New 3080 and suddenly problems

opethdisciple · 24 Jan 2021 at 13:47

I am currently working so haven't had a chance to see what happens when I put things back to stock.

Will do later and use PC for a few days and see what happens.

ltron · 24 Jan 2021 at 13:50

opethdisciple said:
How is that the problem?

I mean I have two separate cables running from the PSU to the adapter that goes in the 3080 FE.

I'm pretty sure that's what your meant to do.

If I switch it to single rail, will I need to change the wiring?

It's not a problem, separate cables mean it couldn't possibly be a problem. This myth about multi rail CPUs being a problem goes back over ten years when multi rail PSUs were new technology and some manufacturers designed them so poorly that there were issues as it was easy to overload a rail (single rail is now used as clever marketing by manufacturers as they know the early issues with multi rail are burned on enthusiasts' brains; of course single rail is easier and cheaper for them to design and manufacture while offering fewer protections for the user, so it suits them down to the ground). This is not an issue since over ten years ago and certainly not from reputable brands.

ltron · 24 Jan 2021 at 13:55

HWInfo now has a handy voltage readout for each GPU power connector and the PCIE slot. Can you post a screenshot at idle and another one after the PC has been under gaming/graphics load e.g. 3DMark?

Nerusy · 24 Jan 2021 at 13:57

As was previously said, you have both RAM and CPU overclocked. First get them to stock to see if games are stable.

Also as was mentioned, 3080 puts out a lot of heat. My CPU without 3080 sits at 60c during load, my CPU with 3080 installed sits at 75c (system and cpu fans set to be super quiet). It is quite possible that the heat generated from 3080 is making your cpu and or ram overclocks unstable.

HazardO · 24 Jan 2021 at 14:08

FWIW when I got my 2080Ti FE, I got regular hard shutdowns on my rig. Swapped my single rail corsair HX850 for a 1KW superflower and have been rock solid ever since. The HX850 was about 4-5 years old though

opethdisciple · 24 Jan 2021 at 14:08

I agree the most likely culprit here is the ram.

It's at 1.47v in bios (more like 1.48-1.5v observed voltage. It oscillates) plus heat from 3080 could be pushing it over the edge.

But I thought it was well tested by reviewers that the new FE design wasn't negatively affecting CPU / RAM temperatures.

tommoT.Striker · 24 Jan 2021 at 14:40

AthlonXP1800 said:
Look like PSU is the issue, Corsair RM 850x did not supported single rail. About a week ago I read somewhere someone with AMD Ryzen CPU had same issues with RTX 3080 and Corsair HXi 750i PSU then hours later after someone posted suggested switch to single rail mode in iCUE software and the issue wass finally resolved with no more freezes and crashes. Guess RTX 3080 do not like PSUs with multi rails.

This was my problem with my 3080. Once i set my PSU to Single Rail, the black screens I was getting on a high GPU load were gone and it behaves itself. I've heard horror stories of EVGA 3080/3090 dying so hopefully this isn't happening.

For your PC hopefully the single rail switch works. But as a baseline to solve issues, put the RAM to stock and the CPU including turning PBO off. There is a chance the 3080 exposed instabilities in the system.

tommoT.Striker · 24 Jan 2021 at 14:42

HazardO said:
FWIW when I got my 2080Ti FE, I got regular hard shutdowns on my rig. Swapped my single rail corsair HX850 for a 1KW superflower and have been rock solid ever since. The HX850 was about 4-5 years old though

A lot of reports about this, i'm surprised it's not well known. Although in my experience hard shutdowns can be because the PSU or the cabling are damaged and can no longer supply the required power.

deuse · 24 Jan 2021 at 14:56

opethdisciple said:
Good spot mate.

However mine is an RM850x and it doesn't have such a switch.

You're the second person with that power supply having RTX 3000s problems here.
The supply is made by CWT.

opethdisciple · 24 Jan 2021 at 15:12

deuse said:
You're the second person with that power supply having RTX 3000s problems here.
The supply is made by CWT.

Well that would be sad news if true. Do you know who the second person is?

opethdisciple · 24 Jan 2021 at 15:15

Also on the subject of ram and over heating.

Lets imagine the issue is the ram over heating.

Those with say the 8 Pack 3600MHz CL14 ram sticks which are 1.45v out of the box surely they too would overheat if I bought them and used them in the same scenario I am running my current ram.

Or is it they are binned better for high voltages and ability to remain stable as the heat creeps up?

---

I guess I will find out what the issue is over the next few days as I will run my system at stock and even the ram at 2133MHz.

If it is stable, then the PSU/GPU are fine and the issue is with the ram.

If it still crashes then I have big problems and have either a faulty 3080 or the PSU is at fault.

But I bought this PSU back in March 2020 and it's been fine so far.

opethdisciple · 24 Jan 2021 at 15:30

First post here someone mentions that the RMX series has been confirmed to work with the 3000 series apparently.

ltron · 24 Jan 2021 at 15:37

opethdisciple said:
First post here someone mentions that the RMX series has been confirmed to work with the 3000 series apparently.

It should work, I am sceptical the issue that those people had is due to multi rail (unless you are using a single PCIE power connector that is daisy chained), it's much more likely to be a faulty PSU exposed by the increased power draw of the RTX 3080. Asus even put special LEDs on their RTX 30 cards that warn people when there is a transient drop in voltage due to an old or failing PSU that caused the card to crash. Monitoring the PCIE power connector voltages with HWInfo is a decent substitute and should tell you if you are running into problems.

I had crashing issues with a faulty 2080TI and switching between single and multi rail on my Corsair RM1000i made no difference.

turbot1984 · 24 Jan 2021 at 15:48

ltron said:
It should work, I am sceptical the issue that those people had is due to multi rail (unless you are using a single PCIE power connector that is daisy chained), it's much more likely to be a faulty PSU exposed by the increased power draw of the RTX 3080. Asus even put special LEDs on their RTX 30 cards that warn people when there is a transient drop in voltage due to an old or failing PSU that caused the card to crash. Monitoring the PCIE power connector voltages with HWInfo is a decent substitute and should tell you if you are running into problems.

I had crashing issues with a faulty 2080TI and switching between single and multi rail on my Corsair RM1000i made no difference.

Some cheaper 750W psu's have a pair of 20A rails, would that run a 3080?
*not on OCUK i will add.

ltron · 24 Jan 2021 at 15:53

turbot1984 said:
Some cheaper 750W psu's have a pair of 20A rails, would that run a 3080?
*not on OCUK i will add.

That sounds a bit low, what brand is this? Because I did say you won't have issues with a decent brand. However, if the two PCIE power connectors are on separate rails then I think this would be fine (the manufacturer should have taken care of this). Nvidia do recommend a 750W+ PSU, I could swear they recommended 850W at one point.

EsaT · 24 Jan 2021 at 16:03

opethdisciple said:
I agree the most likely culprit here is the ram.

It's at 1.47v in bios (more like 1.48-1.5v observed voltage. It oscillates) plus heat from 3080 could be pushing it over the edge.

But I thought it was well tested by reviewers that the new FE design wasn't negatively affecting CPU / RAM temperatures.

Samsung B-die is known to be temperature sensitive and when overclocked hard it's apparently super sensitive.
For example this mentions 45C:
https://www.igorslab.de/en/does-the...r4-3800-cl14-2x-16gb-put-through-its-paces/3/

And that 1,45V would be certainly in area of needing good airflow for memory cooling.
Even if room is cool and case airflow good long duration gaming could push temps up just enough.
You might check with Thaiphoon Burner/HWinfo etc if you can see temperature of DIMMs.

Then there's the another part in other end of memory wiring: Memory controller.
Whose stability is also affected by CPU temps.
Normal memory testing doesn't stress CPU in any real way, so memory controller has easy time.
But add heavier CPU load and possibly increasing case temperature from all heat of graphics card and memory controller might get into problems.

Having case open with some fan pushing air there could tell, if those problems are about rising temperature.

Jerry64 · 24 Jan 2021 at 16:03

a good 750 ie evga or corsair or sea sonic will run a 3090

Woodsta888 · 24 Jan 2021 at 16:13

Wasn't it you and I that had the same RAM issues? COuld never get my 3600 to run at 3600 and it ran fine at 3577Mhz or whatever is the nevxt speed down. I put in MY 3080 and I do get the odd crash at the beginning. COuple of blue screens and a couple of reboots (oddly only when surfing). Flippin' annoying mate.

opethdisciple · 24 Jan 2021 at 17:11

Woodsta888 said:
Wasn't it you and I that had the same RAM issues? COuld never get my 3600 to run at 3600 and it ran fine at 3577Mhz or whatever is the nevxt speed down. I put in MY 3080 and I do get the odd crash at the beginning. COuple of blue screens and a couple of reboots (oddly only when surfing). Flippin' annoying mate.

Probably, name rings bell.

My ram has been working fine for a good year if not more though, up until I installed the 3080.

turbot1984 · 24 Jan 2021 at 17:36

ltron said:
That sounds a bit low, what brand is this? Because I did say you won't have issues with a decent brand. However, if the two PCIE power connectors are on separate rails then I think this would be fine (the manufacturer should have taken care of this). Nvidia do recommend a 750W+ PSU, I could swear they recommended 850W at one point.

The one I saw was a CiT, to be fair i put a 500W one in a build and its been great, but its running s GT 730, not a beast like 3080.