Question Rtx 5090 + 9800x3d 1000w pay psu isn’t enough?

[syykii]

Hello, so few days ago I got an rtx 5090 astral oc it’s been working great I’ve managed to oc to 3000mhz and it wouldn’t go higher than 62C however when im stress testing both the cpu and gpu at the same time on OCCT power after 5minutes my pc would freeze and i would have to hold the power on putting to restart the pc. I’ve tried stress testing cpu and gpu on extreme separately for 15minutes and it was working flawlessly gpu hitting 62C at 550w and cpu 80C it’s at 5.2ghz 1.1v. Is it possible that when they’re both at a full utilization they’re taking more than 1000w that’s why pc is crashing ?

 

[Eximo]

What PSU?

 

[syykii]

What PSU?

Rm1000x corsair

 

[Lutfij]

Rm1000x corsair

How old is your PSU? In order to rule out your PSU being the issue, might want to source(borrow, not buy) a reliably built, higher wattage PSU, preferably 1.2~1.5KW and see if the issue persists.

On second thoughts, please include the specs to your build like so:
CPU:
CPU cooler:
Motherboard:
Ram:
SSD/HDD:
GPU:
PSU:
Chassis:
OS:
Monitor:
include the age of the PSU apart from it's make and model. BIOS version for your motherboard at this moment of time.

 

[Eximo]

Astral card should be reporting current on each wire going into the card. Verify none of them are getting above 9-10 amps. 550W does seem about right on the stock BIOS though, so it must be holding itself back.

Only the very late model RM1000x will be ATX 3.1, the older ones may still trip with large power draw spikes from the GPU.

So if it isn't the 2024 version, that may very well be the problem.

If the box DOESN'T have the 80+ certification, that would be the new one. Should only have Cybenetics Platinum certification.

 

[syykii]

How old is your PSU? In order to rule out your PSU being the issue, might want to source(borrow, not buy) a reliably built, higher wattage PSU, preferably 1.2~1.5KW and see if the issue persists.

On second thoughts, please include the specs to your build like so:
CPU:
CPU cooler:
Motherboard:
Ram:
SSD/HDD:
GPU:
PSU:
Chassis:
OS:
Monitor:
include the age of the PSU apart from it's make and model. BIOS version for your motherboard at this moment of time.

Cpu: ryzen 7 9800x3d
Cpu cooler: nucleus d 360aio
Motherboard: msi b650 tomhawk
Ram: gskill trideen 6000mhz cl30 16x2
Ssd/hhd: 990 samsung 1tb/ 2tb hhd seagate
Osd: win 11
Minitor: f0274q 360hz, benq xl2568x+ 600hz
I bought psu in 2023-07
I haven’t updated my bios in a half a year.

Astral card should be reporting current on each wire going into the card. Verify none of them are getting above 9-10 amps. 550W does seem about right on the stock BIOS though, so it must be holding itself back.

Only the very late model RM1000x will be ATX 3.1, the older ones may still trip with large power draw spikes from the GPU.

So if it isn't the 2024 version, that may very well be the problem.

If the box DOESN'T have the 80+ certification, that would be the new one. Should only have Cybenetics Platinum certification.

I bought the psu in 2023-07 i no longer have the box

 

[Eximo]

Sounds like it is the older model, so you could very well be running up against the over current protection.

2024 RM1000x would probably be okay and let you leave the cables in place, though you would want to use the 12VHPWR native cable. But if you want more capacity while you are at it:

https://pcpartpicker.com/product/Cn...-modular-atx-power-supply-ps-tpd-1200fnfagu-4

 

[syykii]

Sounds like it is the older model, so you could very well be running up against the over current protection.

2024 RM1000x would probably be okay and let you leave the cables in place, though you would want to use the 12VHPWR native cable. But if you want more capacity while you are at it:

https://pcpartpicker.com/product/Cn...-modular-atx-power-supply-ps-tpd-1200fnfagu-4

Would you recommend me to upgrade?

 

[Flayed]

If it only crashes during stress tests it might be alright for gaming and general usage.

 

[Eximo]

Would you recommend me to upgrade?

Up to you. As Flayed points out, could be fine for normal use. During gaming 9800X3D draw only like 55W.

And if you are like me, I tend to set a lower power limit on my GPUs. 5090 at 450W is still shockingly quick.

 

[syykii]

Up to you. As Flayed points out, could be fine for normal use. During gaming 9800X3D draw only like 55W.

And if you are like me, I tend to set a lower power limit on my GPUs. 5090 at 450W is still shockingly quick.

What do you think about this psu?

MSI MEG Ai1300P 1300w​

 

[Eximo]

What do you think about this psu?

MSI MEG Ai1300P 1300w​

A good unit.

 

[helper800]

I find it odd that the RMx1KW PSU is having OCP issues. I have a Seasonic prime titanium 850w PSU from 2019 that does not have these issues with a 5800X3D and a 5090. I wonder if something else is going on. Usually when a PSU trips a protection, the PC has a shutdown, not a freeze that forces a hard shutdown. I would look more at drivers, firmware, non-PSU hardware issues.

 

[ridwan123]

Is it your gpu clock beyond specification during stress test compare to gaming?

 

[Eximo]

Could just be as simple as extreme voltage droop under a heavy load.
I go out of my way to make sure that doesn't happen. 3 individual cables running from my RM850x to a 3080Ti.
If you have the room to add more PCIe cables, that is another thing to try.

I suppose it would depend on exactly where they put the OCP limits as well. Total unit should be okay, even with the whole system plus a 5090 power spiking to 740W, should still be under 1k. But if they put OCP limits on the individual PCIe connectors, those I could see being overloaded.

 

[helper800]

Could just be as simple as extreme voltage droop under a heavy load.
I go out of my way to make sure that doesn't happen. 3 individual cables running from my RM850x to a 3080Ti.
If you have the room to add more PCIe cables, that is another thing to try.

I suppose it would depend on exactly where they put the OCP limits as well. Total unit should be okay, even with the whole system plus a 5090 power spiking to 740W, should still be under 1k. But if they put OCP limits on the individual PCIe connectors, those I could see being overloaded.

I think each PCIe cable from corsair is rated for 25 amps sustained on each port and cable. I still believe something other than the PSU is going on. A new PSU to sanity check the RMx1kw PSU is also a good idea in the off chance the PSU is faulty. If the Corsair unit is fine, they can just return the new PSU. If the RMx unit is the issue, they can RMA it with corsair for a new unit they can either sell or use.

 

[Eximo]

I think each PCIe cable from corsair is rated for 25 amps sustained on each port and cable. I still believe something other than the PSU is going on. A new PSU to sanity check the RMx1kw PSU is also a good idea in the off chance the PSU is faulty. If the Corsair unit is fine, they can just return the new PSU. If the RMx unit is the issue, they can RMA it with corsair for a new unit they can either sell or use.

Yes, 600W, but the 5090 has measured spikes up to 900W according to Igor. That should be under the OCP time threshold, but he also measured spikes that could trip that threshold. The whole point of ATX 3.1 was to allow for longer time periods to prevent the issues we saw from the 30/40 series cards.

https://tpucdn.com/img/eqUfVoGQW9zuOuLH_thm.jpg

Removing the overclock and setting a manual power limit on the GPU would be a fairly simple test as well to rule out the PSU.

I agree though, testing with a known good supply would be ideal. A new one is pretty close to that, since chances are low to get two malfunctioning units.

 

[VizzieTheViz]

If the gpu is overclocked as you write the first thing I’d do is try to stress the system at stock speed.

If that doesn’t give this issue your oc is likely too high.