PC system max power draw?

[Veseleil]

I was always wondering about this. So now as I got this energy monitor, I've started collecting numbers.



I play my games at 1440p on my Purple Haze rig (system specs);
Disabled VSync;
Unlocked FPS (if no cap option at 144 or 165);
Maxed out graphics;
HWinfo, Adrenaline at the second (1080p) monitor;
The 6900's frequency slider is set between 2300MHz and 2500MHz most of the time, as gaming with a turboprop at a half meter from me isn't my kind of fun. The fan curve is set to keep GPU and GPU hotspot at max 60C and 80C respectively.


In idle, it consumes from 80W to 120W, which depends if the both monitors are used. And sometimes it uses 120W with just one on, until you power cycle the unused one.

Days Gone - 460W
Age of Wonders 4 - 360W
Warhammer Rogue Trader - 350W
Baldur's Gate 3 was around 480W iirc

The list will be updated on the fly, as I try different games.

I'm curious about yours. Welcome!

 

[Klemc]

It would be a good idea to test old (DSR), ugly graphically, or badly optimised games, tha make the GPU/CPU (aio wc) fans crazy, like : Pathologic 2 (oh yeah, this one)...

... so we would know if it takes a lot power too !

 

[Veseleil]

It would be a good idea to test old (DSR), ugly graphically, or badly optimised games, tha make the GPU/CPU (aio wc) fans crazy, like : Pathologic 2 (oh yeah, this one)...

... so we would know if it takes a lot power too !

Graven at 144FPS locked used more than 400W, I know by the sound of the GPU fans. Had to move that slider to 2300MHz to endure. Low poly indie game...

 

[GerKNG]

my GPU alone pulls more power

 

[Klemc]

Graven at 144FPS locked used more than 400W, I know by the sound of the GPU fans. Had to move that slider to 2300MHz to endure. Low poly indie game...

Game just updated thre days ago btw :

9 February 2024
Even More Bugs Fixed - New Patch for GRAVEN is Live

 

[Veseleil]

my GPU alone pulls more power

Bring it on! TBH, most of my numbers are coming from a GPU side, as 5800X3D, fans and drives are sipping power.

 

[Bill_Bright]

Yeah, most people would be surprised how "little" their computers actually consume most of the time. This happens because, even when tasked, it is very rare for the CPU, GPU, motherboard, RAM, fans and connected drives and other devices to all demand maximum power at the exact same point in time.

In idle, it consumes from 80W to 120W, which depends if the both monitors are used.

If you see your monitors' power consumption being reflected in your power meters readings too, that indicates you have your computer and monitors plugged into a common power strip or something similar. That's fine, however, it means you are not actually reading the consumption of just the computer.

BTW, those kill-a-watt type meters are nice but I note most of the better and larger UPS with AVR have a very informative status display panel too. Two birds, one stone.

 

[JIWIL]

Been using a UPS with LCD screen to monitor power usage over the past ~15 years; maybe ~5w margin of error, not horribly accurate but a good ballpark.

Current system idles around 100w +/- 10w for everything connected to that UPS (Ryzen sig comp, 27" IPS/1440p, DAC, ~65w Class D stereo amp, 250w Class D sub amp, sim wheel, etc).

Older comp consumed about 20w less, just due to the Ryzen idling @25w vs 10w on the 8600k. For reasons I can't fathom, the 4080 idles around 10w less than the 2080ti... and virtually half of what the 2080ti drew in-game (I generally cap most non-FPS content @72fps).

As for fans, I really dislike any kind of variations in the white noise, so I tend to find a RPM % that sounds pleasant at my sitting position and then lock it there (case/CPU/GPU fans). I don't have any aural clue if the system is drawing 100w or 400w, which is cherry.

What's really surprising is how little power the audio stuff uses. The amps idle at basically nothing, and at uncomfortably loud levels the 2ch amp hits maybe 5w and the sub around 10-20w. My old AVR idled at 60w 24/7 and under load 100+... I guess it had a Class A or A/B amp in it?

I usually run a power limit on the 4080 (175/200/250w depending on profile), 7600x (80w), and the aforementioned FPS cap, so totally system power (+peripherals) rarely exceeds 300w. I think I broke 500w immediately after build, stress testing Cinebench and Furmark without limits to make sure everything was working as advertised, but haven't seen more than 350w since then.

Generally the single thing on my desk that draws the most power is my hotplate for coffee mug (~40w).

 

[Bill_Bright]

Considering the fact the power demands of the various components inside a computer can go from idle to maximum in just a few clock cycles, and since there could easily be 3 Billion+ clock cycles every second, I think ~5W margin of error is plenty accurate for these purposes.

In addition to the demand/load from the computer is constantly changing, the sample rate of the power meter is set to specific intervals and that could easily be 1 to several seconds apart. This means even is the error of margin was much tighter, by the time power consumption was displayed, it would be superseded and outdated anyway.

Last, the most common reason to check the power consumption is to determine if the power supply is [at a minimum] "adequate" and/or you are not at risk of overloading the facility circuit.

"IF" a reading with ±5% error of margin gave concern for consumption being too close to those limits, you are already too close and already need a bigger supply and/or need to divide the load onto separate circuits.

 

[Veseleil]

If you see your monitors' power consumption being reflected in your power meters readings too, that indicates you have your computer and monitors plugged into a common power strip or something similar. That's fine, however, it means you are not actually reading the consumption of just the computer.

Measuring just the PC internals. The different power consumption with a different combinations of monitors (how many, what res, what refresh rate) is a Radeon thing. Fine wine, move along, nothing to see here.

What's really surprising is how little power the audio stuff uses. The amps idle at basically nothing, and at uncomfortably loud levels the 2ch amp hits maybe 5w and the sub around 10-20w. My old AVR idled at 60w 24/7 and under load 100+... I guess it had a Class A or A/B amp in it?

Probably, as class D is cold all the time.

 

[Tomgang]

It can really differ about wattage use. Is the pc running Stock, overclock or maybe undervolt/underclock on cpu/gpu. Is fps locked or not in games, dlss/fsr can save some wattage to and so on.

That has a lot to say. I made eksperimente with my own Two pc's.

The little with 5600x/rtx 4060. Ilde use 125 watt, thats including everything. Monitor, speakers, the pc it self. My power saving settings in games is among 220 watt. Full Stock around 300 watt. Overclocking around 325 watt all includes monitor and speakers. So total powerdraw for everything. Gpu is power locked and cpu cooler significantly limits powerdraw out of stock spec.

My big machine however can have a large difference and again all results including monitor and speakers power use. thats a 5950X/rtx 490

Ilde/light load 160 watt with with out rgb light. with rbg on 180 watt.

Power saving settings in games 250 watts to 350 watt. Including undervolt on cpu and gpu, cpu cores deaktivated and some other things.

Stock 400 watt to 600 watt depending on game and power settings.

With overclock/pbo active. Power can go from 700 to well over 800 watt. Cpu alone 200 watt and gpu up to 600 watt + everything else like monitor, speakers, motherboard, ssd/hdd, rgb light. Peak power can go to 950 in total power draw. Rtx 4090 becomes really in efficient when overclock. Much higher power use for relative little performance gain.

All results above is with games uncapped fps. I control my power draw by limit cpu/gpu power usage and undervolt. Else i let the hardware run lose and provide as much fps as possible at the given max allowed watt consumption.

With that said, i can highly recommend to undervolt a high end gpu. I save 100 to 150 watt by running my gpu at 0.900 mv clock locked to 2600 MHz compared to stock 1.050 volt and 2775 MHz gpu clock with minimal performance loss. The higher end and power hungry a gpu is. This more a underclock/undervolt will save you wattage. on the other hand a card like rtx 4060 is a minimal saving. Same settings as 4090 gave me 15-20 watt saving. So the more power hungry gpu, the more you would benefit from it.

 

[QuietBob]

These are the maximum momentary values recorded at the wall for the system in my profile. PC only, without the monitor and other peripherals. Full stock GPU.
Testing with the in-game benchmark in native 4K with manually maxed out settings (AA and RT disabled where possible). No upscaling, no dynamic resolution/quality scaling, no frame generation, with the game's default FOV.

Game	Uncapped frame rate [W]	60 fps cap [W]	Difference [%]
Assassin's Creed: Valhalla	447	280	-37
Borderlands 3 (dx12)	450	346	-23
Deus Ex: Mankind Divided (dx12)	454	381	-16
F1 23	438	246	-44
Far Cry 6	459	311	-32
Forza Horizon 4	451	219	-51
Forza Horizon 5	452	331	-27
Ghost Recon: Breakpoint (Vulkan)	449	416	-7
GTA V	474	297	-37
Horizon: Zero Dawn	452	318	-30
Immortals: Fenyx Rising	445	338	-24
Metro Exodus: Enhanced Edition	452	450	0
Rainbow Six: Extraction	479	351	-27
Red Dead Redemption 2 (dx12)	446	417	-7
Shadow of the Tomb Raider	462	331	-28
The Division 2	458	370	-19
The Riftbreaker	468	249	-47
Watch Dogs: Legion	459	382	-17

Yes, I game in 60 fps
This build idles at 56 W, out of which 28 W is the CPU, 5 W the GPU and 23 W for the rest of the system.

 

[Sithaer]

This is my PC only, the monitor is not plugged into my power meter.
Idle with TPU open: ~60-65W
Playing Diablo 4 end game content with maxed out settings and DLAA: ~185W-spikes up to around 200W.
Cyberpunk High settings with RT reflections and Lights on Ultra + DLSS Quality: 280-300W. 'This is the highest draw I've seen from gaming on my system'

Do keep in mind that I cap my FPS via Nvidia controll panel to 73 since I only have a 75 Hz monitor + my 3060 Ti is undervolted to 950 mV.

 

[JIWIL]

With that said, i can highly recommend to undervolt a high end gpu. I save 100 to 150 watt by running my gpu at 0.900 mv clock locked to 2600 MHz compared to stock 1.050 volt and 2775 MHz gpu clock with minimal performance loss. The higher end and power hungry a gpu is. This more a underclock/undervolt will save you wattage. on the other hand a ved like rtx 4060 is a minimal saving. Same settings as 4090 gave me 15-20 watt saving. So the more power hungry gpu, the more you would benefit from it.

My 200w profile, which I use most of the time, is 2650mhz @ 950mv +1200 on mem. Performs about 80-90% of stock (320w). The other profile that I use often is just a straight 50% PL (160w) for things that I know should never use any real amount of power draw but end up doing silly stuff anyways... like Super Resolution that generally uses 30-50w on setting 2 or 3 but sometimes pegs PL for no reason.

In addition to the demand/load from the computer is constantly changing, the sample rate of the power meter is set to specific intervals and that could easily be 1 to several seconds apart. This means even is the error of margin was much tighter, by the time power consumption was displayed, it would be superseded and outdated anyway.

Yeah polling rate on the UPS is about 10 second intervals.

System only idle power is ~70w -- CPU 30 // GPU 5 // 35 for MB, Fans, Drives, Etc. Honestly if I had known Zen4 idled like this I'd have been very tempted to go 13400/600 instead of the 7600x. Am planning on dropping a Zen5 X3D chip in it eventually, so... eh. The only CPU tweaking I've done is -20 All Core PBO and SOC voltage from default (1.35v?) to 1.25v.

 

[natr0n]

I can maintain 120fps playing forzaH5 at around 300 watts

Idle 125 watts

I have tuned the sh!t out of my system

 

[mechtech]

all bios and windows power settings are default

Terraria 4k max graphics 135W including the monitor and printer plugged into the UPS

desktop idle about 80-85w

monitor sleeping about 57w-62w

 

[Deleted member 237813]

I was always wondering about this. So now as I got this energy monitor, I've started collecting numbers.

just so you know you can add 10-20% on top of what that monitor shows you to be accurate. i cant explain why because my english is not good enoughfor that . but its just how it is^^

my pc needs at least

Gpu: 285 Watt max
Cpu: 65 Watt max
Rest: 50-70 Watts

with an nvidia rtx 4070 undervolted i managed below 250 watts for the whole thing , but how do we say power comes from power .

 

[Fokker]

I've never measured to see what the entire setup pulls, but my main rig pulls ~140W at idle and 550-700W depending on what I'm playing. I see 600-630 most often.

 

[Deleted member 237813]

I've never measured to see what the entire setup pulls, but my main rig pulls ~140W at idle and 550-700W depending on what I'm playing. I see 600-630 most often.

damn that would be expensive in europe lol

 

[SPDIF]

I've never measured to see what the entire setup pulls, but my main rig pulls ~140W at idle and 550-700W depending on what I'm playing. I see 600-630 most often.

140W? Damn, that can't be right... My system uses only 60-65W idle... Are you sure you measured with a TRUE RMS Wattmeter? Those cheap stock Wattmeters can't measure exactly from a Switch Power Supply.

 

[Deleted member 237813]

modern pcs easily suck 100 watts and more he probably has like 42 fans 2 oled tvs as rgb and what not .

 

[dgianstefani]

My UPS can pull 700 W +, but that's also powering two monitors and some LED lighting.

 

[Bill_Bright]

32 inch monitors can easily add 60+ watts each. LED lighting "should" almost be negligible - unless you have several 100! Looking at your impressive system specs, no doubt your CPU, GPU and cooling are your biggest hogs.

 

[mechtech]

damn that would be expensive in europe lol

Depends how long it’s used at that. Just checked my steam. 0.1hrs past 2 weeks.

 

[SPDIF]

You can do a lot in the bios to get the power drain down and also in Windows. I have around 60-65W {monitor not included} idle. Measured with a T-RMS meter. Those cheap Wattmeters you find everywhere are only good to measure a resistive load. Switch Power Supply's it does a very bad job, and you're not sure what you measure. Do the test, put in the wall and connect a small Phone Charger in it without any load connected. Does your Wattmeter shows about 1W to 7W or more, then you have a cheap meter that can only measure resistive loads.

A good T-RMS meter will show something around 0.1 or 0.3W even less, for a charger without load connected. More then 1W is impossible because at that power draw it would burn up itself...