What AMD Didn't Tell Us: 21.4.1 Drivers Improve Non-Gaming Power Consumption By Up To 72%

[shadow3401]

Here we go again with the AMD = bad, nVidia = sun shines out of its ass everyday. Now that this MINOR problem of increased power consumption has been identified i'm sure it will be resolved in the 21.4.2 or 21.5.1 driver update so no need to have a mental breakdown over it, move on.

 

[thesmokingman]

Isn't it obvious? It's because they're the good guys

I wish they would toot their horn more often like some other companies playing that media hype train to manipulate their stock price... lol.

 

[zlobby]

What about mobile platforms?

 

[HisDivineOrder]

I wish they would toot their horn more often like some other companies playing that media hype train to manipulate their stock price... lol.

AMD tooting their own horn about something they've done terribly for years and years and finally getting around to fixing it--and not even completely--is like Dominos advertising the fact that they "previously" delivered pizza that tasted like cardboard...

...not a wise PR move. Best to just fix the obvious and move on.

 

[brink]

i have my eizo 4k@60 via display port on an rx 6900 xt with as low as 5W idle. usually around 10W.

the idle power consumption directly correlates with the vram frequency. some visual changes trigger vram to ~900 mhz -> 20-30W. eventually, it cools down to ~200 nhz -> 10W. or even almost 0 mhz -> 5W.

when i connect in addition my lg 48" oled 4k@120 via hdmi 2.1, the vram goes up to 2000 mhz.
when "switch off" the lg, i.e. put into standy. no change. vram 2000 mhz.
when I unplug power cable from lg, vram frequency goes down.

when I unplug the eizo and have just the lg 4k@120. vram 2000 mhz.
when change to 4k@100. no change. vram 2000.
when change to 4k@60. vram goes down.

i.e. as long as only one monitor/tv is connected to the graphics card and only one device is electrically powered, then 4k@60hz has as low as 5W power consumption.
with higher refresh rate or more monitors (standby monitor/tv counts too), the vram goes to 2000 mhz and so the idle power consumption

 

[Dyatlov A]

Finally, it was my reason to not buy AMD GPU. As it is corrected, i would take one, if i could

 

[Frick]

Here we go again with the AMD = bad, nVidia = sun shines out of its ass everyday. Now that this MINOR problem of increased power consumption has been identified i'm sure it will be resolved in the 21.4.2 or 21.5.1 driver update so no need to have a mental breakdown over it, move on.

Waste of power isn't a minor problem. And AMD had quite bad non gaming power numbers long after Nvidia imroved theirs. These days there is no good reason why AMD has worse numbers tham Nvidia.

 

[zlobby]

Finally, it was my reason to not buy AMD GPU. As it is corrected, i would take one, if i could

As if you could pass the power tests. Now that everyone is catching up with their cryptomining productivity quotas... Protip: АЗ-5 doesn't do miracles!

 

[THANATOS]

I remembered different Idle values, so I checked the old reviews and what did I find?
AMD Radeon RX 6900 XT Review - The Biggest Big Navi
AMD Radeon RX 6700 XT Review

In RX 6900XT review the power consumption was this:
RX 6800 & 6800 XT: 20.45.01.12-11.6 Press Driver
RX 6900 XT: 20.45.01.14 RC8 Press Driver

	Idle	Multi-Monitor	Media Playback
RX 6800:	7 W	8 W	48 W
RX 6800 XT:	7 W	7 W	47 W
RX 6900 XT:	7 W	7 W	48 W

In RX 6700XT review the power consumption was different:
RX 6700 XT: March 3 Press Driver
AMD: 21.2.3 Beta

	Idle	Multi-Monitor	Media Playback
RX 6700 XT:	7 W	33 W	21 W
RX 6800:	28 W	40 W	48 W
RX 6800 XT:	29 W	41 W	50 W
RX 6900 XT:	28 W	40 W	52 W

Drivers were different, but the testing for Idle and Multi-monitor power consumption also changed, so I am not sure what had the biggest impact on power consumption.

 

[Robin Seina]

Is it only with RDNA2? Has anyone tried to measure with RX5700XT or older generation yet?

 

[turbogear]

AMD does what Nvidia doesn't ~ 7nm nom nom bois!!!



did you report it through amd software? i did. and i know one other did. the more who report this issue the sooner it gets fixed. you can't complain unless you do the bug report tool in amd software.

I reported it multiple times since November on 6800XT but unfortunately no fix until now.
For me also if I set default monitor refresh rate of 165Hz the memory remains at full speed.
If I drop refresh rate to 144Hz them memory clock drops.

I was ones reading that it could be due to the fact that the VRAM does not support intermediate voltage and frequency levels.

 

[brink]

Is it only with RDNA2? Has anyone tried to measure with RX5700XT or older generation yet?

i previously had the rx 5700 connected to my eizo 4k@60hz. idle also around 10w. afar vram clock was lower and so the power consumption during media playback, eg youtube.

I reported it multiple times since November on 6800XT but unfortunately no fix until now.
For me also if I set default monitor refresh rate of 165Hz the memory remains at full speed.
If I drop refresh rate to 144Hz them memory clock drops.

I was ones reading that it could be due to the fact that the VRAM does not support intermediate voltage and frequency levels.

i think it simply has to do something with the pixel clock. at given resolution and v-refresh, the vram is bumped up.
and/or with the physical layer drivers (for the pixel clock) and so if more cables have to be driven, the vram is also bumped up.
for amd, i suppose, there is little to gain from idle power consumption compared to average gaming.
maybe there are some optimization for few use cases. all others are just full clock..

 

[W1zzard]

Is it only with RDNA2? Has anyone tried to measure with RX5700XT or older generation yet?

Says so in the text, no differences for 5700 XT

I remembered different Idle values, so I checked the old reviews and what did I find?

The old reviews were idle @1080p, now it's 1440p. Multi-monitor 1080p+1280x1024 was outdated, too

 

[Xaled]

Who thought that non-gaming was "minning", before reading the body?!

 

[THANATOS]

The old reviews were idle @1080p, now it's 1440p. Multi-monitor 1080p+1280x1024 was outdated, too

I know, that's why I wrote this.

Drivers were different, but the testing for Idle and Multi-monitor power consumption also changed, so I am not sure what had the biggest impact on power consumption.

At least changing Idle 1080p to 1440p shouldn't have caused 4x higher power consumption in my opinion, and this was already confirmed in this news by using the newest drivers. Idle is once more just a few W.

 

[RIckyXDs]

I think It is because the Ryzen Mobile 2000, 3000, 4000 and 5000 series has battery drain issue when you install APU Drivers and Radeon Software is running in background. If you close Radeon Software you can increase Ryzen's mobile battery by 100% or in others words Double the battery autonomy.

I test the Ryzen 5 2500U, Ryzen 5 3500u and Ryzen 5 4600H... If you install any APU driver, and close manually Radeon Software in task manager, you can Double battery duration

 

[Mysteoa]

There are no changes with multi-monitor.

Why is this still happening? I use 4 screen setup and measure no difference between 1 and 4 monitors plugged in good old EVGA 1080TI (10W) but my gigabyte 5700XT draws +50W in idle with the same 4 screen setup. I wanna keep my power draw to a minimum when not gaming so I can have my open case system completely passively cooled when in idle (I hate PC noise). While 1080TI fans never turn when in idle and case is open, 5700XT fans do turn on from time to time and are quite audible. AMD should fix this ever persisting problem with their GPUs a long time ago. It should not be all that hard to fix it, given the fact that Nvidia has nailed it 6 years ago on much bigger die.

It draws more power because it has to run the memory at full speed. This it by design as it will introduce screen blanking(gliching) otherwise when changing memory frequency. I have seen it, and it is distracting every time I start a game. Overall, It's a different implementation than Nvidia.

If the cards gets over 53C is when the fans will start to spin and will stop when it reaches 45C. The cooler is just not open/big enough on your 5700XT to dissipate the heat. You could try putting a low RPM fan to blow over the card, so it doesn't start its own.

 

[Marecki_CLF]

How about fixing 165hz refresh on monitors, memory clock stays at full speed! Changing to 144 fixes the problem!

This is so weird. I have a 144Hz monitor - when running at 144Hz, the VRAM clock stays at full speed. Lowering the refresh rate down to 120Hz fixes the VRAM clock for me.

 

[defaultluser]

And meanwhile, AMD took 12 months to fix the stability bugs with RDNA1 - are we at-all surprised that their overworked driver team took six months to fix a feature that should have been fixed within a month of launch on Nvidia?

If you EVER wonder why AMD has a bad driver reputtaion - this is it! They always spend the first 9-12 months of a new arch wrestling with big CTD bug , and then spend the next two years attempting to fix the sea of "annoying, but not game-breaking" driver bugs.

At this point, Intel has a better Windows driver team than AMD does!

 

[Deleted member 205776]

I read this wrong just as I woke up, and thought this was about the latest chipset drivers, almost hyped myself lol

 

[watzupken]

And meanwhile, AMD took 12 months to fix the stability bugs with RDNA1 - are we at-all surprised that their overworked driver team took six months to fix a feature that should have been fixed within a month of launch on Nvidia?

If you EVER wonder why AMD has a bad driver reputtaion - this is it! They always spend the first 9-12 months of a new arch wrestling with big CTD bug , and then spend the next two years attempting to fix the sea of "annoying, but not game-breaking" driver bugs.

At this point, Intel has a better Windows driver team than AMD does!

Then by all means, be the first to use Intel's gaming graphics when they are released to prove the point that the driver is better on Intel. Driver situation during RDNA is bad primarily because its their first switch away from VEGA. However, it is still not excuse for bad drivers though. AMD improved driver stability significantly towards the tail end of the RDNA cycle and with the introduction of RDNA2.

I am agnostic to AMD or Nvidia cards as long as they are good value for money. Driver problems/ bugs will always persist. If you think Nvidia is better, then by all means, please stick to Nvidia. FYI, there are still quite a few reports of blacking out RTX 3xxx series cards and only resolved through under clocking. Nvidia appears to have "solved" the problem by means of a driver update, but in reality, some of these cards need a BIOS level update due to reduce aggressiveness of the clock speed boost mechanism. So it all boils down to silicon lottery for potential RTX 3xxx buyers. Between minor software bug vs one that is potentially hardware related, I am happy to stick with a minor software annoyance.

 

[GamerGuy]

Meanwhile, I'm having a fun time reading some of the salty posts from the other users who seems to love raining on AMD's parade...

 

[evernessince]

Looks more like fixing a bug. The rest of this thread is cancer, people always trying to making things Nvidia vs AMD. Improvements / fixes should be welcomed regardless of brand.

 

[Deleted member 205776]

Looks more like fixing a bug. The rest of this thread is cancer, people always trying to making things Nvidia vs AMD. Improvements / fixes should be welcomed regardless of brand.

Tech forums will always be filled with stockholm syndrome fanboys who think companies care about their existence.

 

[watzupken]

Looks more like fixing a bug. The rest of this thread is cancer, people always trying to making things Nvidia vs AMD. Improvements / fixes should be welcomed regardless of brand.

I agree. Any improvement at no cost is always a thing to welcome but not sure why people tend to like to side track and bring in other issues/ brands to compare. Even if it is a single digit % drop in power consumption at idle is still a win in my opinion, and here we are talking about a decent double digit % drop.