I think I need to upgrade PSU?

[Raytsou]

Hi guys, just upgraded from 2700x to 5900x. I am now getting frequent shutdowns while gaming. I used to get them occasionally but happened rare enough that I didn't think too much of it. But now every game I boot up the computer runs fine for a while then just *clicks off* with no warning, then powers back on. This seems like I don't have enough juice on my PSU, which is kinda absurd considering I have a 750 G2. I tried the usual, reinstalling chipset/gpu drivers, memtest86, checking temps, all were fine while gaming. I undervolted the GPU and things started running smoothly, so I really suspect its the power usage. I'm in a bit of a weird position rn since everything points to power issue but I have a hard time believing I hit 750W on 1 GPU... can someone sanity check me? This only makes sense if games are somehow maxing out both CPU and GPU.

Things connected to the PSU are:
5900x
6800XT
x470 mobo
2x32gb ram
ekwb DDC pump and 3 noctua fans
(case mod) Dell 1907fpv panel connected via molex + internal led strips
 

[Raytsou]

oh also I've experienced shutdowns in squad, war thunder, back 4 blood, and 3d mark firestrike. otoh I've been fine in csgo, valorant, and 3d mark time spy

[An0maly_76]

Something wrong here. 5900X needs power but my new build peaks at 450W load on my UPS (with an RTX3060ti). PSU may be going bad, but it's certainly not insufficient. What's your cooling setup (including case)?

Edited April 23, 2022 by An0maly_76
Revised, more info

[Raytsou]

5 minutes ago, An0maly_76 said:

Something wrong here. 5900X needs power but my new build peaks at 450W load on my UPS. PSU may be going bad, but it's certainly not insufficient. What's your cooling setup (including case)?

I have my CPU+GPU both in the same loop on a 120 + 240 rad (nemesis GTS 30mm) in a H510. GPU temps are 40C idle and 65C gaming. 5900X temp is always between 65~75C, regardless of load, for some reason.

[An0maly_76]

5 minutes ago, Raytsou said:

I have a 120 + 240 rad (nemesis GTS 30mm) in a H510. GPU temps are 40C idle and 65C gaming. 5900X temp is always between 65~75C, regardless of load, for some reason.

You've got an airflow / ventilation problem for sure (mine idles around 38C-42C, peaking 76C-86C, air-cooled), but that's still not in the throttling zone, and it will throttle before it shuts down. AIO pump could be getting weak too. As for the shutdowns, double check to make sure you haven't bumped your RAM loose or something (I did this when installing header cable adapters and it caused a shutdown). Still might be a flaky PSU starting to fail.

 

The NZXT H510 isn't a good choice for cooling, that's part of the reason for your higher temps.

[PDifolco]

750W is defo enough, I didn't hit ever 650W with OCed 5900X (200W) and a 3080 (320W) plus a ton of stuff (4 drives, DC pump, 2 AIO pumps, 6 fans, wifi/BT card, radio wireless peripherals)... And games never use full CPU on a 12 cores 5900X, usually less than 100W draw on the CPU

 

Unless your crashes have everything shut down (PSU included), in which case the PSU is faulty, then it's rather some other cause

 

However 65C idle is real high, should be 45ish, 75C full load is normal tho

[Raytsou]

5 minutes ago, An0maly_76 said:

You've got an airflow / ventilation problem for sure (mine idles around 38C-42C, peaking 76C-86C, air-cooled), but that's still not in the throttling zone, and it will throttle before it shuts down. AIO pump could be getting weak too. As for the shutdowns, double check to make sure you haven't bumped your RAM loose or something (I did this when installing header cable adapters and it caused a shutdown). Still might be a flaky PSU starting to fail.

 

The NZXT H510 isn't a good choice for cooling, that's part of the reason for your higher temps.

airflow makes no sense if my CPU and GPU are in the same loop and my GPU sits at 65 while gaming. It doesn't throttle, just cuts power. Furthermore, I can turn my fans to max and take the side panel off -- no difference in temps at all.
ddc pump being weak also doesn't make sense, my config is pump -> CPU -> 120 rad -> gpu -> 240 rad -> res -> pump. If that's the case then GPU would be hot too.
Do EVGA PSU's usually fail in 4 years? Another possible reason is that I have custom sleeved cables from cablemod and those might be going bad?

[Raytsou]

8 minutes ago, PDifolco said:

750W is defo enough, I didn't hit ever 650W with OCed 5900X (200W) and a 3080 (320W) plus a ton of stuff (4 drives, DC pump, 2 AIO pumps, 6 fans, wifi/BT card, radio wireless peripherals)... And games never use full CPU on a 12 cores 5900X, usually less than 100W draw on the CPU

 

Unless your crashes have everything shut down (PSU included), in which case the PSU is faulty, then it's rather some other cause

 

However 65C idle is real high, should be 45ish, 75C full load is normal tho

There's no "crash" or "freeze". Its as if someone just flicked I/O really fast on the PSU. It completely turns off and cold boots up again.

[PDifolco]

5 minutes ago, Raytsou said:

There's no "crash" or "freeze". Its as if someone just flicked I/O really fast on the PSU. It completely turns off and cold boots up again.

Did you check Windows event viewer ? Could point out what happened

[An0maly_76]

All I can tell you is that I run an 850W Corsair in a  Corsair 4000X case with six 120s and a Scythe Mugen 5, and mine maintains the aforementioned temps without issue. I do think your shutdowns could be simple like unseated RAM (as I mentioned happened with mine), or a failing PSU. And anything can fail at any time, especially electrical. Anything higher performance, I would recommend a PSU with 105 degree capacitors and monster cooling.

 

BTW -- if your GPU is first in the cooling loop, it would make total sense that the GPU is running cooler if the loop is GPU - CPU - RAD - RAD. GPU heat would be transferring to the CPU and the CPU heat would have no way to transfer. Something obviously isn't right. And even a 5900X doesn't require liquid cooling -- my build proves that.

Edited April 23, 2022 by An0maly_76
Revised, more info

[Raytsou]

5 minutes ago, PDifolco said:

Did you check Windows event viewer ? Could point out what happened

hmm forgot this tool existed, I'll keep an eye on it for next time I shutdown. Thanks

[Mark Kaine]

evga psus tend to have issues with newer gpus... (power spikes)

so yeah, it could be. try a corsair rmx 850w.

 

But also, your cpu temps are obviously strange (cant help with that, not into putting water into my pc, sorry, but still it seems obvious something is wrong with the loop, or mount) 

 

8 minutes ago, An0maly_76 said:

BTW -- if your GPU is first in the cooling loop, it would make total sense that the GPU is running cooler if the loop is GPU - CPU - RAD - RAD. GPU heat would be transferring to the CPU and the CPU heat would have no way to transfer. Something obviously isn't right.

yup this. but doesn't really explain the idle temps compared to load temps.

[An0maly_76]

5 minutes ago, Mark Kaine said:

evga psus tend to have issues with newer gpus... (power spikes)

so yeah, it could be. try a corsair rmx 850w.

 

But also, your cpu temps are obviously strange (cant help with that, not into putting water into my pc, sorry, but still it seems obvious something is wrong with the loop, or mount) 

Technically it's not water, it's some type of synthetic cooling fluid that's supposed to be nonconductive.

 

However, OP, if your loop runs GPU - CPU - RAD - RAD, that is part of your problem because that would be putting GPU heat directly to the CPU with insufficient capacity for the CPU to dissipate heat. I still think that NZXT H510 case is part of the problem -- it's hardly set up for adequate ventilation, and the 5900X puts off a lot of heat.

Edited April 23, 2022 by An0maly_76
Revised, more info

[Raytsou]

1 minute ago, An0maly_76 said:

Technically it's not water, it's some type of synthetic cooling fluid that's supposed to be nonconductive.

 

However, OP, if your loop runs GPU - CPU - RAD - RAD, that is part of your problem because that would be putting GPU heat directly to the CPU with insufficient capacity for the CPU to dissipate heat.

 

17 minutes ago, Raytsou said:

 my config is pump -> CPU -> 120 rad -> gpu -> 240 rad -> res -> pump.

 

[An0maly_76]

6 minutes ago, Raytsou said:

 my config is pump -> CPU -> 120 rad -> gpu -> 240 rad -> res -> pump.

 

You might check to see which direction the fluid is flowing. If the pump is circulating the other way, it would be pushing res -- 240 -- gpu -- 120 -- CPU. And that might explain it, because that would supercool the GPU and so-so cool the 5900X. I do think you might want to think about a more ventilated case in the future.

Edited April 23, 2022 by An0maly_76
Revised, more info

[Raytsou]

Just now, An0maly_76 said:

You might check to see which direction the fluid is flowing. If the pump is circulating the other way, it would be drawing from res -- 240 -- gpu -- 120 -- CPU. And that might explain it, because that would supercool the GPU and so-so cool the 5900X. I do think you might want to think about a more ventilated case in the future.

not possible for it to go the other way -- res is mounted vertically with the inlet at the top "dropping" the coolant down. I wouldn't've been able to fill the loop if it was the other way.

[Raytsou]

and because you can never trust anyone, just double checked. The coolant is definitely flowing the right way.

[An0maly_76]

2 minutes ago, Raytsou said:

not possible for it to go the other way -- res is mounted vertically with the inlet at the top "dropping" the coolant down. I wouldn't've been able to fill the loop if it was the other way.

You'll have to excuse me if I sound ignorant here -- AIO is not my forte. The case's lack of airflow certainly isn't helping matters, and that 5900X puts out a lot more heat than your previous 2700X. But as to the shutdowns, this could very well be weak capacitors in the PSU, and as someone else has said, EVGA has had issues with theirs, of course, that could happen with anything, really. Any replacement, be sure the dimensions are compatible -- they're not one size fits all. I used a Corsair RM850x in my build.

[An0maly_76]

One other thing -- there could be a blockage in the loop between the CPU and GPU. The GPU doesn't need as much flow to cool because it has its own fans. If the fluid isn't carrying the heat away from the CPU fast enough, that could very well be your problem. I presume all your tubing is the same diameter? What diameter is used?

[Raytsou]

1 minute ago, An0maly_76 said:

One other thing -- there could be a blockage in the loop between the CPU and GPU. The GPU doesn't need as much flow to cool because it has its own fans. If the fluid isn't carrying the heat away from the CPU fast enough, that could very well be your problem. I presume all your tubing is the same diameter? What diameter is used?

no the GPU has no fans any more, I had to remove them to install the waterblock. A blockage anywhere in the system would mess all the temps up. BUT if there say, some gunk built up inside the CPU block, it could cause heating issues. I'm not against pulling it out and taking a look, that's just a lot of work for something that's not really an issue (it doesn't go above 80C, just never goes below 60 for some reason). While its possible the shutdowns are a result of random CPU temperature spikes into the 90's, I feel like its more likely to be a PSU issue. I have a spare CX430M laying around that I'll plug in just the GPU to and test that when I have time...

[RONOTHAN##]

Something that might be worth checking would actually be your RAM. I have seen it before where unstable RAM will just cause the system to shut down during memory intensive loads, and its possible that's what's happening with your system. Try running memtest86 to see if that reports any issues.

[Raytsou]

27 minutes ago, Mark Kaine said:

evga psus tend to have issues with newer gpus... (power spikes)

do you have a source on that? I've never heard of such thing...
Also I thought evga g series are just rebranded seasonic PSUs... do I need to to avoid seasonic as well then?

[Raytsou]

Just now, RONOTHAN## said:

Something that might be worth checking would actually be your RAM. I have seen it before where unstable RAM will just cause the system to shut down during memory intensive loads, and its possible that's what's happening with your system. Try running memtest86 to see if that reports any issues.

yeah first thing I did was rum memtest86 for 8 hours, no errors found.

[Raytsou]

@An0maly_76posting a pic of the build here so you have a better idea of the cooling

 

[191x7]

7 minutes ago, Raytsou said:

@An0maly_76posting a pic of the build here so you have a better idea of the cooling

 

 

These PCI-E power extensions, are they plugged in into the same 2x 6+2 PCI-E cable?