You popped in new RAM, hit the power button, the fans spin, and the board's debug LED stops on DRAM. Nothing on screen. Could be five seconds, could be five minutes — the light just sits there. This is one of the most common calls I get, and 9 times out of 10 it's something dumb and physical. Here's the order I work through it, from most likely to least.
Cause 1: Sticks aren't fully seated (or you used the wrong slots)
I know. You already pushed them in. Push again — harder than feels comfortable. DDR4 and DDR5 retention clips need an audible click on both ends, and on DDR5 the module has to be dead flat. If one end is a millimeter up, the board sees a dead channel and parks on DRAM.
Pull both sticks out and look at the gold fingers. You should see a clean contact pattern with no dark smears. If you see one, clean it with isopropyl alcohol on a lint-free swab — finger oils from handling can absolutely cause a dead channel on a new stick.
Then reseat. On most boards with four DIMM slots, you want A2 and B2 — that's the second and fourth slot from the CPU. Check your manual, it's silkscreened on the board but it's tiny. Running a single stick in A1 will often work on some boards and won't on others; the DRAM LED behavior varies by chipset. While you're in there, make sure the slot itself doesn't have a bent pin or debris. I've pulled a dead moth out of a DIMM slot before. It happens.
If it still won't POST with two sticks, pull one and boot single-stick in A2. If that works, try the other stick alone. A dead stick from the factory is uncommon but real, and this is the fastest way to catch it.
Cause 2: Memory training failure — DDR5 especially hates this
DDR5 boards do memory training on first boot after any RAM change, and it can take 30 seconds to several minutes with no video output. The DRAM LED will be solid the entire time. A lot of people panic and kill the power at the 15-second mark — which is the worst thing you can do, because you interrupt training mid-cycle.
Give it a full five minutes before you touch anything. Yes, five. Yes, it feels broken. On Ryzen 7000 and 13th/14th gen Intel boards with 64GB or 96GB kits, training can genuinely take that long the first time.
If it's been five minutes and still nothing, power down, unplug the PSU, and clear CMOS:
- Unplug the wall cable and hold the case power button for 10 seconds to drain residual.
- Find the CLR_CMOS jumper (2-pin, usually near the front panel header or the battery). Short it with a screwdriver for 10 seconds.
- Or pull the CR2032 battery for 60 seconds.
- Reconnect, boot, and let it retrain. Give it the full five minutes again.
Clearing CMOS wipes the stored memory profile, so the board falls back to JEDEC defaults. If it POSTs after that, you know the RAM itself is fine and the problem is the XMP/EXPO profile — see cause 3.
Some boards have a Memory Try It or Memory Boot Mode switch with a Safe Mode option. Use it. It boots the kit at 4800 MT/s so you can get into BIOS and sort out the profile manually.
Cause 3: XMP/EXPO profile the board or CPU can't actually run
The kit says 6000 CL30 on the box. That's a marketing number that assumes a specific CPU memory controller, a specific board, and a specific BIOS revision. Change any of those and the profile can be a paperweight.
Common triggers I see:
- Four-stick DDR5 kits on AM5. The memory controller on Ryzen 7000 often can't hit 6000 with four DIMMs populated. 3600–4800 is realistic. The QVL won't save you here.
- Intel 12th/13th gen with a 4-DIMM board and a high-speed kit — the board's daisy-chain topology prefers two sticks, and it will fight you on four.
- BIOS on the board predates the kit. This is huge for DDR5. Update the BIOS before you do anything else — AGESA updates exist almost entirely to fix memory compatibility.
If the PC boots with XMP off but hangs on XMP on, go into BIOS and drop the DRAM frequency one notch manually. 6000 down to 5600, or 5600 to 5200. Bump VDDIO / VSOC slightly if your board exposes it, but don't go wild — VSOC above 1.3V on AM5 is asking for trouble. For Intel, XMP II or XMP Tweaked sometimes works when XMP I doesn't. Weird but true.
Check the motherboard's memory QVL on the vendor site. Not the RAM maker's site — the board maker's. If your exact part number isn't listed, that doesn't mean it won't work, but it tells you nobody at the board vendor validated it, and you're on your own for tuning.
Less common but worth checking
Three things I'd rule out before RMA'ing anything:
- Bent CPU socket pins. If you pulled the CPU to install a new cooler at the same time as the RAM, you may have nudged a pin. Memory channel issues are the first symptom. Look with a bright light at a shallow angle. Re-seat the CPU and check the DRAM light again.
- BIOS revision too old for the kit. I said it above but it's worth repeating — Flash BIOS Button on modern boards lets you update without a CPU installed. Use it.
- PSU sag. Rare, but a dying PSU that can't hold 12V under the extra current draw can present as a POST failure. Test with a known-good unit if you've exhausted everything else.
Quick reference
| Symptom | Most likely cause | First fix |
|---|---|---|
| DRAM LED solid, no display, just installed RAM | Not fully seated / wrong slots | Reseat in A2/B2, listen for both clicks |
| DRAM LED solid 30s–5min, then POSTs | DDR5 memory training | Wait. Do not power off. 5 minutes. |
| Boots fine with XMP off, hangs with it on | Profile unstable on this board/CPU | Update BIOS first, then drop one frequency step |
| No POST after also swapping CPU cooler | Bent socket pins | Pull CPU, inspect pins, reseat |
| One stick works, other doesn't | Dead module | RMA the dead stick |
Work top to bottom. Don't skip the seating check because you're sure — I've been sure and been wrong more times than I'll admit. And update that BIOS before you return the kit. Half the "bad RAM" RMA's I've seen come back tested fine.