Bottom line: WHEA Logger events, machine-check exceptions, or random reboots on AMD AM5 after enabling EXPO usually mean the memory overclock is unstable on your CPU/board/population — not that you should immediately enable Memory Context Restore. Disable EXPO, prove stability at JEDEC, reduce sticks or frequency, update BIOS, then retest. MCR is for slow boots, not WHEA repair — see AM5 slow-boot / MCR only after memory is clean.
Prefer stable kits shaped like DDR5-6000 CL30. Spec check date: July 22, 2026.
WHEA vs slow boot (do not mix the fixes)
| Signal | Likely issue | Primary guide |
|---|---|---|
| Event Viewer WHEA-Logger / MCE | Unstable EXPO, IMC stress, bad population | This page |
| Long black screen, then normal Windows | DDR5 training / MCR candidate | AM5 slow boot |
| EXPO option missing or won’t save | Firmware / SPD / recovery | EXPO won’t enable |
| Four sticks only | DIMMs-per-channel load | Four-Stick DDR5 Stability |
Enabling MCR on an unstable EXPO profile can produce confusing intermittent faults. Stabilize memory first.
Symptom → first action
| Symptom | First action | Next step |
|---|---|---|
| WHEA events only with EXPO on | Disable EXPO; retest 24–48h | Update BIOS; retry lower frequency |
| WHEA at JEDEC defaults | Reseat; test one stick | RMA if a stick fails alone |
| WHEA after adding second kit | Remove to one kit | Avoid four-stick 6000+ |
| WHEA under render/game load only | Longer memory stress at EXPO | Drop to 5600; retest |
| WHEA after BIOS update | Clear CMOS; re-train | Re-apply EXPO carefully |
EXPO setup reference: XMP / EXPO BIOS Guide. Health testing: How to Test RAM Health.
Why AM5 throws memory-related WHEA events
On Zen 4 / Zen 5, EXPO raises frequency, timings, and related rails beyond JEDEC. The CPU memory controller, motherboard traces, and module ranks must all tolerate that point. Failures increase when:
- 1UCLK:MCLK leaves a stable 1:1 relationship at high MT/s
- 2Four DIMMs or heavy dual-rank load exceeds the sample’s margin
- 3BIOS/AGESA lacks training fixes for your IC
- 4Manual SoC / voltage guesses from forums overshoot or undershoot
- 5Mixed modules share one compromised timing set
AMD lists lower official JEDEC speeds for four-DIMM configurations on many Ryzen CPUs — the electrical load story matches real-world EXPO WHEA tickets even when two sticks are fine.
Sources: AMD EXPO technology, AMD Ryzen 7 9800X3D specifications.
Stabilization sequence (WHEA-focused)
- 1Document Event Viewer IDs, BIOS version, kit part numbers, and stick count.
- 2Disable EXPO and any manual DRAM/SoC overrides; boot JEDEC.
- 3Run a memory test at defaults — test guide.
- 4If defaults fail → single-stick isolation → RMA path.
- 5If defaults pass → update BIOS, reload defaults, enable EXPO alone.
- 6If WHEA returns → keep EXPO timings but set DDR5-5600 (or the next lower stable step), test again, then try 6000.
- 7Reduce to two sticks whenever capacity allows (2×32GB over 4×16GB).
- 8Only after clean endurance testing consider MCR for boot time.
QVL double-check: How to Read a Motherboard QVL.
Population and rank decisions
| Configuration | WHEA risk at 6000 EXPO | Prefer |
|---|---|---|
| 2×16GB matched | Lowest | Default gaming kit |
| 2×32GB matched | Moderate | Creators / 64GB target |
| 4×16GB | High | Replace with 2×32GB |
| Mixed brands/kits | High | One new matched kit |
Rank background: Dual-Rank vs Single-Rank. Mixing rules: Mixing RAM Compatibility.
What not to do
- Do not “fix WHEA” by maxing SoC voltage from a random screenshot.
- Do not enable MCR to hide EXPO instability.
- Do not chase DDR5-7200/8000 on AM5 to outrun errors — 7200 worth it?.
- Do not ignore WHEA that still appears at JEDEC — that is hardware fault territory.
When to replace the kit or board assumptions
| Evidence | Action |
|---|---|
| One module fails JEDEC alone | Return / RMA that kit |
| Matched two-stick EXPO fails on multiple BIOS versions at 5600 | Try a different QVL kit (different IC) |
| Any kit fails on one slot only | Board/slot damage — RMA board |
| Stable at 5600, fails only at 6000 on a non-QVL kit | Keep 5600 or buy QVL 6000 CL30 |
Commerce recovery loop (return → rebuy → validate)
- 1Prove JEDEC — if errors remain at defaults, treat as hardware fault and start a return.
- 2Return with evidence — Amazon RAM return guide.
- 3Rebuy one matched DDR5-6000 CL30 EXPO two-stick QVL kit — DDR5-6000 CL30 · Best RAM for AM5 · Best RAM Deals.
- 4Avoid four-stick replacements — prefer 2×32GB for 64GB.
- 5Validate with a long memory test before enabling MCR — How to Test RAM Health · AM5 slow boot.
Bottom line
WHEA on AM5 is a stability signal: back off EXPO, cut load, update BIOS, then climb carefully to 6000. Use the slow-boot MCR guide only after memory tests are clean. The durable buy remains a matched two-stick DDR5-6000 CL30 EXPO kit.