Your NVMe drive feels blazing fast for a few seconds, then suddenly stutters, load times spike, and games start hitching or freezing mid-transfer. In Windows you might see the drive disappear for a moment, file copies drop to a crawl, or you get a warning in HWInfo or Samsung Magician that the drive is at 70-80C and throttling. That sharp drop-off after a burst of speed is the classic sign of thermal throttling.
This is not an emergency that will instantly kill your drive, but you should not ignore it either. Modern NVMe controllers will protect themselves by slowing down, yet running at sustained high temperatures accelerates NAND wear, can cause unstable writes, and makes the throttling worse over time as the controller ages. The fixes below are ordered from quickest to most involved so you can stop the throttling without replacing hardware unless you have to, and reflects best practices as of 2026.
Likely Causes
| Cause | How to confirm | Fix difficulty |
|---|---|---|
| Missing or poorly mounted heatsink / thermal pad | Visual inspection: bare drive with no motherboard or aftermarket heatsink, or pad not contacting controller | Easy |
| Poor case airflow / hot GPU blowing on drive | Drive temp drops 10C+ with side panel removed; GPU hotspot directly over M.2 slot | Easy |
| Wrong M.2 slot (under GPU, no heatsink, or PCIe 5.0 drive in PCIe 4.0 slot with no cooling) | Check motherboard manual slot diagram vs. where drive is installed; compare temps between slots | Easy to Moderate |
| Plastic film left on motherboard heatsink or dried-out thermal pad | Remove heatsink and look for clear plastic film or cracked, dry pad that leaves no imprint | Easy |
| Aggressive power plan / no throttling hysteresis (Active Link Power Management disabled) | Check Windows Power Options and BIOS ASPM settings; drive idles above 55C with no load | Moderate |
| Failing drive firmware or dust-clogged chipset / M.2 fan | Firmware version is outdated in CrystalDiskInfo; visible dust blocking M.2 fan or chipset heatsink | Moderate |
Fix 1: Confirm Real Temperatures and Throttle Threshold
What causes it
You cannot fix throttling until you know if it is actually thermal. NVMe controllers typically start throttling between 75C and 82C on the composite/controller sensor, but some utilities report only the NAND sensor which runs cooler. Without the correct reading you may chase the wrong problem.
How to check
Install HWInfo64 and run Sensors-only mode. Scroll to your drive section, for example Samsung 980 PRO or WD Black SN850X. Look for Drive Temperature, Drive Temperature 2, and Controller Temperature if listed. Leave HWInfo logging while you run a 10 GB file copy or use the benchmark in CrystalDiskMark with Settings > Test Data set to Default. Note the peak before performance drops. Also check CrystalDiskInfo > Function > Advanced Feature > AAM/APM Control to see SMART attribute 194 or 190 for thermal events.
What to do
Establish a baseline: at idle on desktop the drive should be 35C to 50C. Under a sustained write it may climb to 60C to 75C. If you hit 76C or higher and see clock or transfer speed drop in HWInfo or Task Manager > Performance > SSD active time spiking to 100% while transfer rate collapses, throttling is confirmed. Screenshot the sensors for comparison after each fix.
How to undo it
No system changes were made. Just close HWInfo and CrystalDiskMark. Delete logs from Documents > HWInfo64 Logs if you do not need them.
Fix 2: Remove Plastic Film and Reseat the Motherboard Heatsink
What causes it
Most motherboards include an integrated M.2 heatsink with a pre-applied thermal pad. From the factory that pad has a clear protective film. If the builder forgets to peel it, heat cannot transfer and the drive will overheat even at idle. A loose heatsink screw creates the same air gap.
How to check
Power off, flip the PSU switch to Off, and hold the power button for 5 seconds to discharge. Remove the GPU if it covers the M.2 slot. Unscrew the M.2 heatsink, usually one or two Phillips screws at the edge. Lift the heatsink. If you see a shiny, clear plastic layer on the pad or on the underside of the heatsink, it was never removed. Check that the pad leaves an oily imprint of the controller and NAND packages when pressed.
What to do
Peel the plastic film off completely. If the pad is torn or does not touch the drive, replace it with a 1.0mm thick thermal pad cut to the exact size of the controller and NAND. Do not stack multiple pads. Place the drive back in the M.2 standoff, secure it with the standoff screw, then reinstall the heatsink and tighten screws until snug. Do not overtighten. Boot and recheck idle temps.
How to undo it
If temperatures worsen, power off and remove the heatsink again. Verify the pad is centered and the drive is fully seated in the slot. Reinstall the original pad orientation or replace with the previous pad you removed.
Fix 3: Move the Drive to a Cooler M.2 Slot
What causes it
The top M.2_1 slot is often directly under the graphics card and between the CPU and GPU exhaust. Hot air from the GPU backplate can add 15C to the drive. Secondary slots lower on the board, often labeled M.2_2 or M.2_3 and connected via the chipset, run significantly cooler and usually still have a full heatsink.
How to check
Open your motherboard manual PDF and find the Storage > M.2 Slot Configuration page. It shows which slots support PCIe 4.0 x4 or PCIe 5.0 x4 and which share bandwidth. Identify where your OS drive is now. In BIOS, go to Advanced > Onboard Devices Configuration > M.2 Configuration to confirm link speed.
What to do
Power off and move the NVMe drive to the lowest M.2 slot that still has a heatsink and full x4 wiring. For most ATX boards, M.2_2 or M.2_3 below the GPU is ideal. If you have a PCIe 5.0 drive, ensure you move it to a PCIe 5.0 certified slot if possible, otherwise it will still run cooler at PCIe 4.0 speeds. Install the heatsink properly as in Fix 2. You can clone the drive first with Macrium Reflect or do a clean Windows install if this is your boot drive and Windows fails to boot due to BIOS boot order changes.
How to undo it
Move the drive back to the original slot. In BIOS > Boot > Boot Option Priorities, reselect Windows Boot Manager for the correct drive as Boot Option #1.
Fix 4: Improve Case Airflow Over the M.2 Area
What causes it
Even with a heatsink, stagnant air around the M.2 area traps heat. Cases with only rear exhaust and no intake, or front panels blocked by dust filters, create a hot pocket around the motherboard where NVMe drives sit.
How to check
With HWInfo logging, run a file copy for 3 minutes with the side panel on, note the max temp. Power off, remove the side panel, and repeat the same copy. If the temperature is more than 8C lower with the panel off, airflow is the limiter. Visually check that front intake fans spin and that dust is not caked on the front filter or bottom intake.
What to do
Set a proper fan curve in BIOS > Monitor > Q-Fan Configuration or via FanControl software. Set front intake fans to 40% at 40C motherboard temp and 70% at 65C, and ensure they are set to PWM and not DC if they are 4-pin fans. Vacuum or wash the front dust filter. If your GPU is blowing hot air straight onto the drive, add a small 40mm or 50mm fan pointed at the M.2 heatsink if your motherboard has an M.2 Fan Header, or reposition a spare 120mm fan as bottom intake. Keep cables tied away from the M.2 area.
How to undo it
In BIOS > Monitor > Q-Fan Configuration, press F5 to Load Optimized Defaults or select Profile > Standard. Reinstall the side panel.
Fix 5: Install a Proper Aftermarket NVMe Heatsink and Thermal Pad
What causes it
Bare drives or thin motherboard heatsinks without enough mass cannot absorb the heat spike from the controller during sustained writes. Double-sided drives with NAND on both sides also need bottom pads that many stock heatsinks lack.
How to check
After Fix 2 and Fix 4, if the controller still hits 75C+ within 30 seconds of load, cooling capacity is insufficient. Check if your drive is bare with no heatsink, or if the motherboard heatsink is a thin decorative strip less than 3mm thick with no fins.
What to do
Buy a dedicated NVMe heatsink with fins and included silicone thermal pads, such as a single or double-sided M.2 2280 heatsink kit. Power off and remove the drive. Clean the controller and NAND with isopropyl alcohol. Apply the 0.5mm pad to the bottom of the drive if double-sided, and the 1.0mm pad to the top so it contacts the controller, DRAM, and NAND. For drives with a factory sticker label, leave the label on unless the heatsink manual specifically says to remove it; the label is thin and acts as an electrical insulator. Secure the heatsink with the included clips or screws, ensuring even pressure. Do not use liquid metal or thick putty.
How to undo it
Remove the aftermarket heatsink and reinstall the original motherboard heatsink with its original pad. Keep the aftermarket pads for reuse; if they tear, replace them before reinstalling.
Fix 6: Adjust Power Settings and Update Firmware
What causes it
Windows High Performance and some BIOS defaults disable PCIe power saving, keeping the controller at full voltage even when idle, which raises baseline temps by 10C. Old firmware may have overly aggressive boost behavior with no idle cool-down.
How to check
In Windows, go to Control Panel > Hardware and Sound > Power Options > Change plan settings > Change advanced power settings. Expand PCI Express > Link State Power Management. If it is Off, power saving is disabled. Check firmware: open CrystalDiskInfo and note Firmware version, then open the manufacturer’s tool: Samsung Magician, WD Dashboard, Crucial Storage Executive, or Sabrent Rocket Control Panel, and see if a newer version is listed.
What to do
In Power Options, set Link State Power Management to Maximum power savings on Balanced plan. Do not use High Performance for NVMe testing. In BIOS, go to Advanced > Platform Power Configuration or Advanced > PCI Subsystem Settings and set PCI Express Native Power Management and ASPM to Enabled and ASPM L1 Substates to L1.1 & L1.2 if available. For firmware, in the manufacturer tool click Drive Details > Firmware Update > Update, and do not power off during flashing. After updating, in Device Manager > Disk drives > Right-click NVMe drive > Properties > Policies, ensure Enable write caching on the device is checked and Turn off Windows write-cache buffer flushing is unchecked unless you have battery backup.
How to undo it
In Power Options > Advanced settings, set Link State Power Management back to Off. In BIOS, set ASPM back to Disabled. Firmware cannot be downgraded easily, so only update if the release notes list thermal or stability improvements and you have backed up data.
Fix 7: Clean and Repaste the Chipset and M.2 Fan Assembly
What causes it
High-end boards with active M.2 or chipset fans get clogged with dust, and the fan stops or spins slowly. The chipset heatsink then saturates and heats the adjacent M.2 slots from below.
How to check
Power off and look at the small 30mm to 40mm fan near the M.2 slots. Shine a flashlight. If dust bunnies block the fins or the fan blades do not spin freely when nudged with a finger, airflow is blocked. In BIOS > Monitor > Fan Speed Monitor, see if Chipset Fan or M.2 Fan RPM is 0 or below 1000 RPM under load.
What to do
Use compressed air in short bursts to clear the fan and heatsink fins while holding the fan to prevent overspin. If the fan profile is set to Silent in BIOS > Monitor > Q-Fan Configuration > Chipset Fan, change it to Standard or Manual at 50% minimum. If you are comfortable, remove the chipset heatsink, clean old thermal pad residue, and apply a new 1.0mm to 1.5mm thermal pad cut to size. Reassemble and verify RPM returns to 2000-4000 RPM under load.
How to undo it
Set the Chipset Fan profile back to Silent or Auto in BIOS. If you replaced pads, keep the old pads measurements so you can revert thickness if the heatsink does not make good contact.
If Nothing Works
If every fix still leaves the drive throttling above 75C within a minute, test the drive in another PC or with a USB NVMe enclosure that has its own thermal pad to rule out a faulty temperature sensor. Run the manufacturer diagnostic long test: Samsung Magician > Diagnostic > Full Scan, WD Dashboard > Tools > Drive Health > Extended Test. If the drive fails or throttles at low load in a different system, back up immediately via File Explorer or Macrium Reflect to another drive and contact the manufacturer for RMA, the drive may have a defective controller. As a temporary mitigation, limit sustained writes by moving your game library and scratch disk to a secondary SATA SSD and keep the hot NVMe as OS-only until it is replaced. Avoid covering the drive with thick decorative heatsinks that have no airflow, they trap more heat.
FAQ
Is it safe to remove the sticker on my NVMe drive before adding a heatsink?
For most consumer drives the sticker is thin and does not hurt cooling, and removing it can void warranty with some brands. Leave it on and place the thermal pad directly over it unless the heatsink instructions explicitly tell you to remove it. The controller heat transfers through the sticker without issue if pad pressure is good.
What temperature should an NVMe drive stay under to avoid throttling?
Aim for under 70C under sustained load as measured on the controller sensor in HWInfo. Idle should be well below 50C with good airflow. If you stay at 65C to 70C during a 5-minute file copy or game install, you have enough headroom and will not see performance drops.
Will a PCIe 5.0 NVMe drive overheat more than a PCIe 4.0 drive?
Yes. PCIe 5.0 controllers run hotter and almost always require a finned heatsink with active airflow. If your motherboard only has a thin flat heatsink for that slot, use an aftermarket PCIe 5.0 rated heatsink or move the drive to a slot with a larger heatsink rather than running it bare.
Can software alone fix NVMe throttling without opening the PC?
Partially. Enabling Link State Power Management and updating firmware can lower idle temps by 5C to 10C and smooth boost behavior, which you can do from Windows and BIOS without opening the case. However if the drive lacks a proper heatsink or the plastic film was never removed, software alone will not prevent throttling under heavy game installs or large file transfers.
