Question Nitro + RX 590 less power draw

darebt123

Commendable
Jan 12, 2019
79
5
1,535
Since I've gotten this card, I've noticed that I get lower FPS on games from people with the same build, the core speed while gaming is 1560Mhz fixed, which is good, but on Furmark it's around 1300 Mhz.
Also, I've noticed that the power draw is not adequate to the examples shown on the internet, following:
https://www.techpowerup.com/reviews/Sapphire/Radeon_RX_590_Nitro_Plus/30.html
Photos:

View: https://imgur.com/a/yXbTkww

View: https://imgur.com/a/uPDDx1Y


Mine never exceeds 185W power draw.
Radeon Chill is turned off(the option for conservative power usage).
My monitor is 60 Hz 1080p connected through HDMI.
PSU: 650w CM Vanguard GOLD
CPU: i7 4790k (stock)
MB: H81M-S1
RAM: 16GB 1600mhz ddr3

Is there anything wrong with it? Does someone else experience the same thing?
 
You can max out the power limit from AMD's WattMan utility that's already running inside the driver software. No need to have a 3rd part app.

You can keep power draw (heat/noise) low while ALSO maintaining your 1560MHz clockspeed simply by applying a manual voltage curve in WattMan. I haven't tested higher, but my undervolted RX480 pulls 100W in FurMark @ 1305MHz/1020mV. Auto voltage wants to apply anywhere from 1075-1150mV @ 1305MHz. RX590 is more efficient than an RX480/580. Notice TPU's review says their card hit 1560MHz @ 1175mV.

You may also be able to eke out some additional performance by using Memory Timing Level 2 from the WattMan utility. Gave me an extra 3-5% performance without increasing power draw.

The following is from a thread I was helping someone with their RX580:
Try this on for size (obviously test for stability, each card is a little different).

600MHz/820mV
900MHz/870mV
1145MHz/925mV
1215MHz/945mV
1255MHz/970mV
1300MHz/1020mV
1355MHz/1100mV

VRAM = 2000MHz/925mV
Leave Memory Profile = "Auto"

I'd suggest checking the voltages are actually showing up like that on cold boot. I've had issues with machines not applying the voltages correctly on cold boot, but a single restart fixes the issue (and yes, you have to do this EVERY time you boot the PC). My personal machine does this.

Do you have a fixed refresh or FreeSync monitor? Using Frame Rate Target Control for fixed refresh and Chill for FreeSync is a good way to save additional power/heat/noise because it allows the GPU to downclock into those lower frequencies if you're playing a lightweight game where your GPU can crank out say 200fps but your monitor is only 60Hz. Likewise, Chill adds the ability to downclock the GPU in scenes/instances where there's little/no movement on screen.
IIRC, @ about 945mV, your GPU will probably only pull about 90W power draw (don't quote me on that, might be less, see what GPUz says)

I'm pretty certain the "Power Limit" slider does nothing once you're into manual voltages. I haven't changed it from 0% in a long time (I don't OC, I just undervolt while keeping the default clocks) and I can't remember the conclusion of my tests back when I investigated it. If you're on Auto voltage, the Power Limit slider will allow the GPU to suck more power and/or voltage. Therefore, TJHooker's advice about that is/was valid, but only for Auto voltage. Manual voltage, on the other hand, reduces the power usage, so you've got more headroom (at stock clocks).

From my experience, it seems AMD tested and found a fitting voltage curve for the majority of cards, then offset by about +50mV for the Auto voltages. Since the RX580 is just a further OC'd RX480, hitting power limits is a somewhat common occurrence. Coincidentally, the RX480 stock clock of 1266MHz sits pretty close to the 950mV voltage inflection point (kudos AMD!). You'll find that most chips have some sort of "safety factor" offset in this same vein (including CPUs, hence why you always see things like "I OC'd my CPU to xxxx @ stock voltage).
 
Last edited:
  • Like
Reactions: darebt123

darebt123

Commendable
Jan 12, 2019
79
5
1,535
You can max out the power limit from AMD's WattMan utility that's already running inside the driver software. No need to have a 3rd part app.

You can keep power draw (heat/noise) low while ALSO maintaining your 1560MHz clockspeed simply by applying a manual voltage curve in WattMan. I haven't tested higher, but my undervolted RX480 pulls 100W in FurMark @ 1305MHz/1020mV. Auto voltage wants to apply anywhere from 1075-1150mV @ 1305MHz. RX590 is more efficient than an RX480/580. Notice TPU's review says their card hit 1560MHz @ 1175mV.

You may also be able to eke out some additional performance by using Memory Timing Level 2 from the WattMan utility. Gave me an extra 3-5% performance without increasing power draw.

The following is from a thread I was helping someone with their RX580:
That is very helpful, thank you a lot, i will test it myself just to see how it works, although my concern is not conserving the power draw at all, on the contrary i want to have as much as possible performance needless of the power draw, that is why I have gotten a 650w Gold.
I was wondering if something is wrong with the GPU itself, since I've noticed that I've been getting 20-30 fps less than the official benchmarks, for example I get around 30 on ultra on GTA V, while the reference says 67 average.
The benchmark is made with my cpu exactly and a ddr 3 ram.
Is it possible something to be wrong with the card itself? I wonder why I have 80 W lower draw than the official score?
 

TJ Hooker

Titan
Ambassador
I wonder why I have 80 W lower draw than the official score?
Measuring power draw with software isn't accurate, and often reports lower values than real power draw.

I've been getting 20-30 fps less than the official benchmarks, for example I get around 30 on ultra on GTA V, while the reference says 67 average.
The benchmark is made with my cpu exactly and a ddr 3 ram.
What "official benchmarks" are you referring to?
 

darebt123

Commendable
Jan 12, 2019
79
5
1,535
Measuring power draw with software isn't accurate, and often reports lower values than real power draw.


What "official benchmarks" are you referring to?
I was referring to the gpucheck benchmarks about the games:

And the 3d benchmark tests:
View: https://m.imgur.com/a/jLgOlVc

View: https://m.imgur.com/a/zrJKCNS


I get a score below the 1060
 
Last edited:

darebt123

Commendable
Jan 12, 2019
79
5
1,535
Please run this and post a link to your results: https://www.userbenchmark.com/
I did that as well, I got 75-80%, as I had ran it multiple times, which was rated as good but it was bugging me that I "was supposed" to get higher fps on games, looking at other benchmarks.
Honestly I wouldn't have been looking at the GPU "with a magnifying glass" if it didn't cost me a fortune for the standards here, plus it is shipped from Canada, so if anything was wrong, it would be a big problem. 😐
 

TJ Hooker

Titan
Ambassador
Hmm, when you game (or run a stress test) for a while, does the GPU core clock stay pretty stable around ~1560 MHz? What is max CPU and GPU temp?

Does GTA V have a built in benchmark? If not, you might be looking at FPS in a different situation in-game than the other benchmark results you're looking at, and FPS can vary a lot depending on what is occurring on-screen.
 
  • Like
Reactions: darebt123

darebt123

Commendable
Jan 12, 2019
79
5
1,535
Hmm, when you game (or run a stress test) for a while, does the GPU core clock stay pretty stable around ~1560 MHz? What is max CPU and GPU temp?

Does GTA V have a built in benchmark? If not, you might be looking at FPS in a different situation in-game than the other benchmark results you're looking at, and FPS can vary a lot depending on what is occurring on-screen.
1. My case is pretty terrible atm, the is only 1 cooler dragging the air out on the rear, and the hottest the gpu has gotten on a stress test was 78C, otherwise is keep it around 60C with manual control.
The CPU is around 60C in-game.
And yes the clock is stable, it's always on 1560 Mhz

2. I've thought of that as well, so I have tried a lot of things, like just walking around, staying in place, driving insanely and shooting, and the most I would get is 50-52, lowest 25-26, which makes the aprox average of 30-40 fps.

Do you think there would be other signs if there was somethinf wrong with it?
 

TJ Hooker

Titan
Ambassador
Nothing else really stands out as being wrong to be honest. It never hurts to try doing a clean driver install though. Uninstall current drivers using DDU (in safe mode) and then reinstall.

Could also try running other standardized graphics benchmarks (e.g. 3Dmark firestrike) and comparing them to values listed in reviews to see if your card seems to perform as expected in other scenarios.
 

darebt123

Commendable
Jan 12, 2019
79
5
1,535
Nothing really stands out as being wrong to be honest. It never hurts to try doing a clean driver install though. Uninstall current drivers using DDU (in safe mode) and then reinstall.

Could also try running other standardized graphics benchmarks (e.g. 3Dmark firestrike) and comparing them to values listed in reviews to see if your card seems to perform as expected in other scenarios.
Alright, thanks a lot for your time and help! I will keep the thread open, and I'll post the results.
Thank you again!
 
  • Like
Reactions: TJ Hooker
Didn't you say your GPU was running at 1300MHz instead of 1560MHz? That's 85% of the highest frequency, which would go hand-in-hand with the percentages you're saying.

Two solutions to that problem:
  1. Bump up the power limit slider while on Auto voltage - high power draw, but limit is increased.
  2. Set manual voltages - lower power draw at each frequency, so you don't even reach the power limit.
 

darebt123

Commendable
Jan 12, 2019
79
5
1,535
Didn't you say your GPU was running at 1300MHz instead of 1560MHz? That's 85% of the highest frequency, which would go hand-in-hand with the percentages you're saying.

Two solutions to that problem:
  1. Bump up the power limit slider while on Auto voltage - high power draw, but limit is increased.
  2. Set manual voltages - lower power draw at each frequency, so you don't even reach the power limit.
I said I have lower core clocks on the Furmark stress test only, otherwise it is constant 1560 in games.
Since we've got the software power draw being mostly incorrect out of the way, I'm only concerned about the lower FPS in games and the passmark benchmark for example, like 8800/8900 out of 9300+
 
This is hard for me, because I have manual voltages set for each frequency step, so my 1300MHz (for example) will always be the same power draw because it's always getting the exact same voltage. If you're on auto voltage (as mentioned before) and you can't be certain whether Auto is supplying 1075mV for 1300MHz or 1150mV, that's a noticeable power difference. I've seen "auto" give my GPU both voltages for 1300MHz.

From my experience, FurMark is NOT as stressful on GPUs as some/most games, so your frequency results seem a little backwards. But hey, you measured the freq. so you ought to know. Perhaps games suck more power, but less consistently, so the GPU could give itself more temporary frequency headroom in games...? If your GPU is running the same frequency in games as other's benchmarks, then as long as you both have the same CPU (and CPU frequency) and RAM quantity/frequency, the frame rates [IN OFFLINE GAMES**] should be the same.

Also, not saying software is perfectly accurate, but...from my observations about my rig, I'm inclined to think it's pretty close. If anything, it can be used as a relative comparison before/after changes.

I really do recommend all AMD RX4xx and up set up a manual voltage curve. AMD's WattMan utility is already running on your computer, and it's comically easy to set and SAVE a voltage curve for yourself and save a good amount of power/heat/noise and/or to increase clockspeeds whilst staying within your current power draw on "auto". Even if GPUz is wrong in absolute terms, for me it showed a 30-35W power draw savings at the same frequencies after applying manual voltages.
 
Last edited:

darebt123

Commendable
Jan 12, 2019
79
5
1,535
This is hard for me, because I have manual voltages set for each frequency, so my 1300MHz will always be the same power draw. If you're on auto voltage (as mentioned before) and you can't be certain whether Auto is supplying 1075mV for 1300MHz or 1150mV, that's a noticeable power difference. I've seen "auto" give my GPU both voltages for 1300MHz.
Also, not saying software is perfectly accurate, but...from my observations about my rig, I'm inclined to think it's pretty close.
Ahh.. The worst part is that I'm away from home atm and can't test it, but I'll try out everything and post here again,
Thanks a lot for your time!