NUMA and Memory Affinity: Inputs and Outputs

Inputs and outputs contract

Inputs and Outputs for NUMA and Memory Affinity centers on remote memory access ratio, NUMA latency delta, and workload affinity score. Tie every claim to a measurable artifact and an owner-controlled action.

INPUTS
  - workload definition and target KPI
  - binary/compile flags/runtime/firmware metadata
  - microarchitecture and silicon assumptions
  - correctness and regression gates

OUTPUTS
  - evidence-backed bottleneck classification
  - owner-signed fix proposal
  - validation matrix with rollback thresholds

Ownership split

CPU OWNERSHIP LAYERS - NUMA and Memory Affinity

artifact area     owner
----------------  ----------------------------
architecture    system software owner
RTL/microarch   platform architect
software/tools  performance engineer

Rule: every regressed metric must map to an explicit owner and closure artifact.

CPU deep dive

Scaling across cores is limited by coherence and interconnect behavior before compute saturation on many workloads.

Concept diagram

MULTICORE SYSTEM VIEW

cores + private caches <-> LLC slices <-> interconnect <-> memory controllers

Metric graph

SCALING EFFICIENCY

ideal scaling        ███████████
observed under load  ███████
after policy tuning  █████████

Reports and artifacts

• coherence traffic matrix

• NoC congestion map

• NUMA locality profile

• synchronization contention report

Mini case study

A lock-heavy service regressed at higher core counts because coherence invalidations and NoC hotspots dominated.

Debug branches

• Classify traffic as coherence, demand miss, or synchronization

• Measure hotspot links instead of aggregate NoC throughput

• Validate thread and page affinity before hardware changes

Senior review question

Ask: which CPI/latency evidence proves this topic is truly closed beyond synthetic benchmarks?

Key takeaways

• Always connect microarchitectural counter changes to product workload outcomes.

• Lock binary, compiler, firmware, and thermal metadata before comparing CPU traces.

Common pitfalls

• Treating average IPC as sufficient proof while ignoring latency tails and outliers.

• Applying predictor or prefetch tweaks without first-failing-stage attribution.

• Declaring closure without reproducible perf, correctness, and power gates.

Handoff explanation

Inputs are broader than knob settings. CPU analysis inputs include workload mix, branch entropy, memory footprint, compiler revision, OS affinity policy, DVFS state, thermal envelope, and stepping.

Outputs must support action: remote memory access ratio, NUMA latency delta, and workload affinity score, artifact packet (NUMA locality profile, page-placement log, and latency percentile report), bottleneck class, owner, expected effect, and rollback scope. "Performance improved" without this packet is not closure-ready.

The safest handoff is before/after evidence: environment tags, counters, traces, hypothesis, chosen change, rejected alternatives, and regression criteria.