“device inventory meets requirements and the test allocates memory without failure” addresses scheduling readiness did not confirm required device memory and capabilities, but the current hardware or infrastructure issue is interconnect readiness is not captured in GPU bring-up acceptance. The option “topology display and local bandwidth checks match the approved HGX design” is more directly tied to the required outcome.

“a sample CUDA application initializes and reports the expected device” addresses driver visibility was mistaken for full application-runtime readiness, but the current hardware or infrastructure issue is interconnect readiness is not captured in GPU bring-up acceptance. The option “topology display and local bandwidth checks match the approved HGX design” is more directly tied to the required outcome.

“nvidia-smi lists the intended GPU count with no driver initialization errors” addresses host driver registration has not established visibility of installed GPUs, but the current hardware or infrastructure issue is interconnect readiness is not captured in GPU bring-up acceptance. The option “topology display and local bandwidth checks match the approved HGX design” is more directly tied to the required outcome.

“compare BMC asset PCIe listing and nvidia-smi before driver replacement” addresses software-level symptoms were not isolated from hardware enumeration faults, but the current hardware or infrastructure issue is hardware discovery and memory inventory do not match approved BOM. The option “isolate memory slot and DIMM faults using diagnostics and compare installed capacity” is more directly tied to the required outcome.

“diagnose and replace the failing memory component before burn-in” addresses boot completion cannot justify acceptance with critical hardware errors, but the current hardware or infrastructure issue is hardware discovery and memory inventory do not match approved BOM. The option “isolate memory slot and DIMM faults using diagnostics and compare installed capacity” is more directly tied to the required outcome.

Useful for local disk capacity and namespace layout do not match the deployment bill of materials, but it does not remove the specific condition: hardware discovery and memory inventory do not match approved BOM.

“expected ports are up at negotiated speed with no fault counters” addresses network acceptance omitted direct NIC port and transceiver state verification, but the current hardware or infrastructure issue is hardware discovery and memory inventory do not match approved BOM. The option “ECC and system memory reports show all expected DIMMs with no active faults” is more directly tied to the required outcome.

“disk inventory and usable capacity equal approved requirements” addresses local disk capacity and namespace layout do not match the deployment bill of materials, but the current hardware or infrastructure issue is hardware discovery and memory inventory do not match approved BOM. The option “ECC and system memory reports show all expected DIMMs with no active faults” is more directly tied to the required outcome.

“ECC hardware counters remain clear under a controlled memory stress test” addresses boot completion cannot justify acceptance with critical hardware errors, but the current hardware or infrastructure issue is hardware discovery and memory inventory do not match approved BOM. The option “ECC and system memory reports show all expected DIMMs with no active faults” is more directly tied to the required outcome.

“compare BMC asset PCIe listing and nvidia-smi before driver replacement” addresses software-level symptoms were not isolated from hardware enumeration faults, but the current hardware or infrastructure issue is a network adapter is missing from PCIe enumeration or its slot is faulty. The option “inspect adapter seating PCIe slot state and BMC inventory against BOM” is more directly tied to the required outcome.

“move or configure the adapter in the qualified host slot” addresses pCIe lane width or generation constrains measured device bandwidth, but the current hardware or infrastructure issue is a network adapter is missing from PCIe enumeration or its slot is faulty. The option “inspect adapter seating PCIe slot state and BMC inventory against BOM” is more directly tied to the required outcome.

“inspect NIC identity link negotiation and cabling at both ends” addresses network acceptance omitted direct NIC port and transceiver state verification, but the current hardware or infrastructure issue is a network adapter is missing from PCIe enumeration or its slot is faulty. The option “inspect adapter seating PCIe slot state and BMC inventory against BOM” is more directly tied to the required outcome.

“disk inventory and usable capacity equal approved requirements” addresses local disk capacity and namespace layout do not match the deployment bill of materials, but the current hardware or infrastructure issue is a network adapter is missing from PCIe enumeration or its slot is faulty. The option “lspci and NIC inventory return four approved adapter identities” is more directly tied to the required outcome.

“both firmware and operating system enumerate the intended GPU” addresses software-level symptoms were not isolated from hardware enumeration faults, but the current hardware or infrastructure issue is a network adapter is missing from PCIe enumeration or its slot is faulty. The option “lspci and NIC inventory return four approved adapter identities” is more directly tied to the required outcome.

“negotiated PCIe width and throughput meet the acceptance baseline” addresses pCIe lane width or generation constrains measured device bandwidth, but the current hardware or infrastructure issue is a network adapter is missing from PCIe enumeration or its slot is faulty. The option “lspci and NIC inventory return four approved adapter identities” is more directly tied to the required outcome.

“replace approved fan modules and verify redundancy status” addresses active fan hardware faults are masked by transient service availability, but the current hardware or infrastructure issue is boot completion cannot justify acceptance with critical hardware errors. The option “diagnose and replace the failing memory component before burn-in” is more directly tied to the required outcome.

“compare BMC asset PCIe listing and nvidia-smi before driver replacement” addresses software-level symptoms were not isolated from hardware enumeration faults, but the current hardware or infrastructure issue is boot completion cannot justify acceptance with critical hardware errors. The option “diagnose and replace the failing memory component before burn-in” is more directly tied to the required outcome.

“isolate memory slot and DIMM faults using diagnostics and compare installed capacity” addresses hardware discovery and memory inventory do not match approved BOM, but the current hardware or infrastructure issue is boot completion cannot justify acceptance with critical hardware errors. The option “diagnose and replace the failing memory component before burn-in” is more directly tied to the required outcome.

“ECC and system memory reports show all expected DIMMs with no active faults” addresses hardware discovery and memory inventory do not match approved BOM, but the current hardware or infrastructure issue is boot completion cannot justify acceptance with critical hardware errors. The option “ECC hardware counters remain clear under a controlled memory stress test” is more directly tied to the required outcome.

“expected ports are up at negotiated speed with no fault counters” addresses network acceptance omitted direct NIC port and transceiver state verification, but the current hardware or infrastructure issue is boot completion cannot justify acceptance with critical hardware errors. The option “ECC hardware counters remain clear under a controlled memory stress test” is more directly tied to the required outcome.

“BMC and power telemetry show both power modules available under load” addresses system POST masks power redundancy loss, but the current hardware or infrastructure issue is boot completion cannot justify acceptance with critical hardware errors. The option “ECC hardware counters remain clear under a controlled memory stress test” is more directly tied to the required outcome.

“diagnose and replace the failing memory component before burn-in” addresses boot completion cannot justify acceptance with critical hardware errors, but the current hardware or infrastructure issue is active fan hardware faults are masked by transient service availability. The option “replace approved fan modules and verify redundancy status” is more directly tied to the required outcome.

“repair PSU or feed fault and verify dual-feed healthy state” addresses system POST masks power redundancy loss, but the current hardware or infrastructure issue is active fan hardware faults are masked by transient service availability. The option “replace approved fan modules and verify redundancy status” is more directly tied to the required outcome.

“compare BMC asset PCIe listing and nvidia-smi before driver replacement” addresses software-level symptoms were not isolated from hardware enumeration faults, but the current hardware or infrastructure issue is active fan hardware faults are masked by transient service availability. The option “replace approved fan modules and verify redundancy status” is more directly tied to the required outcome.

“both firmware and operating system enumerate the intended GPU” addresses software-level symptoms were not isolated from hardware enumeration faults, but the current hardware or infrastructure issue is active fan hardware faults are masked by transient service availability. The option “BMC sensors report healthy fans and cooling recovery within limits” is more directly tied to the required outcome.

“BMC and power telemetry show both power modules available under load” addresses system POST masks power redundancy loss, but the current hardware or infrastructure issue is active fan hardware faults are masked by transient service availability. The option “BMC sensors report healthy fans and cooling recovery within limits” is more directly tied to the required outcome.

“ECC and system memory reports show all expected DIMMs with no active faults” addresses hardware discovery and memory inventory do not match approved BOM, but the current hardware or infrastructure issue is active fan hardware faults are masked by transient service availability. The option “BMC sensors report healthy fans and cooling recovery within limits” is more directly tied to the required outcome.

“inspect adapter seating PCIe slot state and BMC inventory against BOM” addresses a network adapter is missing from PCIe enumeration or its slot is faulty, but the current hardware or infrastructure issue is network acceptance omitted direct NIC port and transceiver state verification. The option “inspect NIC identity link negotiation and cabling at both ends” is more directly tied to the required outcome.

Useful for installed security component is not properly enabled and exposed, but it does not remove the specific condition: network acceptance omitted direct NIC port and transceiver state verification.

“diagnose and replace the failing memory component before burn-in” addresses boot completion cannot justify acceptance with critical hardware errors, but the current hardware or infrastructure issue is network acceptance omitted direct NIC port and transceiver state verification. The option “inspect NIC identity link negotiation and cabling at both ends” is more directly tied to the required outcome.

“ECC and system memory reports show all expected DIMMs with no active faults” addresses hardware discovery and memory inventory do not match approved BOM, but the current hardware or infrastructure issue is network acceptance omitted direct NIC port and transceiver state verification. The option “expected ports are up at negotiated speed with no fault counters” is more directly tied to the required outcome.

“ECC hardware counters remain clear under a controlled memory stress test” addresses boot completion cannot justify acceptance with critical hardware errors, but the current hardware or infrastructure issue is network acceptance omitted direct NIC port and transceiver state verification. The option “expected ports are up at negotiated speed with no fault counters” is more directly tied to the required outcome.

“attestation software detects the intended TPM with a valid measured state” addresses installed security component is not properly enabled and exposed, but the current hardware or infrastructure issue is network acceptance omitted direct NIC port and transceiver state verification. The option “expected ports are up at negotiated speed with no fault counters” is more directly tied to the required outcome.