CPU/mobo/RAM upgrade results in BSOD

martbean

n00b
Joined
Apr 15, 2021
Messages
28
Hi, I'm hoping someone can help me. I recently purchased the following bundle to upgrade the PC in my sig:

AMD Ryzen 5 7600X
Gigabyte B650 GAMING X AX
2 x 16GB Corsair Vengeance DDR5

Having installed everything and booted Windows, after a few minutes it crashed with a BSOD and I've not been able to get it stable. I've tried all the usual things like updating drivers, chkdsk, sfc, Windows memory diagnostic, etc. Thinking it could be a memory issue, I double-checked the RAM is seated correctly, and tried each stick individually. Nothing seems to make a difference.

I figured it could be a software issue since it's a relatively old install of Windows so I downloaded Win 11 and tried installing that on a new drive. The initial install worked but during the setup I got another BSOD and that keeps happening.

The error seems to change each time, eg. "DPC Watchdog Violation", "Memory Management", "IRQL not less or equal", something about "kernel" that I didn't catch.

I have a 630W power supply which should be more than enough I believe.

I'm using the same cooler as before, which is the standard AMD Wraith. I've read that the 7600X runs quite hot so I might replace this, but if it's overheating it should throttle back, right? It would crash whilst idle too, so it's not like I'm thrashing it.

Anyone have any ideas?

Thanks,

Martin.
 
Last edited:
Smells of bad memory. Did you actually run MemTest? Did you try changing the memory profile to slower..?
 
Hi,

That was my first thought too, though it seems odd that both sticks are doing the same thing (though I guess they'd be manufactured together so maybe not that odd).

I ran the Windows memory diagnostic but not MemTest86. I'll try that at the weekend.

The BIOS seems to have two profile options. I forget the wording but it was along the lines of "default" or "XMP". I was using the default (slower) one but tried the faster one too.

Thanks,
Martin.
 
Reset to bios, load defaults, test. Don't touch anything else unless you need to for booting. One stick. No add on cards.

Watch the temps.
 
No mention of a GPU are you using the onboard graphics from the cpu? Load into bios and use an accurate digital volt ohm meter to test the power supply output at a molex connector for 12v yellow and 5v orange to see if the psu is carrying the load properly. Then try to boot windows while monitoring the 12v line to see if it sags when loading windows. Memory modules are on the qvl list for the mainboard?
 
I'm using the GPU from my current setup (Radeon RX 6700 XT). I did try using the onboard graphics but got no output so I might need to change a BIOS setting. I'm going to try resetting it tomorrow as Mega6 suggests.

The PSU's from my current rig too so shouldn't be a problem, but I'll try what you're suggesting if I'm feeling confident.

I checked the QVL the other day and mine is CMK32GX5M2B5600Z36 and the nearest on the list is CMK32GX5M2B5600C36 (so just Z36 instead of C36). I'm not sure what to make of that.
 
Yes try onboard grahics as that should decrease the load on the 12v rail and may allow it to function in a limited capacity. You will need to remove the gpu from the mainboard and enable integrated graphics in bios for the attempt.
 
I checked the QVL the other day and mine is CMK32GX5M2B5600Z36 and the nearest on the list is CMK32GX5M2B5600C36 (so just Z36 instead of C36). I'm not sure what to make of that.
Hard to say, that C instead of Z could just be about the color, or it could be something else. The corsair product pages for the two look pretty similar, but C seems targetted towards AMD and Z towards Intel. Might be worth checking the ram voltage, I think that was mostly a DDR4 problem, but lots of systems would auto configure for a voltage too low and things didn't work so well as a result.
 
OK, so I reset the BIOS but couldn't get it to output using the integrated graphics so had to use a GPU card. I didn't attach any other expansion cards or drives though and ran Memtest86 which failed on the first stick (see below). There were around 300 errors. I'm just testing the second stick now and straight away it's showing errors.

toast0 Thanks for this. I'll check the voltage next and I might try contacting Gigabyte to see what they say.

One other thing I noticed is regarding the ATX_12V connectors. There are two sockets - 2x2 and 2x4. I'm using the 2x4 but do I need to use both?

IMG_20230603_232546.jpg
IMG_20230603_232642.jpg
 
No you should be ok with a 7600X with the 2X4 cpu power connector. A fully occupied 8 pin connector on the cpu power header should be enough as long as the 12V rail doesn't sag under load.
 
Memtest crashed about about half an hour on the second stick. I ran it again and after two passes there were hundreds of errors so I cancelled it. I note the average temp was around 70 but the max was 96 so I should probably get a better cooler but I don't think that's the issue.

The DDR voltage reported by Memtest and the BIOS is 1.1V, which is what Corsair says it should be. More details from Memtest attached.

I've contacted Gigabyte and Corsair to ask about the module compatability.

I'd like to get to the point where I'm confident it's a hardware fault so I can return the relevant part(s). I haven't tested the power output but I don't really know what I'm doing with a multimeter so am reluctant to mess around too much. It's annoying I can't test without a discrete GPU so have contacted Gigabyte about this too. I guess there can be other reasons for RAM failing but are we thinking faulty modules?

Thanks again everyone (y)
 

Attachments

  • IMG_20230604_115301.jpg
    IMG_20230604_115301.jpg
    243 KB · Views: 0
The odds are against having two faulty memory modules and more likely to be a compatibility issue butt...
Here's the test procedure for using a molex to test a psu for sagging rails.
 
The DDR voltage reported by Memtest and the BIOS is 1.1V, which is what Corsair says it should be.

Corsair says Tested Voltage: 1.25V; SPD Voltage: 1.1V.

I'd bump it up to 1.25V and see if it helps, maybe in increments if you're patient.
 
The odds are against having two faulty memory modules and more likely to be a compatibility issue butt...
Here's the test procedure for using a molex to test a psu for sagging rails.
Thanks, I'll give it a go. It seemed unlikely to me too, to have two faulty modules.

Corsair says Tested Voltage: 1.25V; SPD Voltage: 1.1V.
Well I had it with XMP disabled (4800mhz) so assumed 1.1V was right but if enable XMP/EXPO 1, Memtest still says it's 1.1V so perhaps you're on to something. I'm struggling to work out what to change though (see attached). "DDR_VDD" and "DDR_VDDQ Voltage" perhaps? Second attachment is the Advanced Voltage Settings page.
 

Attachments

  • IMG_20230604_214437.jpg
    IMG_20230604_214437.jpg
    301.5 KB · Views: 0
  • IMG_20230604_214506.jpg
    IMG_20230604_214506.jpg
    319.8 KB · Views: 0
I'm struggling to work out what to change though
Yeah, I'd try bumping both DDR_VDD and DDR_VDDQ, and maybe CPU_VDDIO_MEM. I'd do 1.15v, then 1.20v, then 1.25v. I wouldn't touch anything in the advanced menu. If bumping all three gets things stable, you can set some of them back to stock and see which one(s?) are important and report back.
 
Yeah, I'd try bumping both DDR_VDD and DDR_VDDQ, and maybe CPU_VDDIO_MEM.
I assume you mean with XMP/EXPO1 disabled? I set the ones you mentioned to 1.15v though MemTest still says "Configured voltage: 1100mV" / "Module Voltage: 1.1V" and instantly reported errors. I tried 1.25v too with the same result.

What bios version is the board running currently?
F7a, which was the latest up until a few days ago.

As an aside, I noticed the 7600X product page says the max memory speed is DDR5-5200, which seems low. That might explain why I can't get it working at 5600 but it should at least be working at 4800.
 
Hi all. An update:
  • Corsair have confirmed that my RAM is compatible with my mobo.
  • Gigabyte couldn't help with the integrated graphics issue and suggested returning it for inspection.
  • I updated to the latest BIOS (F7b) just in case.
  • I removed and reseated the CPU. Checked for bent pins and couldn't see any.
  • As per learners permit 's suggestion I tested the voltage. It's consistently 5V/12V at boot and under load.
The only thing I have slight doubts about now is whether the 630W PSU is powerful enough. I tried a variety of PSU calculators and got these estimates: 420W, 480W, 490W, 530W, 550W, 600-699W. Quite a range. I know it's generally advised to have a bit more than required but 630W should be enough, shouldn't it?

The vendor has offered me an RMA but I want to make sure I've covered everything first.

Thanks again.
 
When you reset the bios was it done via software or bios reset jumper? HUB or GN reported a bug on Gigabyte boards that software bios resets failed to reset some settings unless the clear cmos jumper was used. Just remember to turn off the power supply switch first.
 
Try it via the cmos reset pins with the battery still in. The location of the two pin header is clearly illustrated in the manual PDF. Do not forget to turn off the psu or unplug it.
 
I need the computer for work so have had to put the old kit back in. Doesn't using the reset pins and removing the battery do the same thing though? It reset the settings for me as far as I could tell.
 
When removing the battery sometimes takes a bit of time for the board to discharge and reset to defaults. Pressing the power button when the battery is removed and box unplugged will discharge the board immediately. Using the clear cmos jumper immediately resets the bios to defaults.
 
Last edited:
Hi,

Just to update this, I returned the bundle to the vendor and they tested it. They found the motherboard and RAM to be fine but the CPU to be faulty. They've replaced it and all is working fine.

Thanks again for your help.

Martin.
 
Hard to say, that C instead of Z could just be about the color, or it could be something else. The corsair product pages for the two look pretty similar, but C seems targetted towards AMD and Z towards Intel. Might be worth checking the ram voltage, I think that was mostly a DDR4 problem, but lots of systems would auto configure for a voltage too low and things didn't work so well as a result.
The Z is the AMD-optimized (EXPO) version, while the C is the regular XMP version.

At any rate, the OP had the CPU replaced. CPU memory controller defects are rarer than faulty RAM modules.
 
Back
Top