Symptom Hub · All Boards
Overheating Diagnosis
Overview
Overheating at board level stems from three distinct causes: thermal interface degradation (dried paste or pad), blocked cooling path (heatsink/fan obstruction), or elevated heat generation from a fault condition (VRM inefficiency, leakage current). The diagnostic entry point is always a temperature log under controlled load — a board in normal operation should not exceed 85°C on CPU/GPU die, 75°C on VRM hot spot, or 70°C on memory. Sustained temperatures above these thresholds cause accelerated electromigration and solder joint fatigue.
Diagnostic Methodology
Follow these steps in sequence. Each step eliminates an entire fault zone — do not skip ahead.
1. Log temperatures under load
Run a sustained CPU+GPU stress test (Prime95 + FurMark for 15 minutes). Log all temperatures. Identify which component first reaches thermal limit.
2. Check fan operation
RPM should scale with temperature. Fan not spinning at 80°C+ = failed fan, failed tachometer signal, or SMC/T2 fan control fault.
3. Heatsink seating and contact
Remove heatsink. Inspect thermal paste condition: if dried, cracked, or separated from die surface, it has failed. Inspect heatsink fins for dust blockage.
4. Repaste CPU/GPU
Clean die and heatspreader with IPA + lint-free cloth. Apply 0.5mm thin layer of paste (Kryonaut, MX-6, or equivalent). Do not use silicone-based pastes — they cure hard and are difficult to remove.
5. Measure VRM temperatures
VRM MOSFETs and inductors are major heat sources. Probe with thermocouple or FLIR under load. VRM temperature > 100°C = inadequate cooling or failing capacitors increasing switching losses.
6. Check for abnormal current draw
A fault condition (partial short, leakage) generates heat without useful work. Measure current draw and compare to spec — excess current = heat source investigation required.
Per-Board Fault Trees
Board-specific checks ordered by failure likelihood. Most common root cause listed first.
MacBook Pro A1989 — Overheating
Start:
No rail — thermal/mechanical diagnosis
-
CPU throttling < 2.0 GHzpowermetrics shows CPU capped well below base clock. 9th-gen Intel base is 2.4 GHz. Check thermal paste — dried paste on 2018+ MBP is extremely common cause.
-
Fan at max RPM, still hotFan works but cooling is inadequate. Dust accumulation in heat sink fins blocks airflow. Remove and clean with compressed air and soft brush.
-
T2 chip overheatingT2 has its own thermal pad, separate from CPU. T2 thermal throttling causes keyboard/Touch ID failures before CPU throttles. Repaste T2 separately.
GTX 1060 — Overheating
Start:
PCIe 12V under thermal load
-
GP106 > 88°C under loadGTX 1060 thermal limit is 94°C. Sustained > 88°C = replace paste with 0.5mm application. Fan curve may need manual adjustment for adequate cooling.
-
GDDR5 temp not displayedGDDR5 temps require GPU-Z or HWiNFO64 with GDDR5 monitoring enabled. Memory overheating causes artifacting. Add 1.0mm thermal pads between GDDR5 and backplate.
-
VRM hot spot > 110°CGTX 1060 VRM has minimal cooling. Under full load, VRM MOSFETs can exceed 110°C on reference cards. Add thermal pads between VRM area and backplate.
Frequently Asked Questions
What thermal paste should I use for MacBook repaste?
Kryonaut (Thermal Grizzly), MX-6 (Arctic), or Thermal Compound X23-7762 (Shin-Etsu). All provide 8–12 W/m·K conductivity suitable for soldered CPU packages. Avoid phase-change pads on thin-die CPUs — uneven application can damage die edges. Apply 0.3–0.5mm uniform layer. Do NOT use conductive pastes (IC Diamond, etc.) on Apple boards — spill risk on exposed pads.
Can I improve cooling without hardware changes?
Yes: (1) Undervolt the CPU/GPU using Intel XTU or AMD Ryzen Master — reduces heat generation by 10–20% at same performance. (2) Create a custom fan curve in GPU software or SMCFanControl (Mac). (3) Use a laptop cooling pad for mobile boards. These are useful first steps but do not replace proper thermal paste application if paste has degraded.