Received by 3090 FE today - having weird issue with my z390 asrock taichi mb - "b2" error code

Why would using 2 cables be better than 1 cable anyway?
Each GPU end of the cable has 8 wires connected to it. They aren't daisy chained. They come straight from the PSU. The only difference should be that using separate cables won't use a shared plastic connector.
 
Last edited:
I spoke too soon. My new motherboard worked with the 3090 FE for a few hours. Then I turned my computer off for the night. when I woke up, it would not boot - this new motherboard (msi) gives an error of "VGA - indicates GPU is not detected or fail." - if I put my 2080ti in, it boots just fine. I can't seem to clear the error with the 3090 now at all, fans turn on, but the Geforce LED lighting on it doesn't light up either (maybe normal since it can't boot). I swapped out power supplies for a 1200w just to be sure, same result. Started the RMA process with nvidia. ugh.

kind of odd that it worked for awhile, and now doesn't after being powered off all night.
 
I spoke too soon. My new motherboard worked with the 3090 FE for a few hours. Then I turned my computer off for the night. when I woke up, it would not boot - this new motherboard (msi) gives an error of "VGA - indicates GPU is not detected or fail." - if I put my 2080ti in, it boots just fine. I can't seem to clear the error with the 3090 now at all, fans turn on, but the Geforce LED lighting on it doesn't light up either (maybe normal since it can't boot). I swapped out power supplies for a 1200w just to be sure, same result. Started the RMA process with nvidia. ugh.

kind of odd that it worked for awhile, and now doesn't after being powered off all night.
Not surprised it was the card all along. Mobo rarely is the problem if your other card works flawlessly. Sucks tho that you gotta RMA now.
 
Not surprised it was the card all along. Mobo rarely is the problem if your other card works flawlessly. Sucks tho that you gotta RMA now.
weird that removing the cmos battery seemed to make it work again for him though. Curious problem to say the least.
 
Why would using 2 cables be better than 1 cable anyway?
Each GPU end of the cable has 8 wires connected to it. They aren't daisy chained. They come straight from the PSU. The only difference should be that using separate cables won't use a shared plastic connector.
Refer to DejaWiz post. If only using one cable then each of the 3 12V cables is running 10 amps, the max amps for 18AWG and probably double the spec for any lead on PSU side (PCI-E spec probably calls for max 4-6 amps per 12V lead since one cable only spec'd for 150w). Maybe fine for short loads, but I wouldn't risk it as for sustained loads (multi hour game session with uncapped frame rate) may melt the wire or connector leads causing a short. Though I'm no electrician or electrical engineer, just going off basics I know from home wiring.
 
Refer to DejaWiz post. If only using one cable then each of the 3 12V cables is running 10 amps, the max amps for 18AWG and probably double the spec for any lead on PSU side (PCI-E spec probably calls for max 4-6 amps per 12V lead since one cable only spec'd for 150w). Maybe fine for short loads, but I wouldn't risk it as for sustained loads (multi hour game session with uncapped frame rate) may melt the wire or connector leads causing a short. Though I'm no electrician or electrical engineer, just going off basics I know from home wiring.
If each 8 pin carries 150w for a total of 300w delivered, that is 25A in a 12V system. 25/3 conductors = 8.3A per conductor. It is tough to find amp ratings for wire as short as PSU leads, but I would guess using a single lead instead of two would have slightly more voltage drop end to end. It is not a matter of overloading the wire with current though. There is no risk of that, the wires and connectors are in spec, or they wouldn't exist. The use of two separate cables is recommended for optimal performance, not for safety. If there were any liability whatsoever, no PSU maker would make dual connector cables (the no name ones with bogus ratings aside).
 
I spoke too soon. My new motherboard worked with the 3090 FE for a few hours. Then I turned my computer off for the night. when I woke up, it would not boot - this new motherboard (msi) gives an error of "VGA - indicates GPU is not detected or fail." - if I put my 2080ti in, it boots just fine. I can't seem to clear the error with the 3090 now at all, fans turn on, but the Geforce LED lighting on it doesn't light up either (maybe normal since it can't boot). I swapped out power supplies for a 1200w just to be sure, same result. Started the RMA process with nvidia. ugh.

kind of odd that it worked for awhile, and now doesn't after being powered off all night.
have you tried turning csm off in bios and looking to see if there is a bios update? you seemed to just to "swapped board" and havent even said what they were.
 
yeah the BIOS on both boards was the latest version. The new one is MSI Gaming Edge (z490 w/ new processor as well). I took the card out for a day, getting ready to RMA it back.. decided to put it back in the computer for 1 more test before I sent it of....and now its been working fine for 2 days, even after running stress tests. I have no idea what to think now.
 
Cable from video card to monitor.
it's not that.. i ran into a similar issue with a friends 1080ti that would randomly throw the B2 error code, i've never been able to figure out why it does and still does after being in that system for almost a year now. out of the blue it'll just decide that it no longer wants to be detected, clear cmos works fine for a week+ then just decides it doesn't want to work again. i definitely think it's an issue with the gpu but never was able to pinpoint it. the only board that absolutely refused to work period with it was an asrock x370 taichi, current board it's on is an msi tomahawk b450 and worked fine on my x570 elite for the two times i booted up with that card installed.

yeah the BIOS on both boards was the latest version. The new one is MSI Gaming Edge (z490 w/ new processor as well). I took the card out for a day, getting ready to RMA it back.. decided to put it back in the computer for 1 more test before I sent it of....and now its been working fine for 2 days, even after running stress tests. I have no idea what to think now.

try pendragons fix but personally if you have the option to RMA it i would.
 
Has anyone tried with just 1 cable and did any benchmarks and or checked voltages and such? I always see reviewers use 2 separate cables when benching.
 
After having some issues with Borderlands, I started running some tests. My 3090 has boosted my entire system's temperature by around 10%. With no other changes. That's with some pretty impeccable cable management and airflow in my Carbide 540. I'm tinkering with some things and am about to re-apply my cooler and thermal paste, but this might be happening to others.
 
After having some issues with Borderlands, I started running some tests. My 3090 has boosted my entire system's temperature by around 10%. With no other changes. That's with some pretty impeccable cable management and airflow in my Carbide 540. I'm tinkering with some things and am about to re-apply my cooler and thermal paste, but this might be happening to others.
have you tried undervolting?
 
have you tried undervolting?

Nope. I only just discovered this yesterday night. Right now I mainly want to make sure it's the card doing it and not something else. I have some new thermal paste on the way from Amazon right now. I'd kinda rather not undervolt the card if I can help it.
 
If each 8 pin carries 150w for a total of 300w delivered, that is 25A in a 12V system. 25/3 conductors = 8.3A per conductor. It is tough to find amp ratings for wire as short as PSU leads, but I would guess using a single lead instead of two would have slightly more voltage drop end to end.
Adding to your post to help general understanding:

As a cable and the connections at each end heat up due to current flowing through them, the overall resistance increases slightly (more than the R increase already added).
The more current flowing, the higher this temperature increase and thus the higher the resistance, varying the voltage supplied to the gfx card.
This alone negatively impacts the power supplied using a single cable vs dual cables.
But on top, so does the rapidly changing current flow caused by the rapidly changing load of a GPU and its memory.

When less current flows through a resistance, the voltage drop is lower. Or the higher the current flow, the more voltage will drop across this resistance.
So as the current flow changes rapidly with the GPU load, the voltage drop across the cable connection also fluctuates rapidly.
This stops the expensive, very stable power supply from being so stable.
When pushing the boundaries of capability (such as with already overclocked 3080/90 cards as bought) this starts to matter more.
While it can work fine with some cards, others might exhibit stability problems using two power outs from a single cable.
Double the current flow = much higher power fluctuation level.
AND
with the 2 connections supplying different areas on the card/GPU that could be on different clocks, the change in current flow will not necessarily be the same for both, increasing the rate of voltage changes over that single cable.

Summing up:
A single cable supplying 2 connections to a gfx card will provide noisier power and thus less stable power than using 2 cables.
Whether that matters depends how close to the edge the gfx card is being pushed.

Also, a bad connection has a much more profound effect when supplying more current, the heat created can be extreme.
I burned a connection to my Corsair AX750 PSU when the gfx card wire wasnt plugged into it properly.
I only found out after my PC shut off while gaming, I sniffed around for any issues and found the connector had melted after it wasnt properly clipped in place.
Luckily, after cleaning it up, the plug clipped in place ok and still worked fine. The same PC is still running well a good 3 years+ later.
If this GPUs PCIe wire had been doing double duty while being maxed out, it could have been much worse.
 
Back
Top