AMD RX 7000 series GPU Owners' Club

[Vya Domus]

However, the coil whine is loud

That's actually a good thing believe it or not, buildzoid explained in an older video that the types of inductors that whine a lot are the best in terms of quality.

 

[remekra]

That's actually a good thing believe it or not, buildzoid explained in an older video that the types of inductors that whine a lot are the best in terms of quality.

I need to look for that video from him. It might be true, however they are really audible. Even when there are some changes in frametime, they squeak in different frequencies. It is annoying, to the point where I would rather have worse inductors but the ones that are not that loud.

 

[tabascosauz]

I need to look for that video from him. It might be true, however they are really audible. Even when there are some changes in frametime, they squeak in different frequencies. It is annoying, to the point where I would rather have worse inductors but the ones that are not that loud.

It's a bit of a stretch to call coil whine "good" lol, but the point of the video is that choice of inductors like choice of mosfets needs to fit the application, and the chokes commonly used on GPUs are usually just prone to coil whine. Changing pitch/character based on load/fps/power is expected.

You won't find those "worse but quiet" inductors he talks about because they just don't make the cut for 300W+ GPUs. You probably won't even find them in the Vcore VRM of budget boards nowadays after everyone completely switched to 50A DrMOS minimum.

Choose the Gaming OC for its cooling solution, not in hopes that it has no coil whine. I certainly hope yours will be a quiet card, but coil whine is as unpredictable as the silicon lottery - I'd bet that every 250W+ GPU SKU sold on the market today will have at least one owner complaining about coil whine, regardless of brand and model.

XT and XTX MBA cards share the exact same VRM component choices - mine is quiet.

 

[remekra]

It's a bit of a stretch to call coil whine "good" lol, but the point of the video is that choice of inductors like choice of mosfets needs to fit the application, and the chokes commonly used on GPUs are usually just prone to coil whine. Changing pitch/character based on load/fps/power is expected.

You won't find those "worse but quiet" inductors he talks about because they just don't make the cut for 300W+ GPUs. You probably won't even find them in the Vcore VRM of budget boards nowadays after everyone completely switched to 50A DrMOS minimum.

Choose the Gaming OC for its cooling solution, not in hopes that it has no coil whine. I certainly hope yours will be a quiet card, but coil whine is as unpredictable as the silicon lottery - I'd bet that every 250W+ GPU SKU sold on the market today will have at least one owner complaining about coil whine, regardless of brand and model.

XT and XTX MBA cards share the exact same VRM component choices - mine is quiet.

Found that video, I know what he might have meant about that the better inductors are more prone to coil whine.
Yeah it is a lottery, in the case of my MBA I lost pretty badly, as you could record a symphony of it. Just like buildzoid mentioned in the video, it's frequencies are all over the place.
Problem is exaggerated as I have my case on my desk, with Liquid Freezer 360 AIO and 5800X3D and some P14 fans so overall it's dead quiet,
I don't mind some coil whine, like I said had a 3080 that did it, but it was not that kind of bad.

EDIT
Yeah the cooling solution is main point. From the looks only I would choose MBA, but for performance I will try that Gaming OC.

EDIT2

Picked up the Gaming OC as mentioned. On the OC Bios by default it draws around 10W more than reference. Hotspot then reaches max 80 degrees. But the fan curve is a little bit too aggressive for me. Once it reaches over 50% (>2000RPM) its getting into loud territory. Little coil whine, but a lot less than the MBA.

At the stock it reaches 15 800 points in Port Royale (100 more than reference on stock). So far I applied almost the same settings as on MBA, so 1100mv, +10% PL, 2600MHz with Fast Timings on the memory. But since PL is higher even on stock it sits at 400W. Limited the fan speed to 50% and voila - max hot spot temp at around 87 but that depends on the load. Fans are audible but not loud.
Also reaches Zero RPM mode more easily. MBA after watching some videos and browsing could start the fan to spin at 500RPM, this one as soon as you exit the game goes to 0RPM.

Score jumped to 16800, with raising the power limit to +15 - 17000. But to run it at that fans would need to spin faster as hotspot creeps over 90 degrees, but still not reaching over 105 like MBA did, if someone doesn't mind the noise then it's doable. Improvement in games varies but around 5 fps or more, free performance is free performance (well it draws more electricity).
If someone prefers efficiency then you can leave it at default and limit the fan speed to 35-40%, then it's dead quiet.

Will need to test stability more, as in benchmarks I could easily go as down as 1050mv and it would not crash but run some games and it will, AMD clocks are a lot more dynamic than nvidia I would say, with a nvidia card you just set the clock in Afterburner and if it will not crash in 3dmark then it probably won't anywhere else, but with Radeon one game will run on lower end of the clocks spectrum while in the other it will go to the upper limit and crash, overall its a lot more random and requires more testing.

Overall build quality is good. Shroud is plastic but the backplate is metal with a large pad sitting on the core. Still like the design of MBA more, but it's not a bad looking card by any means.
And since it's a little over 2 slot it would be a great alternative if someone wants to do SFF build.
A shame that it doesn't get any reviews as at least in my country it's priced very good compared to other brands and MBA and it has a lot better cooling than the MBA.

 

[QuietBob]

Officially joining the club with an Asrock MBA 7900XTX

I've already done some testing, mostly running benchmarks to gage the power draw and maximum temperatures. Zero problems so far with the card or the driver. I'm using the latest 23.1.1, stock settings for the card, and a high quality uncertified DP 1.4 cable to drive a 4K60 panel. Running newest HWinfo 7.37 beta which enables more monitoring features for Navi 31.

The card doesn't seem to be affected by the vapor chamber quality issue. In a 21c room the highest temps I've seen when running benchmarks are 67c edge and 81c hotspot. I'm a little worried about the VRAM temp, as it reached 92c. Still lower than the 110c limit, but enough for me to get concerned. Might be poor quality or carelessly placed thermal pads. In general, when benchmarking the temps stay below 65/79c for the GPU and 88c for the memory.

When fully idle, the card draws around 15w at 36-51c edge, 39-53c hotspot and 60-72c VRAM. The reason for the great variation is the Zero RPM feature. The temps will creep up until the fans kick in, bringing them back down. Again, I don't like the memory temp. I'll see how disabling Zero RPM or setting a custom fan curve will affect it.

Maximum TBP in benchmarks was 362w, with the "GPU power maximum" figure at 546w, as measured by HWinfo with a 1000ms polling rate. Honestly, I don't know what to think of this number. It's way off the 375w provided by the two 8-pin connectors and the slot. At the same time my watt meter recorded a maximum of 456w for the whole 5800X3D system. Right now I'm clueless as to what that statistic represents.

YouTube power consumption is 40-45w at 720p and 1080p, 45-50w at 1440p, and 55-60w at 2160p. These numbers are still high, buth AMD promised further refinements in that area.

By the way, I'm using a Focus GX-650. I've got a Prime TX-750 set aside, but wanted to test with my current PSU first. The minimum recommendation is 800w. Still, it would appear that even a quality 550w unit might run a reference 7900XTX at stock speeds.

Under load, the fans stay below 2000 rpm most of the time. They're audible, but not disturbing. My case sits on the desk just beside me, to my right. It's an old mesh model with a single 1100rpm intake and a single 1300rpm exhaust. It's got two empty 140mm fan mounts in the left panel -- one directly above the graphics card and the other over the CPU cooler. This means that I hear the GPU and CPU fans far more distinctly than in a typical modern case. The card has minor coil whine, which intensifies when running ridiculous frame rates. I imagine most users won't be able to hear the card in an RGB case with ten fans

I'll be doing more testing. My aim is to run the 7900XTX at 4K60 with best efficiency

 

[remekra]

When fully idle, the card draws around 15w at 36-51c edge, 39-53c hotspot and 60-72c VRAM. The reason for the great variation is the Zero RPM feature. The temps will creep up until the fans kick in, bringing them back down. Again, I don't like the memory temp. I'll see how disabling Zero RPM or setting a custom fan curve will affect it.

Maximum TBP in benchmarks was 362w, with the "GPU power maximum" figure at 546w, as measured by HWinfo with a 1000ms polling rate. Honestly, I don't know what to think of this number. It's way off the 375w provided by the two 8-pin connectors and the slot. At the same time my watt meter recorded a maximum of 456w for the whole 5800X3D system. Right now I'm clueless as to what that statistic represents.

YouTube power consumption is 40-45w at 720p and 1080p, 45-50w at 1440p, and 55-60w at 2160p. These numbers are still high, buth AMD promised further refinements in that area.

Wouldn't worry about mem junc temps. Had the same on MBA, no matter if overclocked or not. Even on 2700Mhz on them they always stopped at 96.
Also have the same high temp wattage in HWinfo. Don't know if those are transient spikes or something else. I'm running 750W PSU and so far it didn't trip the safety, even when running the new card at 400W.

 

[tabascosauz]

Officially joining the club with an Asrock MBA 7900XTX

View attachment 280191

I've already done some testing, mostly running benchmarks to gage the power draw and maximum temperatures. Zero problems so far with the card or the driver. I'm using the latest 23.1.1, stock settings for the card, and a high quality uncertified DP 1.4 cable to drive a 4K60 panel. Running newest HWinfo 7.37 beta which enables more monitoring features for Navi 31.

The card doesn't seem to be affected by the vapor chamber quality issue. In a 21c room the highest temps I've seen when running benchmarks are 67c edge and 81c hotspot. I'm a little worried about the VRAM temp, as it reached 92c. Still lower than the 110c limit, but enough for me to get concerned. Might be poor quality or carelessly placed thermal pads. In general, when benchmarking the temps stay below 65/79c for the GPU and 88c for the memory.

When fully idle, the card draws around 15w at 36-51c edge, 39-53c hotspot and 60-72c VRAM. The reason for the great variation is the Zero RPM feature. The temps will creep up until the fans kick in, bringing them back down. Again, I don't like the memory temp. I'll see how disabling Zero RPM or setting a custom fan curve will affect it.

Maximum TBP in benchmarks was 362w, with the "GPU power maximum" figure at 546w, as measured by HWinfo with a 1000ms polling rate. Honestly, I don't know what to think of this number. It's way off the 375w provided by the two 8-pin connectors and the slot. At the same time my watt meter recorded a maximum of 456w for the whole 5800X3D system. Right now I'm clueless as to what that statistic represents.

YouTube power consumption is 40-45w at 720p and 1080p, 45-50w at 1440p, and 55-60w at 2160p. These numbers are still high, buth AMD promised further refinements in that area.

By the way, I'm using a Focus GX-650. I've got a Prime TX-750 set aside, but wanted to test with my current PSU first. The minimum recommendation is 800w. Still, it would appear that even a quality 550w unit might run a reference 7900XTX at stock speeds.

Under load, the fans stay below 2000 rpm most of the time. They're audible, but not disturbing. My case sits on the desk just beside me, to my right. It's an old mesh model with a single 1100rpm intake and a single 1300rpm exhaust. It's got two empty 140mm fan mounts in the left panel -- one directly above the graphics card and the other over the CPU cooler. This means that I hear the GPU and CPU fans far more distinctly than in a typical modern case. The card has minor coil whine, which intensifies when running ridiculous frame rates. I imagine most users won't be able to hear the card in an RGB case with ten fans

I'll be doing more testing. My aim is to run the 7900XTX at 4K60 with best efficiency

So far the AIB cards run VRAM similarly warm, and the coolers and VRAM baseplates on there are no slouch. Probably just another Ryzen thermals type paradigm shift that we have to get used to. About the same as GDDR6X.

If you want you can adjust your fan curve to be a bit flatter so that at low loads you get more fan speed, without changing load fan speeds. VRAM should run a bit cooler and card should hit fan stop a bit faster after games (usually now a minute or two with the custom curve for me).

I agree on the fan speed - it's not as quiet as other cards but I much prefer hearing a bit of pleasant whoosh than silent fans and screaming coil whine.

Don't forget that the XTX MBA also incorporates the fan intake temp sensor beneath the third fan, which AIBs and XT MBA do not have. I don't think it changes fan speed or performance characteristics at the moment, but AMD hinted that it might change.

I just ignore the max power metric in HWInfo. It's consistently 100W+ higher than TBP, even at idle.

 

[QuietBob]

I disabled Zero RPM and it made a huge difference to idle temps. The fans now run at inaudible 620 rpm. Edge is 29-35c, hotspot 32-38c and VRAM 54-60c. A whopping 15c difference in max temps for the GPU and 12c for the memory. Or, to put it another way, the new maximum is the old minimum. I'm finally happy with the VRAM temp, seeing 72c at idle was plain absurd. I would definitely recommend disabling Zero RPM, at least on MBA cards.

Took some measurements during a full Shadow of the Tomb Raider benchmark with maximum quality at 4K (no RT/AA). The first run is ungated and the second with 60Hz V-sync:

The differences are huge. 196w vs. 346w TBP for 150w savings, or 43% lower power consumption on average Edge temp is 5c down and hotspot 13c. Fans were at 1000rpm as opposed to 1500rpm for most of the bench. CPU uses 43w rather than 50w and the temp is a bit lower.

The efficiency results look very promising so far, but I've come across some errors. Final Fantasy XV benchmark would sometimes exit for no reason, on occasion leaving the system unresponsive and forcing a reset. I don't think it's a power delivery issue, as limiting the fps to 60 had no effect. SOTTR benchmark does the same everytime I set RT to high or ultra. Curiously, the game itself starts fine with these settings.

Need to do more testing

 

[Space Lynx]

I disabled Zero RPM and it made a huge difference to idle temps. The fans now run at inaudible 620 rpm. Edge is 29-35c, hotspot 32-38c and VRAM 54-60c. A whopping 15c difference in max temps for the GPU and 12c for the memory. Or, to put it another way, the new maximum is the old minimum. I'm finally happy with the VRAM temp, seeing 72c at idle was plain absurd. I would definitely recommend disabling Zero RPM, at least on MBA cards.

Took some measurements during a full Shadow of the Tomb Raider benchmark with maximum quality at 4K (no RT/AA). The first run is ungated and the second with 60Hz V-sync:

View attachment 280456View attachment 280457

The differences are huge. 196w vs. 346w TBP for 150w savings, or 43% lower power consumption on average Edge temp is 5c down and hotspot 13c. Fans were at 1000rpm as opposed to 1500rpm for most of the bench. CPU uses 43w rather than 50w and the temp is a bit lower.

The efficiency results look very promising so far, but I've come across some errors. Final Fantasy XV benchmark would sometimes exit for no reason, on occasion leaving the system unresponsive and forcing a reset. I don't think it's a power delivery issue, as limiting the fps to 60 had no effect. SOTTR benchmark does the same everytime I set RT to high or ultra. Curiously, the game itself starts fine with these settings.

Need to do more testing

so you didn't adjust fan curve at all, you just disabled zero rpm and got those temperature decreases in game? that is impressive if so...

 

[tabascosauz]

The efficiency results look very promising so far, but I've come across some errors. Final Fantasy XV benchmark would sometimes exit for no reason, on occasion leaving the system unresponsive and forcing a reset. I don't think it's a power delivery issue, as limiting the fps to 60 had no effect. SOTTR benchmark does the same everytime I set RT to high or ultra. Curiously, the game itself starts fine with these settings.

Need to do more testing

If you're 100% sure power is not an issue, might be time to check your DP cables. That, or just another driver bug

Mine slowly creeps up to about 70C VRAM idle too (before fans inevitably come on again from >50C edge temp) but I really don't see it as an issue warranting disabling fan stop. Neither XTX nor XT MBA backplate helps out with VRAM cooling, but even cards that do (Sapphire) aren't much different.

Delta Superflo may be a highly upgraded sleeve bearing but they're a far cry from ball bearing fans (Asus, XFX and Sapphire). I'd much prefer to just leave them off when they don't need to be on.

so you didn't adjust fan curve at all, you just disabled zero rpm and got those temperature decreases in game? that is impressive if so...

lol no I wish

the temp/power improvements come from 60Hz framecap

 

[QuietBob]

If you're 100% sure power is not an issue, might be time to check your DP cables. That, or just another driver bug

I can't be sure, I'm running a 650w PSU against the 800w minimum recommendation But I've yet to see more than 500w at the wall. In W1zzard's testing the card spiked to 455w, so theoretically I should be covered. I will eventually replace the power supply with a 750w Prime.

so you didn't adjust fan curve at all, you just disabled zero rpm and got those temperature decreases in game?

Disabling Zero RPM brought idle temps down. What you see in the charts are the results of enabling V-sync at 60 Hz.

 

[AusWolf]

I disabled Zero RPM and it made a huge difference to idle temps. The fans now run at inaudible 620 rpm. Edge is 29-35c, hotspot 32-38c and VRAM 54-60c. A whopping 15c difference in max temps for the GPU and 12c for the memory. Or, to put it another way, the new maximum is the old minimum. I'm finally happy with the VRAM temp, seeing 72c at idle was plain absurd. I would definitely recommend disabling Zero RPM, at least on MBA cards.

Took some measurements during a full Shadow of the Tomb Raider benchmark with maximum quality at 4K (no RT/AA). The first run is ungated and the second with 60Hz V-sync:

View attachment 280456View attachment 280457

The differences are huge. 196w vs. 346w TBP for 150w savings, or 43% lower power consumption on average Edge temp is 5c down and hotspot 13c. Fans were at 1000rpm as opposed to 1500rpm for most of the bench. CPU uses 43w rather than 50w and the temp is a bit lower.

The efficiency results look very promising so far, but I've come across some errors. Final Fantasy XV benchmark would sometimes exit for no reason, on occasion leaving the system unresponsive and forcing a reset. I don't think it's a power delivery issue, as limiting the fps to 60 had no effect. SOTTR benchmark does the same everytime I set RT to high or ultra. Curiously, the game itself starts fine with these settings.

Need to do more testing

That's why I always recommend setting an FPS cap instead of tumbling down the bumpy road of undervolting and downclocking.

 

[Space Lynx]

That's why I always recommend setting an FPS cap instead of tumbling down the bumpy road of undervolting and downclocking.

Yeah, I run my fps cap at 160 on a 165hz monitor. Works like a charm

 

[tabascosauz]

DCS is in the testing rotation this week. Pushes the GPU harder than any other game I play.

Or technically, it's very VRAM intensive. VRAM peaked at about 88C, compared to 82C on the 3070Ti's GDDR6X. It doesn't hit the core as hard but still 100% util and core clocks are high (stuck like glue to 2800MHz).

• 3070 Ti: 80-100fps on the ground, 165fps cap in the air
• 7900XT: 100-140fps on the ground, 165fps cap in the air

Surprisingly, it's a butter experience if I just stick to the usual gaming routine (NOTHING open/unminimized on main monitor aside from game, NOTHING open/unminimized on secondary monitor). Smoother than either of my previous Geforce cards. CPU load and temps seem much lower than on Geforce.

DLSS and Vulkan are both coming soon to DCS, however. Might shake up the status quo.

I can't be sure, I'm running a 650w PSU against the 800w minimum recommendation But I've yet to see more than 500w at the wall. In W1zzard's testing the card spiked to 455w, so theoretically I should be covered. I will eventually replace the power supply with a 750w Prime.

Kinda hard to draw much power with a 60fps cap, though.

 

[Vya Domus]

Ok, so the fans do turn off sometimes on my reference model, it's pretty random, I still can't figure out what the criteria is, I think what happens is that there must be some program that puts some utilization on the GPU and it causes the fan stop to not trigger.

It's very odd because I look at the fan speed as temps reach a new minimum and for some reason this time they don't shut off even though at some point previously they did.

 

[tabascosauz]

Ok, so the fans do turn off sometimes on my reference model, it's pretty random, I still can't figure out what the criteria is, I think what happens is that there must be some program that puts some utilization on the GPU and it causes the fan stop to not trigger.

It's very odd because I look at the fan speed as temps reach a new minimum and for some reason this time they don't shut off even though at some point previously they did.

RGB software? I slimmed down my HWInfo and it doesn't negatively affect fan stop.

Fan control is still a bit buggy atm. Not much can do aside from get temps low as possible, get power low as possible, and avoid guaranteed fan spin-up loads (any form of video playback).

If I'm afk and my monitor is always on, it will fan stop as usual (20W idle), but if my monitor is allowed to turn itself off (without comp going into sleep), I will come back to find the GPU fans spinning for no good reason.

 

[remekra]

I've had exactly the same behaviour on my MBA, so I think it will stay like that until AMD fixes power management in driver updates.

 

[brutlern]

I just replaced my old Gigabyte 2080 Super with a brand spanking new XFX AMD Radeon RX 7900 XT MERC310

P.S: The 2080S is NOT a small card.

 

[AusWolf]

Just a thought: what if the high idle power consumption is due to the chiplet design, and cannot be fixed?

I mean,
1. Chiplet-based AMD CPUs consume more power than monolithic Intel ones at Idle. 15-20 W isn't unreasonable on a Ryzen sitting at the Windows desktop.
2. Past Radeons consumed way less at idle. Why did the idle power jump now?

 

[tabascosauz]

Just a thought: what if the high idle power consumption is due to the chiplet design, and cannot be fixed?

I mean,
1. Chiplet-based AMD CPUs consume more power than monolithic Intel ones at Idle. 15-20 W isn't unreasonable on a Ryzen sitting at the Windows desktop.
2. Past Radeons consumed way less at idle. Why did the idle power jump now?

It's not a complete impossibility, since a fair bit of idle power comes from an unreported source, just like chiplet Ryzen (not from core, SOC, or two Vmem rails). I doubt Navi 31 will reach single digit idle power for that reason.

Then again, multi monitor is the only outlier now. On 23.1.1, depending on monitor you can run 1 x 1440p165 + 1 x 1440p60 screen and still idle at the same 20W you would be with just one screen. Video playback power is now also the same as Navi 21 (40-50W) - the main "bug" that AMD was after. You just can't step up Screen 2 above 60Hz.

Also, faster GDDR6 + 50-100% wider memory interface.

More likely the multimonitor "bugs" will never be fixed because it's the same problem as Vega and Navi 1/2 and AMD stopped caring. Like with those GPUs it's always a matter of keeping VRAM clock high enough to prevent artifacting on a given screen setup.

Nvidia cards face the same problem, but 30/40 series can generally handle "more" monitor before clocking up VRAM.

Also, Nvidia usually still has an intermediate Pstates for VRAM, so while 1000MHz is not minimum idle power it still is only 10-20W more. For RX7000 there is no middle ground, it is either VRAM idle or game clock, there is no in-between.

 

[AusWolf]

For RX7000 there is no middle ground, it is either VRAM idle or game clock, there is no in-between.

That's very strange. Even my 6750 XT can change its VRAM clock to suit the need.

 

[AnotherReader]

More likely the multimonitor "bugs" will never be fixed because it's the same problem as Vega and Navi 1/2 and AMD stopped caring. Like with those GPUs it's always a matter of keeping VRAM clock high enough to prevent artifacting on a given screen setup.

I used to run my Vega 64 with a 1440p 144 Hz and an ultrawide 1080p at 75 Hz. It stayed at idle clocks.

 

[tabascosauz]

That's very strange. Even my 6750 XT can change its VRAM clock to suit the need.

I think there do exist intermediate steps for game clock, but seemingly mostly around the 200MHz mark (which is not very useful as it barely offers any more bandwidth). Might be another at 900-1000MHz. But the GPU currently does not use any in-between clock above 200MHz as far as idle is concerned.

Pretty stupid if you ask me. AMD allows partial VRAM downclock in-game and expectedly just results in an incredibly shitty stuttery experience. AMD does NOT allow it in idle, where it would actually make a positive difference. I doubt half clock is "insufficient" to drive 2 1440p165 displays.

Hard to tell on OSD alone, without actively looking at HWInfo which I don't want to do because of the stutters.

I used to run my Vega 64 with a 1440p 144 Hz and an ultrawide 1080p at 75 Hz. It stayed at idle clocks.

Vega 64 had like a 15W multimonitor draw. Navi 21 is 40W, Navi 31 is 80-100W. Can't find the reddit thread anymore talking about whether it was due to HBM.

 

[AnotherReader]

Vega 64 had like a 15W multimonitor draw. Navi 21 is 40W, Navi 31 is 80-100W. Can't find the reddit thread anymore talking about whether it was due to HBM.

It probably is the HBM. HBM is significantly more efficient than GDDR6.

 

[AusWolf]

I think there do exist intermediate steps for game clock, but seemingly mostly around the 200MHz mark (which is not very useful as it barely offers any more bandwidth). Might be another at 900-1000MHz. But the GPU currently does not use any in-between clock above 200MHz as far as idle is concerned.

Pretty stupid if you ask me. AMD allows partial VRAM downclock in-game and expectedly just results in an incredibly shitty stuttery experience. AMD does NOT allow it in idle, where it would actually make a positive difference. I doubt half clock is "insufficient" to drive 2 1440p165 displays.

Hard to tell on OSD alone, without actively looking at HWInfo which I don't want to do because of the stutters.

It is pretty stupid, considering that 6000 series does the exact opposite - adjust the VRAM clock dynamically on the desktop and during video playback, and use the maximum during games.