“NIC enumerates and ports complete full-rate diagnostics after targeted repair” addresses nIC firmware power or PCIe enumeration path may be faulty, but the current hardware or infrastructure issue is nIC host-side receive queue processing may be saturated or misconfigured. The option “corrected queue settings remove host drops while physical link remains error free” is more directly tied to the required outcome.

“system remains stable under sustained approved load with both PSUs healthy” addresses power-delivery faults coincide with load rather than application software bugs, but the current hardware or infrastructure issue is nIC host-side receive queue processing may be saturated or misconfigured. The option “corrected queue settings remove host drops while physical link remains error free” is more directly tied to the required outcome.

“fault timestamps identify the affected GPU and post-fix workload runs without new Xids” addresses gPU diagnostic events were not correlated with the reported workload failures, but the current hardware or infrastructure issue is nIC host-side receive queue processing may be saturated or misconfigured. The option “corrected queue settings remove host drops while physical link remains error free” is more directly tied to the required outcome.

Useful for power-delivery faults coincide with load rather than application software bugs, but it does not remove the specific condition: aggregate system power condition is the likely fault domain.

“correlate AER errors with the device slot and verify hardware seating” addresses pCIe slot or electrical signaling fault is blocking reliable device transfers, but the current hardware or infrastructure issue is aggregate system power condition is the likely fault domain. The option “correlate GPU launch power draw with PSU and feed alert logs” is more directly tied to the required outcome.

“isolate HCA PCIe status firmware and host driver initialization” addresses nIC firmware power or PCIe enumeration path may be faulty, but the current hardware or infrastructure issue is aggregate system power condition is the likely fault domain. The option “correlate GPU launch power draw with PSU and feed alert logs” is more directly tied to the required outcome.

“acceptable inlet and device temperatures return after airflow correction” addresses inadequate intake cooling is external to the GPU fan modules, but the current hardware or infrastructure issue is aggregate system power condition is the likely fault domain. The option “an all-GPU stress run has no current limit or PSU fault after correction” is more directly tied to the required outcome.

“NIC enumerates and ports complete full-rate diagnostics after targeted repair” addresses nIC firmware power or PCIe enumeration path may be faulty, but the current hardware or infrastructure issue is aggregate system power condition is the likely fault domain. The option “an all-GPU stress run has no current limit or PSU fault after correction” is more directly tied to the required outcome.

“fault timestamps identify the affected GPU and post-fix workload runs without new Xids” addresses gPU diagnostic events were not correlated with the reported workload failures, but the current hardware or infrastructure issue is aggregate system power condition is the likely fault domain. The option “an all-GPU stress run has no current limit or PSU fault after correction” is more directly tied to the required outcome.

“check swap-qualified optics and receive power against a known-good link” addresses external optics and cable faults have not been ruled out before card failure diagnosis, but the current hardware or infrastructure issue is the failure is localized to the missing GPU or its PCIe and power path. The option “compare PCIe enumeration slot power and GPU UUIDs to isolate the physical board” is more directly tied to the required outcome.

“swap only controlled feed paths and inspect PSU input telemetry before condemning module” addresses the suspected failure may be the feed or cable rather than the PSU itself, but the current hardware or infrastructure issue is the failure is localized to the missing GPU or its PCIe and power path. The option “compare PCIe enumeration slot power and GPU UUIDs to isolate the physical board” is more directly tied to the required outcome.

Useful for service identification lacks a validated PCIe-to-chassis location record, but it does not remove the specific condition: the failure is localized to the missing GPU or its PCIe and power path.

“tests identify the failing electrical stage instead of replacing unrelated hardware” addresses power symptoms could arise from internal module output or upstream management controls, but the current hardware or infrastructure issue is the failure is localized to the missing GPU or its PCIe and power path. The option “slot-level diagnostics identify the defective component before replacement” is more directly tied to the required outcome.

“error condition follows the defective NIC rather than the known-good path” addresses fault isolation should distinguish adapter card from external cable and switch, but the current hardware or infrastructure issue is the failure is localized to the missing GPU or its PCIe and power path. The option “slot-level diagnostics identify the defective component before replacement” is more directly tied to the required outcome.

“targeted zone inspection finds the defective fan assembly” addresses fault identification must use physical fan-zone mapping instead of a generic fan alert, but the current hardware or infrastructure issue is the failure is localized to the missing GPU or its PCIe and power path. The option “slot-level diagnostics identify the defective component before replacement” is more directly tied to the required outcome.

“inspect both AC input and DC output under controlled load” addresses power symptoms could arise from internal module output or upstream management controls, but the current hardware or infrastructure issue is the suspected failure may be the feed or cable rather than the PSU itself. The option “swap only controlled feed paths and inspect PSU input telemetry before condemning module” is more directly tied to the required outcome.

“run controlled per-GPU diagnostic and inspect its ECC counters” addresses device-specific memory faults can be isolated by per-GPU diagnostics, but the current hardware or infrastructure issue is the suspected failure may be the feed or cable rather than the PSU itself. The option “swap only controlled feed paths and inspect PSU input telemetry before condemning module” is more directly tied to the required outcome.

“perform controlled port and card substitution with preserved cable conditions” addresses fault isolation should distinguish adapter card from external cable and switch, but the current hardware or infrastructure issue is the suspected failure may be the feed or cable rather than the PSU itself. The option “swap only controlled feed paths and inspect PSU input telemetry before condemning module” is more directly tied to the required outcome.

“error condition follows the defective NIC rather than the known-good path” addresses fault isolation should distinguish adapter card from external cable and switch, but the current hardware or infrastructure issue is the suspected failure may be the feed or cable rather than the PSU itself. The option “input fault follows the cable or is isolated to the correct PSU” is more directly tied to the required outcome.

“slot-level diagnostics identify the defective component before replacement” addresses the failure is localized to the missing GPU or its PCIe and power path, but the current hardware or infrastructure issue is the suspected failure may be the feed or cable rather than the PSU itself. The option “input fault follows the cable or is isolated to the correct PSU” is more directly tied to the required outcome.

“defective transceiver or path is isolated while NIC hardware passes tests” addresses external optics and cable faults have not been ruled out before card failure diagnosis, but the current hardware or infrastructure issue is the suspected failure may be the feed or cable rather than the PSU itself. The option “input fault follows the cable or is isolated to the correct PSU” is more directly tied to the required outcome.

“check swap-qualified optics and receive power against a known-good link” addresses external optics and cable faults have not been ruled out before card failure diagnosis, but the current hardware or infrastructure issue is fault isolation should distinguish adapter card from external cable and switch. The option “perform controlled port and card substitution with preserved cable conditions” is more directly tied to the required outcome.

“inspect both AC input and DC output under controlled load” addresses power symptoms could arise from internal module output or upstream management controls, but the current hardware or infrastructure issue is fault isolation should distinguish adapter card from external cable and switch. The option “perform controlled port and card substitution with preserved cable conditions” is more directly tied to the required outcome.

“swap only controlled feed paths and inspect PSU input telemetry before condemning module” addresses the suspected failure may be the feed or cable rather than the PSU itself, but the current hardware or infrastructure issue is fault isolation should distinguish adapter card from external cable and switch. The option “perform controlled port and card substitution with preserved cable conditions” is more directly tied to the required outcome.

“defective transceiver or path is isolated while NIC hardware passes tests” addresses external optics and cable faults have not been ruled out before card failure diagnosis, but the current hardware or infrastructure issue is fault isolation should distinguish adapter card from external cable and switch. The option “error condition follows the defective NIC rather than the known-good path” is more directly tied to the required outcome.

“slot-level diagnostics identify the defective component before replacement” addresses the failure is localized to the missing GPU or its PCIe and power path, but the current hardware or infrastructure issue is fault isolation should distinguish adapter card from external cable and switch. The option “error condition follows the defective NIC rather than the known-good path” is more directly tied to the required outcome.

“targeted zone inspection finds the defective fan assembly” addresses fault identification must use physical fan-zone mapping instead of a generic fan alert, but the current hardware or infrastructure issue is fault isolation should distinguish adapter card from external cable and switch. The option “error condition follows the defective NIC rather than the known-good path” is more directly tied to the required outcome.

“perform controlled port and card substitution with preserved cable conditions” addresses fault isolation should distinguish adapter card from external cable and switch, but the current hardware or infrastructure issue is fault identification must use physical fan-zone mapping instead of a generic fan alert. The option “correlate BMC sensor number with service-manual fan positions” is more directly tied to the required outcome.

“use platform service topology and serial mapping to locate the reported card” addresses service identification lacks a validated PCIe-to-chassis location record, but the current hardware or infrastructure issue is fault identification must use physical fan-zone mapping instead of a generic fan alert. The option “correlate BMC sensor number with service-manual fan positions” is more directly tied to the required outcome.

“compare PCIe enumeration slot power and GPU UUIDs to isolate the physical board” addresses the failure is localized to the missing GPU or its PCIe and power path, but the current hardware or infrastructure issue is fault identification must use physical fan-zone mapping instead of a generic fan alert. The option “correlate BMC sensor number with service-manual fan positions” is more directly tied to the required outcome.

“error condition follows the defective NIC rather than the known-good path” addresses fault isolation should distinguish adapter card from external cable and switch, but the current hardware or infrastructure issue is fault identification must use physical fan-zone mapping instead of a generic fan alert. The option “targeted zone inspection finds the defective fan assembly” is more directly tied to the required outcome.

“slot-level diagnostics identify the defective component before replacement” addresses the failure is localized to the missing GPU or its PCIe and power path, but the current hardware or infrastructure issue is fault identification must use physical fan-zone mapping instead of a generic fan alert. The option “targeted zone inspection finds the defective fan assembly” is more directly tied to the required outcome.

“live PCIe device ID maps unambiguously to the chassis service bay” addresses service identification lacks a validated PCIe-to-chassis location record, but the current hardware or infrastructure issue is fault identification must use physical fan-zone mapping instead of a generic fan alert. The option “targeted zone inspection finds the defective fan assembly” is more directly tied to the required outcome.