GPU configuration causing no-boot

[Cereal5]

Hey guys, I recently purchased a used computer. I've been having some issues with PCIe slots and GPUs. Specs:

Xeon E5-2620 V3

Some cheap cooler (BIOS and Afterburner reading good temps)

Asus Z10PA-U8

4x4GB kit of some RAM, no brand name, no heat spreader

ADATA SU800 256GB SSD (brand new)

Rosewill Photon 1200 (brand new)

GPUs tried: GTX 1080 FE, Asus GTX 1080 Strix, Gigabyte RTX 2070 Super Windforce

 

Only the 1080 FE will boot properly, and whenever I try installing a second card, it will either boot to desktop and show the second GPU as a "basic display adapter", hang on windows, or hang on POST with a PCIe bus enumeration error code (code 94) with the two 1080s, or code 40 with the 2070 super (some kind of waking from a sleep state, but windows never loads and doesn't "wake" when I move the mouse or hit a key). Both of the 1080s are confirmed good, and I just pulled the 2070 super out of the box to put it in this computer, but based on the results I'd say it's not the GPU that's causing the issue.

 

I have tried all GPUs in all slots, all combinations of GPUs, I've tried no GPUs and gotten occasional RAM error codes, sometimes it will boot just fine. I've reset CMOS, updated the BIOS. The only thing I can think of that I haven't done is check the socket for bent pins. Am I missing anything here?

 

Thanks in advance.

[Joseph Stacklin]

Have you already reached out to the Asus support team or is the board out of warranty?

 

[Cereal5]

1 minute ago, Joseph Stacklin said:

Have you already reached out to the Asus support team or is the board out of warranty?

 

Long out of warranty. This is haswell, dog

[Joseph Stacklin]

Didn't pay attention to CPU or pull it up in ARK, sorry.

 

Have you confirmed that it isn't your top PCIe lane? IE: trying other lanes after resetting the BIOS to default?

[Cereal5]

Just now, Joseph Stacklin said:

Didn't pay attention to CPU or pull it up in ARK, sorry.

 

Have you confirmed that it isn't your top PCIe lane? IE: trying other lanes after resetting the BIOS to default?

The top PCIe Lane is the one that isn’t causing problems. The problem starts when I install a second GPU in the second slot

[Joseph Stacklin]

18 minutes ago, Cereal5 said:

The top PCIe Lane is the one that isn’t causing problems. The problem starts when I install a second GPU in the second slot

Power delivery is good? (Tested with multiple PSUs?) Because according to ARK your CPU can handle up to 40 lanes of PCIe

[Cereal5]

9 minutes ago, Joseph Stacklin said:

Power delivery is good? (Tested with multiple PSUs?) Because according to ARK your CPU can handle up to 40 lanes of PCIe

I don’t have any other PSUs to test with. Well that’s not true, I have some non-modular POS that came with what I bought but I’m not plugging that in. The way this PSU is, it has two PCIe 8+8 leads off of one single 12 pin connector, and it powers one GPU at 100% load just fine, so powering two at idle shouldn’t be a problem in comparison

[Joseph Stacklin]

5 minutes ago, Cereal5 said:

I don’t have any other PSUs to test with. Well that’s not true, I have some non-modular POS that came with what I bought but I’m not plugging that in. The way this PSU is, it has two PCIe 8+8 leads off of one single 12 pin connector, and it powers one GPU at 100% load just fine, so powering two at idle shouldn’t be a problem in comparison

The only reason I ask is because rosewill has left a bad taste in my mouth with their products. If you can, I would really try powering the board with one PSU and the GPUs with another. It's jank but it should help you determine if your load is too much.

[Cereal5]

7 minutes ago, Joseph Stacklin said:

The only reason I ask is because rosewill has left a bad taste in my mouth with their products. If you can, I would really try powering the board with one PSU and the GPUs with another. It's jank but it should help you determine if your load is too much.

It’s possible yes but I doubt it due to the RAM post codes I’ve received while trying to get it to post originally, it’s a top rated PSU (not saying you can’t get a bad product every now and then). But it’s a 1200W PSU, I really REALLY hope it can handle a couple of pascal cards at idle. I did mention that it has gotten into windows before with just a slightly longer post (kinda expected from a server board), and it just shows up as a “basic display adapter”. Lights and fans and all that come on like normal. So it’s definitely getting PSU power and PCIe information, just not enough PCIe information.

 

I had a similar problem with an Asus Z370-A motherboard that ended up being what I believe was a chipset problem, that caused certain memory configurations to cause the system to hang and no-start.

 

I don’t really know where I’m going with this. I’m not plugging that other PSU into any of my hardware, that’s for sure. 

[Joseph Stacklin]

1 hour ago, Cereal5 said:

It’s possible yes but I doubt it due to the RAM post codes I’ve received while trying to get it to post originally, it’s a top rated PSU (not saying you can’t get a bad product every now and then). But it’s a 1200W PSU, I really REALLY hope it can handle a couple of pascal cards at idle. I did mention that it has gotten into windows before with just a slightly longer post (kinda expected from a server board), and it just shows up as a “basic display adapter”. Lights and fans and all that come on like normal. So it’s definitely getting PSU power and PCIe information, just not enough PCIe information.

 

I had a similar problem with an Asus Z370-A motherboard that ended up being what I believe was a chipset problem, that caused certain memory configurations to cause the system to hang and no-start.

 

I don’t really know where I’m going with this. I’m not plugging that other PSU into any of my hardware, that’s for sure. 

I'm not sure. It sounds to me that you are kinda stuck because you are unable to troubleshoot with equipment you feel "stable" enough to test with. Honestly I have seen this happen to me a couple times on my systems (Ran Xeons a couple times, they are power hogs) and one of three things happened:

1. PSU was defective/wasn't pushing clean enough power/was hitting some sort of power cap and was unable to properly deliver power
2. PSU cables weren't seated properly, causing a micro short which fried several components on my chipset and bridges
3. The power load was too much and to prevent over-volting the system, it kicked off multiple times until I reduced the load this lead me to purchase a higher wattage PSU.

My guess is still power delivery is messed up. If you can afford a couple (around $100 total) decent 750W units ( or a single 1500W S-Grade PSU) and test with one plugged into the GPUs and one into your board, then you should be able to better troubleshoot your issue. If it isn't the PSU, you most likely can take back the 750W units and continue your troubleshooting with other components.

[Joseph Stacklin]

null, double post