A4-SFX Upgrade Failure

[Jawbreaker]

I was rocking a 9900K for a few years up until my rig stopped booting, I was able to get it back up and running but also saw this as an opportunity to upgrade my PC which I once called my dream build. I jumped in headfirst with no research and bought whatever caught my eye (you'll be able to tell from the specs.) I am now going on a week trying to get my new rig to run without crashing. Send help!

A4-SFX V4.1
ROG Strix B760-I Gaming Wifi
G.Skill 64GB (32x2) 6000mhz
Intel i7-13700
Asetek 645LT AIO
Corsair SF750
EVGA 3080 XC

The upgrades I made were to the motherboard, RAM, and CPU, nothing else.

A few things to note:
- The AIO has since been adjusted so that the tubes are lower, unlike this pic.
- I snagged a new 645LT on eBay in case it's my current AIO that's underperforming due to age, waiting for it to be delivered.
- Already ordered a Thermalright Anti-Bending Buckle, waiting for that to be delivered.
- I originally bought and used the 12700k, I had thought getting a non-K CPU with a much lower TDP would solve my issue, this is also strange considering 13700's TDP is lower than 9900K's (65w vs 95w, respectively.) So I returned the 12700K and got the 13700 instead.
- I am still trying to figure out undervolting, I have the sneaking suspicion B760-I isn't the best choice for undervolting.
- CPU idles at 50C on desktop and even freezes on desktop sometimes, that could be another issue entirely.
- Can't play a decades old game like Halo Reach for over a minute without freezing and buzzing before having to shut it off. My last 9900k build also did that, but only after at least 30 minutes and only if I wasn't cranking up the GPU fans via Precision X1, otherwise 9900K had no issues while gaming.
- Ditching the AIO is not something I considered or want, but I may have to just suck it up depending on whether this build can be made to work.

Either something is seriously wrong with my build or the era of modern A4-SFX gaming is reaching it's end. Please tell me I don't have to go back to Micro Center a third time to swap parts.

 

[confusis]

My wager here would be on memory - are you running any XMP settings?

 

[Jawbreaker]

My wager here would be on memory - are you running any XMP settings?

Tried both Auto and XMP. Tried one RAM stick just to see what happens. The result is the same any which way.

 

[confusis]

Is the memory on the QVL list? https://rog.asus.com/motherboards/rog-strix/rog-strix-b760-i-gaming-wifi-model/helpdesk_qvl_memory/


Modern boards seem to be persnickety around RAM compatibility

 

[Jawbreaker]

Is the memory on the QVL list? https://rog.asus.com/motherboards/rog-strix/rog-strix-b760-i-gaming-wifi-model/helpdesk_qvl_memory/


Modern boards seem to be persnickety around RAM compatibility

The only memory that shows up on that list when searching by vendor and kit is:

G.SKILL

F5-5600J2834F32GX2-TZ5RS

2x 32GB

XMP

5600

5600

DS

SK Hynix

28-34-34-89

1.35

1,2

The kit I have in the system right now is:

G.Skill Trident Z5 RGB Series 64GB (2 x 32GB) DDR5-6000 PC5-48000 CL30 Dual Channel Desktop Memory Kit F5-6000J3040G32GX2-TZ5RK - Black​

XMP1 Profile is shown in BIOS, with it running at 5600.

Today, I went out and bought some RAM to swap out test if it'd work. It didn't, same issue. The RAM I went out and bought and tested, and is listed in the link provided is:

CORSAIR

CMH32GX5M2B5600C40 Ver 4.43.02

2x 16GB

XMP

5600

5600

SS

Samsung

40-40-40-77

1.25

1,2

The Thermalright anti-buckle frame was delivered today, I installed it and it resulted in cooler temps but the system still crashed quickly in game.

 

[REVOCCASES]

I assume it's a fresh Windows installation, latest BIOS and you've already tried BIOS default settings?

just a wild guess, but the PCIe riser cable somehow looks like the older PCIe Gen3 one... maybe you could try to set PCIe to Gen3 in BIOS and see what happens...

 

[Legion]

PL 2 may also be the culprit here unless you've already changed it.

By default the 13700 power values are PL1 set at 65w. PL2 however is 219w.
It's likely here that the CPU is constantly trying to boost with PL2 and that dinky 645 92mm AIO is not going to cope with that at all.

I'd strongly suggest you change the PL2 asap to somewhere in the region of 100w. You have to enter that value manually in the bios.

Have you flashed the motherboard with the latest bios? That can also increase compatibility with ram kits.

Set PL2 to 100w and retest.

And as Revo suggested earlier, set the PCI-e lanes on the gfx slot to Gen 3 and see if that changes anything.

Edit: Also turn off Multi Core Enhancement (MCE) in the bios. Just saw it was an Asus board. With MCE turned on the board will constantly try to boost the CPU to the PL2 limits.

 

[tinyitx]

I was rocking a 9900K for a few years up until my rig stopped booting, I was able to get it back up and running

The upgrades I made were to the motherboard, RAM, and CPU, nothing else.

- Can't play a decades old game like Halo Reach for over a minute without freezing and buzzing before having to shut it off. My last 9900k build also did that, but only after at least 30 minutes and only if I wasn't cranking up the GPU fans via Precision X1, otherwise 9900K had no issues while gaming.

From what OP said above, I suspect the problem with this upgrade might have its root in the old system.
Something was not right and now got carried into this 'semi' new system?

1. What did OP do to get the pre-upgrade PC back up and running back then?

2. ''at least 30 minutes (of gaming?)....cranking up the GPU fans'' seems to be a 'heat accumulation' problem?
Too much heat and is not dissipated fast enough? Is the AIO working properly? Pump working at proper rpm?

 

[dondan]

Random shutdowns or freezes could be the SSD or PSU.

 

[Jawbreaker]

I assume it's a fresh Windows installation, latest BIOS and you've already tried BIOS default settings?

just a wild guess, but the PCIe riser cable somehow looks like the older PCIe Gen3 one... maybe you could try to set PCIe to Gen3 in BIOS and see what happens...

Fresh Win11 install, latest BIOS, and yes I have.

The PCIe tip is a good one, I never would have considered that. I'll try that next.

Set PL2 to 100w and retest.

Edit: Also turn off Multi Core Enhancement (MCE) in the bios. Just saw it was an Asus board. With MCE turned on the board will constantly try to boost the CPU to the PL2 limits.

Short and long power delivery windows right? I have them set to 65w. I'll need to re-check if B760-I has MCE because I don't remember seeing it.

1. What did OP do to get the pre-upgrade PC back up and running back then?

2. ''at least 30 minutes (of gaming?)....cranking up the GPU fans'' seems to be a 'heat accumulation' problem?
Too much heat and is not dissipated fast enough? Is the AIO working properly? Pump working at proper rpm?

I honestly couldn't say. That was a different can of worms. It came to the point where it would only boot if I turned the PSU power switch off, then back on, then pressed the PC's power button. Prior to discovering that, I had taken the thing apart, unplugged CMOS, I can't recall everything else I tried but it was a huge hassle which resulted in me going for this upgrade.

AIO pump shows as running max speed (5000+) in BIOS.

Random shutdowns or freezes could be the SSD or PSU.

This board has a M.2 slot beneath the motherboard so it's literally sandwiched between the GPU and motherboard, possibly the m.2s are overheating?

My SF750 is about 3 years old at this point. I don't know the longevity of psus.

I've done more troubleshooting since my last post:

- Was able to run furmark for 20 minutes without crashing, CPU temp was at a constant 75c
- Did a clean boot where the only service/app ticked was steam, launched the game and the issue was still there
- After installing the thermalright buckle, I booted the system cold and went straight to a game to test, it still crashed, frozen screen showed CPU temp at 57c

 

[Legion]

This board has a M.2 slot beneath the motherboard so it's literally sandwiched between the GPU and motherboard, possibly the m.2s are overheating?

My SF750 is about 3 years old at this point. I don't know the longevity of psus.

I've done more troubleshooting since my last post:

- Was able to run furmark for 20 minutes without crashing, CPU temp was at a constant 75c
- Did a clean boot where the only service/app ticked was steam, launched the game and the issue was still there
- After installing the thermalright buckle, I booted the system cold and went straight to a game to test, it still crashed, frozen screen showed CPU temp at 57c

Is that rear NVME drive Gen 4 ?
They get hot enough without doing that to it in an A4 SFX hehe.
Regardless, I'd either remove it completely from the rear slot and either replace it with a Gen 3 drive which run much cooler anyway, or just use a sata SSD.

Some further elimination required here with testing as well:
CPU first with Cinebench R23 for a good 10 mins looping.
Then run TestMem 5 or Karhu ram test and see if the ram craps out in testing.
You don't need to do this for hours and hours, it's just quick and dirty fault finding.

If the CPU and ram don't crap out in those tests, we've eliminated those.
The rear NVME drive could also be causing this, remove and test without it.

See how you get on and report back

 

[Jawbreaker]

Is that rear NVME drive Gen 4 ?
They get hot enough without doing that to it in an A4 SFX hehe.
Regardless, I'd either remove it completely from the rear slot and either replace it with a Gen 3 drive which run much cooler anyway, or just use a sata SSD.

CPU first with Cinebench R23 for a good 10 mins looping.
Then run TestMem 5 or Karhu ram test and see if the ram craps out in testing.

See how you get on and report back

Rear NVME drive was a inland PCie 3.0 drive. I removed it and played a game and it still crashes.

Both Cinebench R23 and TestMem 5 ran without issue.

 

[Legion]

Rear NVME drive was a inland PCie 3.0 drive. I removed it and played a game and it still crashes.

Both Cinebench R23 and TestMem 5 ran without issue.

Cpu, ram and rear NVME eliminated.

Ok, it's onto the Gfx now then.

You're going to have to strip the system down and build it back up outside the case and plug the 3080 directly into the motherboard.
Just do it on a cardboard box preferably, do not do it sat on carpet or fabrics.

Something you just have to go through here while trying to find out the cause of your problems I'm sorry to say.
If it's still crashing outside of the case and plugged directly into the motherboard, we've eliminated riser gone bad.
If after you build it outside the case and it's still crashing there's 2 possibilites left.

It's the 3080, or the SF750.

Those EVGA 3080 XC have 2, 8 pin connectors for power right?
So you want 2, single, 8 pin power cables plugged into that 3080 (not 1 and a splitter)
In the box of the SF750 you get 2 PCIE-e 8 pin cables with splitters on the end of each, are you using any of the cables with the splitter part plugged in to power the 3080?
If you are unsure, you can always post a pic to show us.

I can also see custom cables being used here, use the cables that came with the SF750 for any futher testing.
Especially with the 3080, they are hot running power hungry things. If you've bought and are using custom (cheap chinese) PCI-e cables that just aren't up to the power draw of a 3080, it's a safe bet this isn't helping you either.

The SF750 should still be good at 3 years old.

Test and report back

 

[Jawbreaker]

Cpu, ram and rear NVME eliminated.

I've made some progress after trying different ssds.

First I realized a grave mistake: I did not remove the film from the thermal pad before installing what would be my main drive. I removed it, hoped all was fixed, tested with a game and it still crashed. I'm guessing since I tested so soon after a previous crash, the m.2 was still too hot.

I removed both m.2s (inland and 990) then installed the old m.2 (970) from my previous build. Although it's not optimized since the win10 install on it was used with a different motherboard and parts, I could still run steam. The m.2 (inland) that was in the rear slot contained most of my steam library, I took that and put it in a m.2 enclosure and connected it via USB C. Still on this Win10 drive, I added the steam library from this external drive so I could play a game and see if it crashes. It did not crash. (yay!)

So I think it was the main m.2 (990) heating up that was the culprit. I did a fresh install of Win11 on the problem m.2, installed a game on it, played a long enough session and it didn't crash. So at this point it seems the issue is fixed.

I then played a more demanding game, and it did end up crashing but not as quickly as before. I feel as though if I were to turn up the GPU fans I'll be fine, just like I used to in my last build. If this is the case, I'm happy. From here I'm going to look at still improving my build. I still have a new 645LT to install, I want to replace the gen3 riser with a gen4 one, I need to figure out undervolting, and then also ensure power draw for the CPU is properly configured. I may down the line try the noctua fan with vent route to see if it the temp/noise trade off is worth it. I appreciate everyone's help! Legion especially!

EDIT: It crashed at desktop. 990 M.2 permanently damaged from repeated overheating maybe?

EDIT2: Did more testing, issue isn't fixed. Crashes in-game and occasionally on desktop no matter which m.2 drive I'm running off of. I'll have to set aside a day to do the cardboard bench pc testing.

EDIT3: It was the riser cable. Connected gpu straight to the board, it worked without a hitch.

 

[STALKER-SVK]

I have the same MB as you, does the asetek AIO backplate collide with components on the back of the MB? and I didn't know that the asetek has also LGA1700 version of this cooler

 

[Jawbreaker]

I have the same MB as you, does the asetek AIO backplate collide with components on the back of the MB? and I didn't know that the asetek has also LGA1700 version of this cooler

It doesn't. The only component back there is the m.2 slot and the bracket stays clear of it. There isn't a LGA1700 version per se, since the aio can be used on most boards so long as you have the corresponding retention kit for it.

 

[STALKER-SVK]

I'm thinking about getting the asetek 645LT for my i5-13500, now I'm using the thermalright AXP90-X47 full copper and the fan is very close to the side panel so it's noisy, what TDP can that small AIO handle?