Sudden freeze

[Thuringwethil]

I have a Dell T3500 with 24 GB of ECC memory, a Xeon, and about 10 TB disk space running under 8.1 that developed a habit of sudden freeze lately. I usually notice it when I try to move the mouse -- it won't move. Then I check my *lock keys, and if the associated lights don't toggle, then I know that I need to use the PWR button.

Is there are any systematic way to figure out what is causing it? I already looked in Event Viewer and ran memtest to no avail. Previously, I've had some hardware issues, such as failing power supply, chipset and CPU overheating, but they were all resolved, plus they mostly manifested themselves with sudden reboots.

Aside from the mouse (which may be coincidental) it doesn't seem to be associated with anything easily noticeable.

 

[ChatGPT]

Thanks — that symptom (system fully unresponsive, caps/num-lock lights not toggling, fans/LEDs still on) usually points to a low‑level hardware/firmware/driver hang rather than an ordinary app crash. I’ll give a short, systematic troubleshooting plan you can follow and what to collect so we can narrow the root cause.
Before you start

• Save any work and make a quick system image or at least back up important files. Some tests (power/PSU swaps, Driver Verifier, forcing crashes) can cause reboots or BSODs.
• Tell me: CPU model, GPU, motherboard, PSU make/model, BIOS version, and whether the freeze happens in Safe Mode or during boot. That will let me prioritize tests.

1) Collect evidence right after a freeze

• After the forced restart, open Reliability Monitor (Win+R → perfmon /rel) and Event Viewer → Windows Logs → System. Filter for WHEA‑Logger, LiveKernelEvent, Kernel‑Power, and any Display/driver errors. Those entries are the best clues for hardware/driver problems.
• Copy/save the last ~20 lines from Reliability Monitor and the system log around the crash time and attach them here.

2) Make Windows produce a useful dump (if possible)

• Set System → Advanced → Startup and Recovery → write a Kernel memory dump (or Complete if you have space). Ensure pagefile is at least the size of RAM so dumps can be written.
• Consider enabling the “Crash on Ctrl+Scroll” keyboard feature so you can force a crash/dump quickly if the system becomes unresponsive (note: if keyboard input is frozen this won’t help, but it’s useful if it sometimes still accepts special keys). If you want the exact reg keys/commands, I’ll paste them. Good references and procedures for enabling crash dumps and collecting Event Viewer output are standard troubleshooting steps.

3) Quick checks (fast, low risk)

• Try a different keyboard and mouse (or try PS/2 vs USB if available). Intermittent USB controller freezes can look like a full system hang. Several users report freezes where only input devices stop responding.
• Boot to Safe Mode or do a clean boot (msconfig → selective startup) and run the machine until the freeze would normally happen. If it never freezes in Safe Mode / clean boot, the likely culprit is a driver or third‑party service.
• Disable USB selective suspend and set Power Plan → High performance while testing. Some freezes are power-management related.

4) Hardware isolation tests (take time, but high yield)

• Memory: run MemTest86 (bootable) for at least 2 full passes (overnight if possible). MemTest from inside Windows often misses intermittent faults; more than one pass is important. If you get errors, test one DIMM at a time and test different slots.
• CPU/GPU stress: run Prime95 (or OCCT Small FFTs / IntelBurnTest) for CPU, and FurMark or Unigine/3DMark loop for GPU. Monitor temps and clocks with HWiNFO64 (sensors logging). Look for thermal runaway, throttling, or sudden clock drops.
• Combined PSU test: OCCT power test or a combined CPU+GPU stress to try and provoke the hang. If the machine hard‑freezes during combined load and you see 12V dips in logs, suspect the PSU/power delivery.

5) Firmware / drivers / BIOS

• Update motherboard BIOS/UEFI, chipset drivers, storage controller and GPU drivers from the vendor pages (not just Windows Update). For graphics, do a clean driver uninstall using DDU in Safe Mode then reinstall the latest stable WHQL driver. Firmware/driver bugs commonly produce hard hangs.

6) Reduce variables (minimal configuration test)

• Remove non‑essential PCIe cards and USB devices. Boot with only one stick of RAM (rotate sticks), no extra storage drives, and onboard video (or swap GPUs). If the freeze stops in minimal config, add parts back one at a time until it recurs — that identifies the failing component. Many long troubleshooting threads converge on PSU, motherboard, or a single peripheral.

7) PSU / motherboard checks

• Inspect electrolytic capacitors on the motherboard and PSU for bulging or leaking. If you have a multimeter, check 12V/5V/3.3V rails under idle and load if you can (or try a known‑good PSU). Intermittent power faults often cause exactly the “everything frozen but fans/LEDs still run” symptom.
• If you have a UPS, try bypassing it (plug directly to wall) to rule out UPS issues. Several users had freezes traced to power sources/UPS.

8) Advanced: Driver Verifier / crash analysis

• If logs point to drivers but not which one, run Driver Verifier for suspect drivers only; it will force a BSOD for faulty drivers so you get a dump you can analyze with WinDbg. WARNING: Driver Verifier will cause BSODs while it’s enabled — don’t run it unless you can analyze dumps or are prepared to recover. I can give exact steps and how to undo it.

9) What to post here (so I can help analyze)

• The last 20 lines from Reliability Monitor and any System log entries (WHEA, Kernel‑Power, LiveKernelEvent) that appear at or just after the freeze.
• Results / screenshots of HWiNFO sensor log (voltages, temps, clocks) around a freeze, and MemTest86 results.
• Your full system spec (CPU, GPU, PSU, motherboard, BIOS version, RAM part numbers).

Quick prioritized checklist to run in this order (estimate time):

1. Change mouse/keyboard and test (10 min).
2. Boot Safe Mode / clean boot (30 min test).
3. Enable kernel dump + reproduce (until freeze). Collect logs.
4. MemTest86 overnight (8–12 hours).
5. Stress tests with HWiNFO logging (1–2 hours).
6. Minimal hardware configuration (several hours or until freeze).
7. Swap PSU or test with known-good PSU (if available).

Why this approach: random full freezes with no keyboard LED changes are most commonly caused by PSU/power delivery, motherboard/firmware bugs, or a kernel/driver hang that prevents input handling; the above sequence narrows those down methodically without replacing parts blindly. Many forum threads and checklists recommend exactly this triage path.
If you want, I can:

• Paste the exact registry/command steps to enable “Crash on Ctrl+Scroll” and to set up kernel dumps.
• Walk you through configuring HWiNFO64 logging and what sensor values to capture.
• Help interpret the Reliability Monitor / Event Viewer output if you post the last lines here.

Tell me which of the above you want to try first and post any logs you already captured — we’ll go step‑by‑step.

 

[datawizard0103]

I have a Dell T3500 with 24 GB of ECC memory, a Xeon, and about 10 TB disk space running under 8.1 that developed a habit of sudden freeze lately. I usually notice it when I try to move the mouse -- it won't move. Then I check my *lock keys, and if the associated lights don't toggle, then I know that I need to use the PWR button.

Is there are any systematic way to figure out what is causing it? I already looked in Event Viewer and ran memtest to no avail. Previously, I've had some hardware issues, such as failing power supply, chipset and CPU overheating, but they were all resolved, plus they mostly manifested themselves with sudden reboots.

Aside from the mouse (which may be coincidental) it doesn't seem to be associated with anything easily noticeable.

Old Dell T3500s often freeze if the PSU or motherboard isn’t giving stable power. Use HWMonitor or Dell’s tool to watch temps and voltages. Also, turn off “USB selective suspend” in Power Settings and try another mouse or port. If it still hangs even in Safe Mode or while idle, it’s most likely a power issue, try testing with a different PSU first.

 

[Thuringwethil]

Old Dell T3500s often freeze if the PSU or motherboard isn’t giving stable power.

As I've mentioned in my OP. I've had issues with the PS before, but it was replaced, and that fixed the issue. It's possible that the new PS is going south, too, but the machine is now old enough that I probably won't bother with another replacement. I already have a newer T5820, although the SW install is not complete on that one.

Use HWMonitor or Dell’s tool to watch temps and voltages.

I just installed HWMonitor and it's only showing voltages for the GPU -- I'm not sure that the motherboard has the necessary sensors.

I'm thinking about building a real hardware monitor using some microcontroller and 3 high-speed ADCs that would sample +12C, +5V and +3.3V, and then dump the data over USB to a different computer.

Also, turn off “USB selective suspend” in Power Settings and try another mouse or port. If it still hangs even in Safe Mode or while idle, it’s most likely a power issue, try testing with a different PSU first.

I'm actually using a PS/2 mouse since I have 5 computers hooked up to an 8 port KVM, and the KVM has PS/2 ports. I could try a PS/2 to USB adapter since I have several of those, though I'm not sure if that would make any difference.

One issue is that this hanging is very unpredictable. I've had 3 of them in a 4-day span, but then the PC is working fine ever since I posted.

 

[datawizard0103]

On these older Dell workstations a silent chipset or southbridge lockup is surprisingly common, especially when you’re on a mix of large disks and legacy PS2 input through a KVM. When the I/O hub hangs it won’t reboot or log anything and the keyboard LEDs freeze exactly like you described. A quick way to check this without replacing hardware is to temporarily disable all non-essential onboard controllers in BIOS one at a time. Start with onboard audio, FireWire, extra SATA ports and even the onboard NIC. Run the machine for a day after each change. If the freeze stops after disabling one item you’ve found the flaky controller. It’s a simple test and works well on these aging boards.

As I've mentioned in my OP. I've had issues with the PS before, but it was replaced, and that fixed the issue. It's possible that the new PS is going south, too, but the machine is now old enough that I probably won't bother with another replacement. I already have a newer T5820, although the SW install is not complete on that one.



I just installed HWMonitor and it's only showing voltages for the GPU -- I'm not sure that the motherboard has the necessary sensors.

I'm thinking about building a real hardware monitor using some microcontroller and 3 high-speed ADCs that would sample +12C, +5V and +3.3V, and then dump the data over USB to a different computer.


I'm actually using a PS/2 mouse since I have 5 computers hooked up to an 8 port KVM, and the KVM has PS/2 ports. I could try a PS/2 to USB adapter since I have several of those, though I'm not sure if that would make any difference.

One issue is that this hanging is very unpredictable. I've had 3 of them in a 4-day span, but then the PC is working fine ever since I posted.

 

[jacobophobia]

honestly it sounds like one of those annoying hardware gremlins that doesn’t leave a clear trace. even if memtest and event viewer didn’t show anything, sudden full freezes (mouse stuck, lock-key lights not responding) usually point to something low-level like temps creeping up again, a drive starting to act slow, a flaky usb or gpu driver, or even the power supply dipping under load. windows 8.1 is also pretty old now so drivers can get weird over time. i’d check temps just to be safe, run full SMART tests on all your drives, maybe reseat your ram and try booting with minimal hardware attached. it’s basically a process of elimination with these kinds of freezes, but those are the areas that usually end up being the culprit.

 

[ChatGPT]

You’re right — this is one of those intermittent, low‑level “gremlin” hangs that’s annoying to track down. The symptom you described (full system unresponsive, keyboard lock‑LEDs not toggling) most often points to a kernel/firmware/IO hub or power delivery hang rather than a normal app crash, so we need to triage methodically rather than guessing. That’s the approach I recommended earlier.
Quick summary of what’s been tried (so we don’t repeat work)

• You checked Event Viewer and ran memtest already.
• You’re using a PS/2 mouse through an 8‑port KVM and the freezes are unpredictable (a few in a short span, then nothing).

Recommended prioritized checklist (do these in order — short steps first)
1) Kill the KVM / isolate input

• Temporarily take the T3500 off the KVM and plug a mouse/keyboard directly into the PC (use the back‑panel ports). If that’s inconvenient, try a PS/2→USB adapter or a different PS/2 mouse/port. Many silent hangs on old Dells show up when legacy input is routed through a KVM.
• Expected outcome: if hangs stop, the KVM or how it negotiates PS/2 signals is likely the trigger.

2) Simple software checks (low risk, quick)

• Boot to Safe Mode or do a Clean Boot (msconfig → selective startup). Leave it running for the period you usually see freezes. If it never freezes in Safe Mode/clean boot, suspect a driver/service.
• Disable “USB selective suspend” and set Power Plan to High Performance while testing (in Control Panel → Power Options).

3) Make Windows create usable dumps (if possible)

• Set Windows to write a Kernel memory dump (Start → System → Advanced system settings → Startup and Recovery) and ensure the pagefile is >= RAM size so dumps can be written if Windows can. I can paste the exact registry keys to enable the “Crash on Ctrl+Scroll” hotkey if you want to force a dump when observed.

4) Log sensors while idle and under stress

• Install HWiNFO64 and enable sensor logging to CSV (voltages, temperatures, clocks). On older Dells the board may not expose many sensors, but HWiNFO will capture what’s available. Run a stress test (Prime95 small‑fft for CPU, and optionally a GPU loop) while logging. Look for sudden voltage drops or temperature excursions just before a freeze.
• If the motherboard exposes little, consider the simplest external checks first (see PSU checks).

5) Minimal hardware isolation

• Remove all non‑essential devices: extra PCIe cards, USB dongles, extra HDDs, etc. Boot with one RAM stick (rotate sticks and slots), onboard video (or swap GPU), and no extra controllers. Run until the freeze either recurs or is gone. If it stops, add parts back one at a time until it returns. This often identifies the flaky component.

6) BIOS / onboard controller isolation (very useful on old Dell boards)

• Temporarily disable non‑essential onboard controllers in BIOS one at a time (audio, FireWire, extra SATA controllers, extra NICs). Run a day after each change. On T3500 boards a southbridge/chipset controller can lock and cause exactly the behaviour you see; disabling controllers is a cheap way to catch it.

7) PSU / power delivery checks

• Inspect PSU and motherboard caps for bulging or leakage. If you can, test with a known‑good PSU or at least try powering the system from a different wall outlet (bypass UPS) to rule out an external power issue. Intermittent rail dips produce the type of full hang you describe.

8) Advanced (if logs point to drivers)

• Run Driver Verifier for suspect drivers (only if you can analyze BSOD dumps or recover from them). Driver Verifier forces faulty drivers to cause a dump so you can identify them. (It will cause crashes while active — I’ll give the exact commands if you want to try.

What to collect and post here (so I can analyze)

• Last ~20 lines from Reliability Monitor after a freeze and the System log entries around the time (filter WHEA‑Logger, LiveKernelEvent, Kernel‑Power).
• HWiNFO CSV sensor log (if available) leading up to a freeze.
• MemTest86 (bootable) result if you re-run it for multiple overnight passes.
• Full system spec: exact CPU model, GPU model, motherboard model and BIOS version, PSU make/model, RAM part numbers, list of PCIe cards and external devices.

Quick suggestions based on your notes

• Since you already replaced the PSU once and the machine is old, start with BIOS controller disabling and KVM isolation — both are cheap, non‑destructive tests that have caught this exact symptom in T3500 systems.
• If you don’t want to swap PSUs, forcing better diagnostics (kernel dumps + sensor logs) is the next best step so you don’t replace parts blind.

If you want, I can:

• Paste the exact registry entries and commands to enable Crash-on-Ctrl+Scroll and to configure kernel dumps.
• Walk you through HWiNFO64 sensor logging setup and what columns to watch.
• Help parse the Reliability Monitor / Event Viewer output if you paste the last lines here.

Which of the above do you want to try first? If you prefer, start with “disconnect from the KVM and plug directly” and report back whether the freeze recurs — that’s the quickest high‑value test.