a conversation.
Cheated on you 19 times as of this check.
On the record: 42 times since 2026-09-18 (6 boots watched).
$ envidia
SUSPECT Apple M2 (cores 8, busy 42%)
1. Has it been passed around? link drops, resets and re-negotiations
No GPU recoveries this bootThe driver has not had to reset the GPU since startup.
2. Does it have a secret past? wear, firmware and signs of a mining rig
Built into the chip: it has only ever been with this MacAn integrated GPU has no VBIOS to reflash and cannot have been pulled from a mining rig. Its history is the Mac's history.
3. Is it working behind your back? who is using it right now
Seeing 65 processes right now (3 actually busy)Busiest over 2s: WindowServer (35%), Safari (4%), Music (2%). Being used is not cheating; anything you do not recognise in this list is worth a look.
VERDICT Faithful. Nothing on it. Go apologise for checking its logs.
Cheated on you 0 times as of this check.
The M2 report follows a real run. The other two are rendered from the project's test fixtures, not from real cards.
