Weird issue with motherboard

[The red spirit]

Seriously, I have problems with FM2+ board again. It seemingly worked fine, until it didn't. I have problems with dual channel RAM. After tightening AIO half of RAM stopped being recognized. I untightened it a bit and all RAM was working again. Today I decided to overclock RAM and it eventually became unstable. No big deal. I would just reseat it (to trigger POST screen) and will just dial back on speed. I reseated RAM and none was recognized. I flexed board around RAM are and got one channel working. I tried to flex it back to make another channel work again, but no luck. I tightened AIO again and it worked in dual channel, until computer crashed and only single channel is recognized again. Really? WTF.

I inspected that stupid board closer and I don't see any kind of failure or something being wrong with it. It seemingly is in perfect shape. Can somebody tell me what could be wrong with board, RAM or CPU, because I'm losing my mind here. Last board died in very similar way. I'm starting to think that it's just dodgy RAM. Either way, all hardware visually doesn't look broken and board is few months old at best.

Anyway, specs are:
Athlon X4 870K
Biostar A68MHE
Corsair XMS3 1333MHz RAM from kit of 4x4GB, I only use 2 sticks, because that's all that board can use.

 

[Calenhad]

A motherboard have several layers. Meaning there are more traces than you can see on the top and bottom sides. Which is possibly damaged in your case. A visual inspection can only do so much. With a cheap motherboard you can maybe inspect 50%. But more expensive motherboards have 6+ layers. Issues with memory can also be hidden in pins, socket and internal components.

Since you have a 4x4GB kit, have tried all of them separately? If two or more works in single channel you can see if they will post in dual.

Does that AIO not come with the correct standoffs? Usually you won't have to think about correct mounting pressure. Having to loosen the AIO or tighten it to get the computer to post, is not a good sign either. Damaged pins or socket is plausible.

Bottom line is that this does indeed sound like failing hardware. You need another motherboard and/or processor to troubleshoot more than swapping ram though. This computer hardware is old enough to start failing from old age.

 

[Bomby569]

seems pretty straightforward, you're over tightening, there is no need to use the HULK in you

 

[The red spirit]

A motherboard have several layers. Meaning there are more traces than you can see on the top and bottom sides. Which is possibly damaged in your case. A visual inspection can only do so much. With a cheap motherboard you can maybe inspect 50%. But more expensive motherboards have 6+ layers. Issues with memory can also be hidden in pins, socket and internal components.

It's low end board, so not many layers.

Since you have a 4x4GB kit, have tried all of them separately? If two or more works in single channel you can see if they will post in dual.

Does it even matter? Anyway, out of those two sticks I tried them randomly.

Does that AIO not come with the correct standoffs? Usually you won't have to think about correct mounting pressure. Having to loosen the AIO or tighten it to get the computer to post, is not a good sign either. Damaged pins or socket is plausible.

It comes with both. One of basic AMD mounting and another more complicated mounting.

Bottom line is that this does indeed sound like failing hardware. You need another motherboard and/or processor to troubleshoot more than swapping ram though. This computer hardware is old enough to start failing from old age.

It's not old at all, motherboard is few months old. Was recently released to keep FM2+ hardware running.

seems pretty straightforward, you're over tightening, there is no need to use the HULK in you

No, I'm not. At first I tightened it properly, then I made it quite loose and it basically was hold in place mostly from thermal place suction. Now I tightened it a bit.

 

[OliverQueen]

Had similar problem with a AM3+ system I got given to "refurb" to give to someone that didn't have one. Even though the boards supports "upto 32GB" of RAM, it would only work with 2GB single sided, single rank sticks (unless the actual memory itself was on the QVL list). There is no Corsair memory listed on Biostar's memory QVL list & some higher capacity double sided, double ranked modules would only work intermittently & then only show a single module installed instead of 2. When it showed the correct memory installed, it would be unstable in any OS tried & after a restart only show what was in the first slot.

 

[The red spirit]

After installing more proper backplate and remounting AIO, everything works well. Not only that, but it managed to reach 40% overclock on memory. It really makes me unsure if it's fine or not. On one hand now it works, but on others, I don't want to touch it ever again. I wonder if there's permanent trace damage somewhere, but to be fair, such minimal motherboard bending shouldn't cause issues. I have seen much worse functioning fine.

 

[qubit]

After installing more proper backplate and remounting AIO, everything works well. Not only that, but it managed to reach 40% overclock on memory. It really makes me unsure if it's fine or not. On one hand now it works, but on others, I don't want to touch it ever again. I wonder if there's permanent trace damage somewhere, but to be fair, such minimal motherboard bending shouldn't cause issues. I have seen much worse functioning fine.

That sounds like a real headache and I suspect the mobo is at fault, most likely around the CPU socket as that's where all the mechanical stress is going.

Now it's working don't touch it again. Unfortunately, if there's an electrical break somewhere, then it's likely to show up over time and fail again. Hopefully it will happen while it's still under warranty.

 

[The red spirit]

That sounds like a real headache and I suspect the mobo is at fault, most likely around the CPU socket as that's where all the mechanical stress is going.

But it's just typical PGA socket. You can apply shit ton of pressure, before it should fail. It's not like that wed noodle LGA, that fails if you look at it wrong way.

 

[qubit]

But it's just typical PGA socket. You can apply shit ton of pressure, before it should fail. It's not like that wed noodle LGA, that fails if you look at it wrong way.

Anything can fail, regardless of the platform. On top of that, Biostar is a budget brand so the mobo likely isn't made as well as a more upmarket brand, so I wouldn't rule that out based on the socket. I learned that the hard way when I put my first PC together with a budget PC Shits Chips mobo in 1999* that was trouble from day one. That was the model with the fraudulent "100MHz" memory bus that came out in a scandal. That company went bust a while later, thankfully.

*That's 5 whole years before TPU existed and even some members here. Man how time flies...

 

[The red spirit]

Anything can fail, regardless of the platform. On top of that, Biostar is a budget brand so the mobo likely isn't made as well as a more upmarket brand, so I wouldn't rule that out based on the socket. I learned that the hard way when I put my first PC together with a budget PC Shits Chips mobo in 1999* that was trouble from day one. That was the model with the fraudulent "100MHz" memory bus that came out in a scandal. That company went bust a while later, thankfully.

*That's 5 whole years before TPU existed and even some members here. Man how time flies...

I'm not sure, Biostar seems to be exactly the same as budget Asus, Giga, Asrock stuff. And what you get today isn't remotely comparable to crapola of 90s,when far more depended on board and when we had capacitor plague, Athlon XP mounting, tons of counterfeit CPUs and outright scam brands like PCChips.

 

[TheoneandonlyMrK]

But it's just typical PGA socket. You can apply shit ton of pressure, before it should fail. It's not like that wed noodle LGA, that fails if you look at it wrong way.

Someone suggested you overtightened it , you said no.
Then fitted "the proper backplate"

Then it worked.

I can see clearly what's at fault and it's not the hardware.

And fffk no, you cannot apply a shit Ton of pressure on PGA or LGA, the right amount is fine, a little less pressure could only possibly cause thermal issues, too much pressure, As many including myself has learned, breaks shit, sometimes not permanently, but sometimes for life.

 

[eidairaman1]

Someone suggested you overtightened it , you said no.
Then fitted "the proper backplate"

Then it worked.

I can see clearly what's at fault and it's not the hardware.

And fffk no, you cannot apply a shit Ton of pressure on PGA or LGA, the right amount is fine, a little less pressure could only possibly cause thermal issues, too much pressure, As many including myself has learned, breaks shit, sometimes not permanently, but sometimes for life.

When the board warps and then continue using it, hot/cold cycles can crack traces or lift smds. I did it on this rig but before ever puuting in case i lightened the tension and never had a problem detecting ram other than trying to run 16GBs at 2400 cas 9 lol.

 

[qubit]

I'm not sure, Biostar seems to be exactly the same as budget Asus, Giga, Asrock stuff. And what you get today isn't remotely comparable to crapola of 90s,when far more depended on board and when we had capacitor plague, Athlon XP mounting, tons of counterfeit CPUs and outright scam brands like PCChips.

There's still a difference in quality, but I can't give you specifics. It stands to reason that a budget brand is gonna cut corners even if things have improved overall from the 90s.

Also, I'm afraid that @TheoneandonlyMrK has a point about overstressing it with the mounting. Perhaps it was just that?

 

[The red spirit]

Someone suggested you overtightened it , you said no.
Then fitted "the proper backplate"

Then it worked.

I can see clearly what's at fault and it's not the hardware.

And fffk no, you cannot apply a shit Ton of pressure on PGA or LGA, the right amount is fine, a little less pressure could only possibly cause thermal issues, too much pressure, As many including myself has learned, breaks shit, sometimes not permanently, but sometimes for life.

Maybe, but still new mounting doesn't have thumbscrews and has two additional screws, so I think there must be more pressure now than it was before. And I left that comment about PGA, because in Athlon XP days even stock cooler was putting a shit ton of pressure on CPU and you were more likely to destroy CPU than board. Anyway, I inspected board before remounting AIO and I didn't notice any bending. Neither PCB was bent, neither stock backplate. Truly nothing to indicate that cooler was overtightened. I would have expected clearly bent board or backplate in case of overtightening, but there was nothing this time.

 

[qubit]

And I left that comment about PGA, because in Athlon XP days even stock cooler was putting a shit ton of pressure on CPU and you were more likely to destroy CPU than board.

My Athlon XP 2500+ (yes, I still have it, sigh) actually has a slight chip on the die from my hamfisted handling of it, yet it still works perfectly and still overclocks to 2200+ level. Happy days, sigh.

 

[The red spirit]

There's still a difference in quality, but I can't give you specifics. It stands to reason that a budget brand is gonna cut corners even if things have improved overall from the 90s.

Also, I'm afraid that @TheoneandonlyMrK has a point about overstressing it with the mounting. Perhaps it was just that?

I have no idea what your experience is with budget boards, but imo Asrock and Gigabyte do worse job than Biostar. HWUB always has issues with those two, I personally had VRM overheating, buggy BIOS issues with both. All in all, I'm not impressed with either and don't have much confidence that either could make a proper BIOS. Say what you want, but old school Intel boards were seemingly very well made.

My Athlon XP 2500+ (yes, I still have it, sigh) actually has a slight chip on the die from my hamfisted handling of it, yet it still works perfectly and still overclocks to 2200+ level. Happy days, sigh.

Oh, I forgot that you could just slip screwdriver into board or destroy plastic clips. It's so nice that they fixed all that nonsense with Athlon 64. BTW what do you mena by 2200+ level? Isn't that worse chip than 2500+?

 

[qubit]

I have no idea what your experience is with budget boards, but imo Asrock and Gigabyte do worse job than Biostar. HWUB always has issues with those two, I personally had VRM overheating, buggy BIOS issues with both. All in all, I'm not impressed with either and don't have much confidence that either could make a proper BIOS. Say what you want, but old school Intel boards were seemingly very well made.

I've bought a few mobos in my time and only PC Shits gave me any serious problems. The others were one Gigabyte and a few Abit and Asus boards. There were some niggles occasionally, but nothing serious.

Oh boy, how I miss Abit, sigh.

Oh, I forgot that you could just slip screwdriver into board or destroy plastic clips. It's so nice that they fixed all that nonsense with Athlon 64. BTW what do you mena by 2200+ level? Isn't that worse chip than 2500+?

Oh sorry, typo, silly me. I meant 3200+, the top model. A lot of these chips were actually 3200+ with a lower default multiplier burned into them that worked just fine at 3200+, no extra voltage or cooling needed. It wouldn't go much above that, though.

 

[The red spirit]

I've bought a few mobos in my time and only PC Shits gave me any serious problems. The others were one Gigabyte and a few Abit and Asus boards. There were some niggles occasionally, but nothing serious.

I only use what you would call budget boards. Only now, I have something a bit higher end from Giga and it's still not great. It has some BIOS options bugged out, there's no way to disable CSM, there fore no SMA, C states bellow C3 don't work and if forced, crash computer, there are many leftovers meant for other chips (those with IGPs and K chips), I think it has serial connector options, despite no serial connector or header anywhere, heatsink on VRMs is basically a scam looks proper on website, but once you look at it, there's just a few fins and no heatsink and it's just 90 degree U shaped deal (it's hard to explain with words), memory timing tweaking is broken (some options don't do anything, despite claiming otherwise, some actually work), there's no option to set fans to custom fixed speed at all, quiet fan preset is not quiet at all (for some reason tops out fans at 70C, despite all LGA 1200 chips being capable of 100C before throttling) VRMs are quite weak (good enough for i5 I use it with, but insufficient for K chips, barely any better than on DS or S2P tier boards). Yeah, despite "Aorus Pro" name, it's quite poor board and Aorus pro was two tiers above gaming and just Aorus. I might be wrong, but that was the best B460 board they had back then at all. I got it because it was selling way bellow MSRP, at just 86 EUR. All other Gigabyte boards I had are already dead. To be fair, I pushed one too far with overclocking. That was 78LMT-S2P I think and all it did was 4.3GHz with FX 6300, despite me pushing 1.6V. VRMs were melting and there was brown discoloration around them, despite that it worked for nearly two years before failing. Another Giga was some FM2+ DS3 board, worked for a while (a bit more than one year), until it started to fail and eventually just died and didn't boot again. Pathetic. I even went with oversized cooling and babied it to work longer, but nope.

Here's that fake heatsink:

It's not fake, but rather very deceptive and stupidly innefective. As you can see, heat is supposed to travel all way from the source and then do 180 and somehow reach those fins, that barely increase surface area. Mounting pressure of it is complete joke and it's just on thermal pad. It's there for Aorus pizzaz, rather than for cooling. Considering that it blocks direct airflow on VRMs, it might actually be detrimental to their cooling. It's really stupid design and I feel dumb for not noticing it earlier.

Oh boy, how I miss Abit, sigh.

I honestly miss DFI, they were awesome. Soyo too. Soltek for colors only, maybe Chaintek too. Intel for stability. Also whoever made purple PCB boards in early 2000s, that looked so good. I would love a board with black and purple changing colors, that would be sick.

Oh sorry, typo, silly me. I meant 3200+, the top model. A lot of these chips were actually 3200+ with a lower default multiplier burned into them that worked just fine at 3200+, no extra voltage or cooling needed. It wouldn't go much above that, though.

Don't feel bad, I didn't manage to push my Athlon 64s much either. I managed to fail in the worst possible way. Set CPU clock too high (224) and it nuked Windows install completely. If it had worked, it would only be running at 2.464 GHz and stock speed was 2.2GHz. I learned hard way that you need PCI/AGP locks for clocking socket 754 chips high and my board didn't have that functionality. That or readjusting every speed to be sort of within a spec, but I didn't want to deal with so much crap just for few hundred MHz at best.

I only wonder how much higher those old Athlon XP and Athlon 64 chips would clock if we removed some common era specific limitations. Back in 2003-2004 CPU coolers were generally awful and board circuitry wasn't all that much beefed up even on top models. Tom's hardware did that for Pentium 4 and socket 478 board back then and reached stable 4.9GHz on LN2 and a bit over 5GHz with suicide runs. They also cherry picked good bin. I don't recall anyone modifying board, using way too good cooling and cherry picking AMD chips. Back then maximum limit was around 2.8-3.2 GHz for K8 chips, have no idea what it was for Athlon XP, but maybe 2.6GHz. Considering how well K10 chips clocked later and how small redesign it was from K8, I have a feeling that Athlon 64s might be able to clock a lot higher and that's without LN2, but with just a bit tower cooler, which didn't exist in 2004.