Which part did I kill?

[pio]

Welcome to another round of "Which part did I kill?"........  

On today's episode, we have my good ol' "trusty" AMD FX "retro rig" I built a while back (current sig rig).  Let's dive in, and see if we can find an answer for me shall we?  (Please lol).
 

So.....this all started during the hwbot Team Cup.  We needed a (4P) FX run, which only certain Gigabyte boards were capable of doing (1 module per core active).  I went ahead and took one for the team, and ran it in a 4 phase Gigabyte 990FX UD3, with my good clocking FX 8350.  Chip used to pull 4.8GHz stable at 1.56v no problems.  Well, during that run, the stupid Gigabyte board decided that 1.55v set was going to be 1.75v actual.  Damaged the CPU (I'm assuming).

Swapped the CPU back into my Sabertooth 990FX, figured it was an FX.....it'll be fine.  System SEEMED fine for the week-ish I had it setup like that.  I never really stressed it, but it booted right up at its old clock speeds fine and was working.  Ripped a few thrift store CD's to my collection with it, again seemed fine.

I received an EVGA Capture card in about a week later.  I hooked it up, and I also hooked up a VGA adapter to it and an HDMI splitter.  This way I could run HDMI into the capture card OR a VGA cable into it, and either way I could capture footage.  It SEEMED to be working, but for some odd reason it kept crashing.

When trying to determine the crashing, I swapped rigs out completely at that station, trying my Phenom II rig in its place.  Something wasn't right.  I noticed that with the Phenom II rig connected, my fans were powering up..........WITHOUT the AC power plug even connected to the wall!!!!!  So something, either the capture card or the splitter or VGA adapter, was backfeeding +5v back through the HDMI cable.  I ran the FX rig like this for a week or so before figuring this out.  (See picture for proof of the power problem)

 

(Pictured rig is my 1100T build, I never ONCE powered it on like this, HDMI plug removed, the lights turned off)

(Pictured rig is my 1100T build, I never ONCE powered it on like this, HDMI plug removed, the lights turned off)

 

NOW.........

I've swapped CPU to @neurotix's old FX 8350, thank you @damric.  I've applied PTM to it.  I've swapped in my known good bench PSU (and its showing kinda low +12v according to software, haven't tried a multimeter yet).  The system will NOT even run at stock / auto / turbo.  It just freezes under load.  No errors, no BSOD's.....just flat out freezes.

How screwed am I?  Any idears on what I can check from here?  I really hope I didn't kill the board with that stupid attempt at trying to stream my screen. -_-

Edited December 19, 2023 by pioneerisloud

[pio]

I've stripped the rig down to just the NVME, CPU, RAM, motherboard, and GPU (known good RX 580).  Still doing it.  Nothing on the front panel is connected and no SATA drives either.  I'm thinking its the motherboard still having a problem. -_-

I'll swap CPU's out tomorrow to check, but looking like the board suffered a failure somewhere, possibly from the capture card shenanigans and not the dead CPU from the Gigabyte board like I thought.  The old FX was definitely faulty, it was requiring 1.5v for stock.  But it wasn't freezing like this new CPU is.  The only thing in between them is that HDMI cable carrying power (as shown above).

[UltraMega]

Based on what you describe with the fans powering on from something other than the power cord, seems like that would have to be something the motherboard is responsible for, so I'd assume that's probably the part that's bad. Never heard of that happening before tho. Pretty weird! 

[damric]

It's haunted.

 

Man this is a weird one. How old is the PSU? That's always a logical place to start.

[pio]

1 hour ago, damric said:

It's haunted.

 

Man this is a weird one. How old is the PSU? That's always a logical place to start.

The PSU was originally approximately 2 years old, if that.  EVGA 700BQ.  I pulled it with intents of RMA'ing due to this problem, and have swapped in a EVGA 650GT from my test bench, to find the same problem.  Possibly worse, but definitely not better.  The EVGA 650GT has never had a failure to date, and was used as my "test bench" PSU.  It was also powering Ryzen rigs at one point, its been a solid PSU.  Also about 2 years old, give or take.

[damric]

you taken everything out of the caee yet to test on a piece of cardboard  or something?  Got to start eliminating everything, fans, aios, drives, you name it

[pio]

25 minutes ago, damric said:

you taken everything out of the caee yet to test on a piece of cardboard  or something?  Got to start eliminating everything, fans, aios, drives, you name it

So far, I've got it down to board, CPU (different), RAM (different), PSU (different), GPU (different), and NVME drive installed for OS only, no SATA drives, case unhooked (its still in the case, but its all unhooked).

I think its to the point where I try swapping CPU's back around to determine if the old FX chip acts the same as it WAS, or if it acts the same as this new one is.  If my old FX chip freezes like this one's doing, I'm going to assume board failure.  If the old FX chip acts like it was, "stable" at stock with 1.55v, then I'll know to try yet another CPU before assuming board's dead, which I have a Phenom II I can toss in for troubleshooting.

 

Old parts SUCK when they die btw.  No warranty / RMA.  

Edited November 13, 2023 by pioneerisloud

[bonami2]

Dead mobo my old Intel DP45SG did that. It would even run overclocked but end up freezing randomly. He had a few trip sliding on the carpet that probably static damaged it somewhow...

Edited November 13, 2023 by bonami2

[Sir Beregond]

Given what you tried, I'm inclined to think dead motherboard too.

[damric]

pull everything from the case. we must know if the board is being shorted or bent from standoffs

[LabRat]

2 hours ago, bonami2 said:

Dead mobo my old Intel DP45SG did that. It would even run overclocked but end up freezing randomly. He had a few trip sliding on the carpet that probably static damaged it somewhow...

ESD damage is insidious.
IIRC, ESD doesn't immediately 'kill' a lot of ICs, it 'erodes' and moves elements/ions @ the nano scale.

 

TBH, I'd been wondering if +5V (pulsed/unfiltered) DC on the ground plane* (the HDMI 'dapter incident) damaged capacitors (or other components) from reverse polarity (and/or 'noisy') current.

 

*Was it ever verified if the anomalous 5VDC was actually on ground, or was it just overriding the PSU's +12VDC?

I've had a couple cheapie FP USB bays directly bridge +5VSB (motherboard-allocated to USB) to +5V; which, caused a similar symptom but, only on +5Vrail-connected devices.

 

It might be worthwhile to further-investigate the suspected cause for the failure: the HDMI adapter. I know you have a DMM (somewhere).

[neurotix]

Good luck, hope you get something working...

[damric]

42 minutes ago, neurotix said:

Good luck, hope you get something 

he'll get it working

[pio]

6 hours ago, LabRat said:

ESD damage is insidious.
IIRC, ESD doesn't immediately 'kill' a lot of ICs, it 'erodes' and moves elements/ions @ the nano scale.

 

TBH, I'd been wondering if +5V (pulsed/unfiltered) DC on the ground plane* (the HDMI 'dapter incident) damaged capacitors (or other components) from reverse polarity (and/or 'noisy') current.

 

*Was it ever verified if the anomalous 5VDC was actually on ground, or was it just overriding the PSU's +12VDC?

I've had a couple cheapie FP USB bays directly bridge +5VSB (motherboard-allocated to USB) to +5V; which, caused a similar symptom but, only on +5Vrail-connected devices.

 

It might be worthwhile to further-investigate the suspected cause for the failure: the HDMI adapter. I know you have a DMM (somewhere).

You and I both know how deathly afraid I am of hooking that stupid adapter back up.  It's in the Labrat pile for now.  

[pio]

6 hours ago, damric said:

pull everything from the case. we must know if the board is being shorted or bent from standoffs

If I'm pulling the board, it'll be when I have a replacement on hand.    Everything is in this rig so incredibly tight, push / pull fans on a 240 rad on top, cables tucked nicely (not shorting).  It was going to be my main rig when I move the 5800x / 6900XT to sim rig duties here in a few weeks.  I definitely can find SOMETHING to daily drive.  I'll daily drive Windows 98 if I have to, I'm building my sim rig.  

I really REALLY doubt its shorting on the standoffs.  Rig's been built and was running solidly for 2 years, and I've never had to pull it apart until now.  This was a failure of multiple reasons from the looks of it.  Cooked the old CPU from voltage / weak VRM's on the Gigabyte UD3, and PROBABLY hurt the board from the backfeeding +5v a week or so later (is my guess).  I'm sure power back feeding through the HDMI cable wasn't exactly assisting these old parts any at all, especially since I ran it like that (unknowing) for about a week. -_-

 

I didn't have the chance to swap CPU's as of yet to verify.  Half days at school all week, so I've got a motor mouth all afternoon preventing me from concentrating on anything.  

[pio]

Bump (please read full OP for history):

Pulled out the FX rig again last night to give it a try.  Here's how it went. -_-

 

1)  Ran it at bone stock "auto" everything.  4.2 boost, 4.0 all core (auto), around 1.35v, DDR3-1333 @ 11-11-11-30-1T timings (RAM is 1866 10-10-10-30 stock but used to pull 9-10-9 at 2000).  System installed Windows 10 just fine like this, all my main programs, and even ran Prime95 for over an hour.

2)  Okay.jpg, seems better right?  Tried overclocking it.  Just 4.5GHz, should be easy.  Freeze, 10 minutes in.  Didn't change ANYTHING else, just multiplier and upped voltage to a reasonable 1.45v.

3)  Okay, something's not right, MORE VCORE.  Slowly started raising, up to 1.55v (in increments because it kept freezing).  Freezing is just happening faster and faster now.

4).  Tried again at 4.2GHz (stock boost), but set to all core.  Still nothing but freezing, faster yet than before.

5).  End the evening with it set 110% back to all auto everything (where it passed fine earlier in the evening).  It's now freezing about 1 minute - 2 minutes in.

System is:
Sabertooth 990FX

FX 8350 (I've tried 2)
2x8GB GSkill DDR3-1866
eVGA B5 750w PSU (had a 650GT and the old 750BQ in there too)
MSI RX 580 8GB, Sapphire 290x, Sapphire 380, I could try an Nvidia card but I doubt its the GPU at this point (580 is still in it)

1x 256GB NVME SSD in a PCIe slot adapter
NO SATA INSTALLED YET

So again, I've swapped CPU's, GPU's, PSU's, and none have fixed it yet.  Does this random freezing problem sound more like a motherboard or a memory problem?  I'd imagine if it were the SSD or RAM it'd be crashing not just halting the system.  I don't get it.  And its happening faster and faster the more I use it.  I'm ALMOST thinking its caps for power delivery on the motherboard maybe from getting backfed that voltage through HDMI?  I really don't know!

ANY IDEAS?  I know its a long read.

[ENTERPRISE]

1 hour ago, pioneerisloud said:

Bump (please read full OP for history):

Pulled out the FX rig again last night to give it a try.  Here's how it went. -_-

 

1)  Ran it at bone stock "auto" everything.  4.2 boost, 4.0 all core (auto), around 1.35v, DDR3-1333 @ 11-11-11-30-1T timings (RAM is 1866 10-10-10-30 stock but used to pull 9-10-9 at 2000).  System installed Windows 10 just fine like this, all my main programs, and even ran Prime95 for over an hour.

2)  Okay.jpg, seems better right?  Tried overclocking it.  Just 4.5GHz, should be easy.  Freeze, 10 minutes in.  Didn't change ANYTHING else, just multiplier and upped voltage to a reasonable 1.45v.

3)  Okay, something's not right, MORE VCORE.  Slowly started raising, up to 1.55v (in increments because it kept freezing).  Freezing is just happening faster and faster now.

4).  Tried again at 4.2GHz (stock boost), but set to all core.  Still nothing but freezing, faster yet than before.

5).  End the evening with it set 110% back to all auto everything (where it passed fine earlier in the evening).  It's now freezing about 1 minute - 2 minutes in.

System is:
Sabertooth 990FX

FX 8350 (I've tried 2)
2x8GB GSkill DDR3-1866
eVGA B5 750w PSU (had a 650GT and the old 750BQ in there too)
MSI RX 580 8GB, Sapphire 290x, Sapphire 380, I could try an Nvidia card but I doubt its the GPU at this point (580 is still in it)

1x 256GB NVME SSD in a PCIe slot adapter
NO SATA INSTALLED YET

So again, I've swapped CPU's, GPU's, PSU's, and none have fixed it yet.  Does this random freezing problem sound more like a motherboard or a memory problem?  I'd imagine if it were the SSD or RAM it'd be crashing not just halting the system.  I don't get it.  And its happening faster and faster the more I use it.  I'm ALMOST thinking its caps for power delivery on the motherboard maybe from getting backfed that voltage through HDMI?  I really don't know!

ANY IDEAS?  I know its a long read.

 

The degradation you describe in length of time of stability whether stock or overclocked sounds like the motherboard to me.  I had a quick read through and perhaps I missed it but did you have any other RAM to swap out ? 

 

Have you ran Memtestx86 on the RAM in the possibly faulty motherboard? Interested to see if it shows any issues. 

 

I doubt its the Nvme but you could run some bench tests outside of the Windows environment and excluding your Nvme to see if the system remains stable.  Have you heard of Medicat ? 

 

Medicat USB | Computer Diagnostic and Recovery Toolkit

MEDICATUSB.COM

---

A toolkit that helps compile a selection of the latest computer diagnostic and recovery tools.

 

 

I would look it over as it has a great set of tools. 

 

 

 

 

[damric]

Sounds like SSD controller possibly. I would rule that out by trying another SSD

Edited December 19, 2023 by damric

[pio]

Alright I'll try a different SSD next.  I do have other RAM I can try too, I really don't think its RAM but I can swap it out.  Normally I'd expect RAM to cause a BSOD not a whole system lockup.  SSD, I've never had a SSD fail before so I wouldn't know what to expect there.  So yeah I'll try another SSD first.  Just grrr. -_-

[neurotix]

Yeah this must be pretty terrible for you. I have issues with my 2nd CCD on my 7900X3D not boosting properly in benchmarks (it stays at 5.2 instead of going to 5.6) and I've tried everything to solve it, worked with Asus as well as AMD and nothing has fixed it. It sucks when you have a problem with a rig that you cannot fix.

[pio]

Okay, so I can confirm its the motherboard.

Swapped SSD's out this evening, and it still won't run without freezing completely.  I was able to get Windows installed (all settings auto / stock), but it froze during installation of programs.  

I could swap RAM out, but I really REALLY doubt its RAM.  It's not BSOD'ing or crashing.  It's locking up completely.

Yup, looking like after Christmas I might be ebay searching for another good AM3+ board now.  I don't trust the Gigabyte I have with an FX (besides, its running my Phenom II 1100T happily).  FX was definitely a part of my "collection".  Sigh. -_-  

So let this be a lesson guys, be wary of cheap adapters for HDMI things. -_-  I'm 99.9% sure that backfeeding HDMI cable is what caused my Sabertooth's failure here.

[UltraMega]

1 hour ago, pioneerisloud said:

Okay, so I can confirm its the motherboard.

Swapped SSD's out this evening, and it still won't run without freezing completely.  I was able to get Windows installed (all settings auto / stock), but it froze during installation of programs.  

I could swap RAM out, but I really REALLY doubt its RAM.  It's not BSOD'ing or crashing.  It's locking up completely.

Yup, looking like after Christmas I might be ebay searching for another good AM3+ board now.  I don't trust the Gigabyte I have with an FX (besides, its running my Phenom II 1100T happily).  FX was definitely a part of my "collection".  Sigh. -_-  

So let this be a lesson guys, be wary of cheap adapters for HDMI things. -_-  I'm 99.9% sure that backfeeding HDMI cable is what caused my Sabertooth's failure here.

Called it 

[Bones]

On 20/12/2023 at 02:05, pioneerisloud said:

Okay, so I can confirm its the motherboard.

Swapped SSD's out this evening, and it still won't run without freezing completely.  I was able to get Windows installed (all settings auto / stock), but it froze during installation of programs.  

I could swap RAM out, but I really REALLY doubt its RAM.  It's not BSOD'ing or crashing.  It's locking up completely.

Yup, looking like after Christmas I might be ebay searching for another good AM3+ board now.  I don't trust the Gigabyte I have with an FX (besides, its running my Phenom II 1100T happily).  FX was definitely a part of my "collection".  Sigh. -_-  

So let this be a lesson guys, be wary of cheap adapters for HDMI things. -_-  I'm 99.9% sure that backfeeding HDMI cable is what caused my Sabertooth's failure here.

Grab another Asus or another Gigabyte, those are the only two brands that made an AM3+ board than can run anything without issue.
If you know you'll be doing alot of Super PI and things like that the UD series like the one you have now is fine but I do suggest either an UD5 or UD7 if you can find one.
UD3 is OK but not the best for larger chips and unfortunately, all UD's regardless do tend to run very warm to hot with their VRM setups.

For an Asus you have a few choices:
Crosshair V (Original) Crosshair V-Z, Sabertooth 2.0 or the newer 3.0 version (Has RGB LED's under the PCI-E latches) and it also has an m.2 drive port as well.

I know your Sabertooth 2.0 died but I'll also say due to your statement about the HDMI adapter that could have been the cause and I'm not aware of any board that can hold up to backfeed for long - It just gets nasty when that happens.
And to that end the old 2.0 I have here (And posting with now) inspite of being ran hard and put up wet right out of the box onward as an open box item from the egg has really held up well, even pushing a few chips to or over 8GHz, over the course of about 7 years of abuse that way.

I did mention the other morning about it always thinking a new chip has been installed and it was finally retired from benching because of that. 
Aside from that it still works fine and useable for daily rig use.

If possible I'd look for one of the newer 3.0 Sabertooths.
Great board and it's a real treat to work with. Yes, it too can run a 9590 without issues from the VRMs unlike a Gigabyte could run into, which is why I never ran one of those, going all Asus here.
If you can find a Crosshair V-Z that's the ticket out of all but I'd be picky about the price no matter what you decide on.

As I had told you it's possible to find one for a good price but I'd pay no more than about $150 for the average Crosshair V-Z found, even those in great condition aren't worth $200+ or more because they sold for around $225-250 NEW and I should know, considering I bought at least two that way back then.
I will say concerning the newer Sabertooth 3.0's like mine, those can be pricey due to their rarity and that's more of a "Take what you can get" kind of thing because of it -But I'd still have a price range in mind and not overpay for one.

Hit me up if you have any questions - You know where to find me.

Edited December 30, 2023 by Bones

[pio]

7 hours ago, Bones said:

Grab another Asus or another Gigabyte, those are the only two brands that made an AM3+ board than can run anything without issue.
If you know you'll be doing alot of Super PI and things like that the UD series like the one you have now is fine but I do suggest either an UD5 or UD7 if you can find one.
UD3 is OK but not the best for larger chips and unfortunately, all UD's regardless do tend to run very warm to hot with their VRM setups.

For an Asus you have a few choices:
Crosshair V (Original) Crosshair V-Z, Sabertooth 2.0 or the newer 3.0 version (Has RGB LED's under the PCI-E latches) and it also has an m.2 drive port as well.

I know your Sabertooth 2.0 died but I'll also say due to your statement about the HDMI adapter that could have been the cause and I'm not aware of any board that can hold up to backfeed for long - It just gets nasty when that happens.
And to that end the old 2.0 I have here (And posting with now) inspite of being ran hard and put up wet right out of the box onward as an open box item from the egg has really held up well, even pushing a few chips to or over 8GHz, over the course of about 7 years of abuse that way.

I did mention the other morning about it always thinking a new chip has been installed and it was finally retired from benching because of that. 
Aside from that it still works fine and useable for daily rig use.

If possible I'd look for one of the newer 3.0 Sabertooths.
Great board and it's a real treat to work with. Yes, it too can run a 9590 without issues from the VRMs unlike a Gigabyte could run into, which is why I never ran one of those, going all Asus here.
If you can find a Crosshair V-Z that's the ticket out of all but I'd be picky about the price no matter what you decide on.

As I had told you it's possible to find one for a good price but I'd pay no more than about $150 for the average Crosshair V-Z found, even those in great condition aren't worth $200+ or more because they sold for around $225-250 NEW and I should know, considering I bought at least two that way back then.
I will say concerning the newer Sabertooth 3.0's like mine, those can be pricey due to their rarity and that's more of a "Take what you can get" kind of thing because of it -But I'd still have a price range in mind and not overpay for one.

Hit me up if you have any questions - You know where to find me.

Well, I won the offer on that Crosshair V with the FX 6200 combo.  $100 shipped (plus tax).  So I'll be putting a CHV in my FX rig soon, I'll just live with one of the two 8350's.  I'm sure one of the two CPU's probably still works fine.  I'll probably buy a better AIO cooler for it before I'll buy a FX 9590.  Maybe the 8350 will clock sufficiently for me again.

Edited December 31, 2023 by pioneerisloud

[Bones]

15 hours ago, pioneerisloud said:

Well, I won the offer on that Crosshair V with the FX 6200 combo.  $100 shipped (plus tax).  So I'll be putting a CHV in my FX rig soon, I'll just live with one of the two 8350's.  I'm sure one of the two CPU's probably still works fine.  I'll probably buy a better AIO cooler for it before I'll buy a FX 9590.  Maybe the 8350 will clock sufficiently for me again.

The one I showed you the other night?
That's cool - And did you try an offer as I had suggested or just won it outright?

Should be good for what you will be using it for.

Yeah..... You'll need water for a 9590 chip when you get one, but for any non-XOC use an 8350 is just as good - Maybe better due to lower overall thermals but it CAN be tweaked down to help with that since the chip's default CPU voltage specs are really high in the first place.

Clock it down a bit and drop the voltage alot by comparison, that's how I did mine. 
At one time I had it running with my Susanoo aircooler around 4.4-4.5GHz and it was fine.