Jump to content

Running a 3090 with a 650W PSU

Origami Cactus
Go to solution Solved by Spotty,
29 minutes ago, Origami Cactus said:

I am willing to limit the power slider and voltage to the GPU a bit, because my experience with Pascal is that Nvidia gpus are way out of efficiency curve, so even if you limit their power by 10%, the performance is pretty much the same, maybe 2-3% slower.

No harm in giving it a try if you already have the PSU sitting around doing nothing. If you have issues it'll just shut down randomly during games or whatever which isn't really the end of the world. Use 2 separate PCIe cables and you'll likely want to switch the RMi to single rail mode via Corsair iCue software. Power limiting the card slightly might help but it's probably going to be you either have issues or you don't, like Kilrah mentioned it's the transient spikes and noise on the +12V rail that is causing the PSUs to shut down and not necessarily sustained load exceeding the OCP/OPP protections.

So I have a Corsair RM650i sitting in a box for like 3-4 years already, eating away it's 10 year warranty, so I thought that I should put it to good use and into a 3090 build.

So while first my plan was to have it in a threadripper build and I even got an extra 8pin connector for a cpu, I understand that with current gen GPUs it is not feasible.

So I am thinking of pairing it with a Ryzen 7 5700x or something like that, not a 12core.

BEFORE you say ANYTHING about how you need a 750W PSU minimum, I'm gonna need to see some proof.

 

Here is the power output for RM650i, 648W on the 12V rail, seems to be more than enough, wouldn't you say? From a quick glance I would even say that a threadripper could work, 250W for the CPU, 400W to the GPU, but I will stick to a 125W Ryzen for now, giving 525W to the GPU, and if that is not enough, then I don't know what is.

I am willing to limit the power slider and voltage to the GPU a bit, because my experience with Pascal is that Nvidia gpus are way out of efficiency curve, so even if you limit their power by 10%, the performance is pretty much the same, maybe 2-3% slower.

image.png.44da4d416d54e191e93a4e5cdd6f06b3.png

More info:

Quote

Efficiency is 80 PLUS Gold certified and the PSU can deliver its full power continuously at up to 50°C. All necessary protection features are present, and the user can disable OCP on the +12V rails through the Corsair Link application in order to turn this PSU into one with a single +12V rail.

image.thumb.png.d0725a6453d08528fd9f0ececff72968.png

 

Even in Multi Rail mode, each 12V has 480W.

I only see your reply if you @ me.

This reply/comment was generated by AI.

Link to comment
Share on other sites

Link to post
Share on other sites

The issue isn't the GPU's continuous power draw, it's the power spikes that can go significantly higher for a fraction of a second and trip the overload protection/cause instability on lower wattage PSUs even though they should be enough for the average load.

Lots of reviews / explanation articles and videos around on the topic.

F@H
Desktop: i9-13900K, ASUS Z790-E, 64GB DDR5-6000 CL36, RTX3080, 2TB MP600 Pro XT, 2TB SX8200Pro, 2x16TB Ironwolf RAID0, Corsair HX1200, Antec Vortex 360 AIO, Thermaltake Versa H25 TG, Samsung 4K curved 49" TV, 23" secondary, Mountain Everest Max

Mobile SFF rig: i9-9900K, Noctua NH-L9i, Asrock Z390 Phantom ITX-AC, 32GB, GTX1070, 2x1TB SX8200Pro RAID0, 2x5TB 2.5" HDD RAID0, Athena 500W Flex (Noctua fan), Custom 4.7l 3D printed case

 

Asus Zenbook UM325UA, Ryzen 7 5700u, 16GB, 1TB, OLED

 

GPD Win 2

Link to comment
Share on other sites

Link to post
Share on other sites

8 minutes ago, Kilrah said:

The issue isn't the GPU's continuous power draw, it's the power spikes that can go significantly higher for a fraction of a second and trip the overload protection/cause instability on lower wattage PSUs even though they should be enough for the average load.

Lots of reviews / explanation articles and videos around on the topic.

From what I have seen is the spikes are about 550w, so they should fit into my power envelope. And if I underclock it somewhat, I see no reason for the spikes to remain the same size.

As I said, I know that Nvidia wants to extract that extra 1% at 50% the power draw, they have been doing it for a while, but I see no reason to not be able to run the GPU more efficiently.

And even with the peaks, I am not going to run Furmark and Prime95 at the same time, gaming loads don't use max cpu and gpu at the same time.

 

I only see your reply if you @ me.

This reply/comment was generated by AI.

Link to comment
Share on other sites

Link to post
Share on other sites

I mean you can try, just know there's a good chance you'll have issues. 

 

In a recent LTT intel tech upgrade there was mention that basically everyone who had tried powering a 3080 with a borderline PSU ended up having to replace it. 

F@H
Desktop: i9-13900K, ASUS Z790-E, 64GB DDR5-6000 CL36, RTX3080, 2TB MP600 Pro XT, 2TB SX8200Pro, 2x16TB Ironwolf RAID0, Corsair HX1200, Antec Vortex 360 AIO, Thermaltake Versa H25 TG, Samsung 4K curved 49" TV, 23" secondary, Mountain Everest Max

Mobile SFF rig: i9-9900K, Noctua NH-L9i, Asrock Z390 Phantom ITX-AC, 32GB, GTX1070, 2x1TB SX8200Pro RAID0, 2x5TB 2.5" HDD RAID0, Athena 500W Flex (Noctua fan), Custom 4.7l 3D printed case

 

Asus Zenbook UM325UA, Ryzen 7 5700u, 16GB, 1TB, OLED

 

GPD Win 2

Link to comment
Share on other sites

Link to post
Share on other sites

31 minutes ago, Kilrah said:

I mean you can try, just know there's a good chance you'll have issues. 

 

In a recent LTT intel tech upgrade there was mention that basically everyone who had tried powering a 3080 with a borderline PSU ended up having to replace it. 

Replacing the PSU, or the GPU? The "it" is pretty open to interpretation here.

EDIT: I thought about it after a coffee, and if an underpowered PSU would hurt the gpu you would have been screaming at me not to do it, so you probably meant that the psu needs replacing.

I only see your reply if you @ me.

This reply/comment was generated by AI.

Link to comment
Share on other sites

Link to post
Share on other sites

Try it out, disable OCP if it trips, doesn't help ? You either need a 750/850W PSU or to downvolt/powerlimit the card.

Tag or quote me so i see your reply

Link to comment
Share on other sites

Link to post
Share on other sites

29 minutes ago, Origami Cactus said:

I am willing to limit the power slider and voltage to the GPU a bit, because my experience with Pascal is that Nvidia gpus are way out of efficiency curve, so even if you limit their power by 10%, the performance is pretty much the same, maybe 2-3% slower.

No harm in giving it a try if you already have the PSU sitting around doing nothing. If you have issues it'll just shut down randomly during games or whatever which isn't really the end of the world. Use 2 separate PCIe cables and you'll likely want to switch the RMi to single rail mode via Corsair iCue software. Power limiting the card slightly might help but it's probably going to be you either have issues or you don't, like Kilrah mentioned it's the transient spikes and noise on the +12V rail that is causing the PSUs to shut down and not necessarily sustained load exceeding the OCP/OPP protections.

CPU: Intel i7 6700k  | Motherboard: Gigabyte Z170x Gaming 5 | RAM: 2x16GB 3000MHz Corsair Vengeance LPX | GPU: Gigabyte Aorus GTX 1080ti | PSU: Corsair RM750x (2018) | Case: BeQuiet SilentBase 800 | Cooler: Arctic Freezer 34 eSports | SSD: Samsung 970 Evo 500GB + Samsung 840 500GB + Crucial MX500 2TB | Monitor: Acer Predator XB271HU + Samsung BX2450

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×