Radeon RX 6700, 6700 XT & 6750 XT users club

[amd64skater]

Do you use a custom fan curve?My stock curve seems way too agressive

Yes fan is always on for gaming. Lower rpms in the lower temps 60 plus it curves up. 52 degrees. I limit mine to about 180 fps in the game settings not in adrenaline. It also depends on resolution. I use 1080p 165hz. So if you use 1440p or 4k yeah the card runs harder and hotter. But 60fps thats kinda low unless your using 4k. I am mainly playing mw2 so I play at a 180 limit on it. Looks great bairly 50 degrees range plus or minus a few degrees. These cards can handle 1440p and lower at higher fps. Not sure on Ray tracing as I dont use it.

 

[cheermate]

Yes fan is always on for gaming. Lower rpms in the lower temps 60 plus it curves up. 52 degrees. I limit mine to about 180 fps in the game settings not in adrenaline. It also depends on resolution. I use 1080p 165hz. So if you use 1440p or 4k yeah the card runs harder and hotter. But 60fps thats kinda low unless your using 4k. I am mainly playing mw2 so I play at a 180 limit on it. Looks great bairly 50 degrees range plus or minus a few degrees. These cards can handle 1440p and lower at higher fps. Not sure on Ray tracing as I dont use it.

do you mind sharing your curve so ı can apply it to mine?I also just play cod mw 2

 

[amd64skater]

Hopefully that helps.

 

[Noci]

In my system specs it shows my benchmark scores. What CPU and ram are you running? Also make sure your resize bar is on and above 4g coding is enabled it makes a difference. Mine does have a slight faster factory stocked overclocked. Each manufacture and model is slightly different in factory settings and setup. But that don't matter the 6750XT is a beast. A little tweaking and yours will be faster. If you look at the TPU GPU database on the 6750XT https://www.techpowerup.com/gpu-specs/radeon-rx-6750-xt.c3879 . Some are the same some are different in each model.

View attachment 309193

I'm new to this thread and browising through these 22 pages I thought; why the hell, lets give this benchmark thing a go on my XFX Speedster QICK 319 RX 6700XT Black a go.

Didn't play around much with the Radeon software yet, but never too late to start. From previous attemts, I noticed the VRAM didn't like user interferrence at all, so I've played around with the sliders a bit except for the VRAM. (sorry the language is in Ducth, but the layout is the same)

When I slide them full to the right, 3DMark Time Spy crashes and no results, except for a message: an error occured. But with the settings in the pic above it pulls through. PPT limit for this setting is 242.650 W, but it peaked at 221.607 W, GPU Core Voltage peaked at 1.120 V and the current peaked at 12.676 A. Frequency peaked @ 2,671 MHz, that's below the 2,753 of the slider.
Temps, GPU peak 43 C, hotspot 74 C peak, nothing to worry about, got a full cover waterblock.

Time Spy Score with this, see below:

Any suggestions how to spice these scores up? For the CPU, it's no problem atm it is undervolted and not running on steriods .

By the way, I noticed somewhere in this thread somebody mentioned his card had Micron IC's for VRAM, GPU-Z tells me I have Samsung. Does this mayby explain why I cannot tickle my VRAM?

And before commenting, I know I'm runnig @ PCIe x16 3.0 instead of PCIe x16 4.0, it's defenitly due to the riser cable that came with the Thermaltake Core P90. Have to look for a replacement, but the A-brands are still f*cking expensive and I'm reluctant to try an AliBaba one, unless somebody can advise a really good one with a reasonable pricetag.

Additional the Superposition BM with the settings above v@ 1080P extreme

 

[cheermate]

Hopefully that helps.
View attachment 309244View attachment 309245View attachment 309251View attachment 309252

Do you mind sharing the screenshot of the ''Fine Tuning''part.Im really new to all this and dont want to make a mistake.Also ı should apply this to ''MW2''right?Do you use any other curves ?Thank you for all this btw.I really appreciate it.

 

[amd64skater]

Do you mind sharing the screenshot of the ''Fine Tuning''part.Im really new to all this and dont want to make a mistake.Also ı should apply this to ''MW2''right?Do you use any other curves ?Thank you for all this btw.I really appreciate it.

I didn't change the core or memory or voltage for mw2 it is factory tuned. Fan curve is usually the same on all games. I tend to lower the fans speeds and lower temps for noise but that is it. It's really not that loud beyond my case fan noise. I also have never heard it go up any rpms in that curve setting. Now I will set the minimum to like 90 percent across for benchmark purposes but that's it.

I'm new to this thread and browising through these 22 pages I thought; why the hell, lets give this benchmark thing a go on my XFX Speedster QICK 319 RX 6700XT Black a go.

Didn't play around much with the Radeon software yet, but never too late to start. From previous attemts, I noticed the VRAM didn't like user interferrence at all, so I've played around with the sliders a bit except for the VRAM. (sorry the language is in Ducth, but the layout is the same)
View attachment 309239

When I slide them full to the right, 3DMark Time Spy crashes and no results, except for a message: an error occured. But with the settings in the pic above it pulls through. PPT limit for this setting is 242.650 W, but it peaked at 221.607 W, GPU Core Voltage peaked at 1.120 V and the current peaked at 12.676 A. Frequency peaked @ 2,671 MHz, that's below the 2,753 of the slider.
Temps, GPU peak 43 C, hotspot 74 C peak, nothing to worry about, got a full cover waterblock.

Time Spy Score with this, see below:
View attachment 309240

Any suggestions how to spice these scores up? For the CPU, it's no problem atm it is undervolted and not running on steriods .

By the way, I noticed somewhere in this thread somebody mentioned his card had Micron IC's for VRAM, GPU-Z tells me I have Samsung. Does this mayby explain why I cannot tickle my VRAM?

View attachment 309243

And before commenting, I know I'm runnig @ PCIe x16 3.0 instead of PCIe x16 4.0, it's defenitly due to the riser cable that came with the Thermaltake Core P90. Have to look for a replacement, but the A-brands are still f*cking expensive and I'm reluctant to try an AliBaba one, unless somebody can advise a really good one with a reasonable pricetag.

Additional the Superposition BM with the settings above v@ 1080P extreme

View attachment 309253

That is awesome. See the 6700 series in general is a beast. Similar results on a non 50 version and they are about the same. The 6700 series in all is so underrated compared to the team green equivalent they hold their own.

 

[seoguru]

Hi guys, I think it's worth mentioning a few things that some might not be aware of, having a stronger CPU really boosts these cards I have built many with 6650, 6700 and 6750XT Variants and going from say 3600x to a 7900x or similar can push these cards by as much as 20% above card average performance scores not clocks, I will edit this post soon to prove it as I am writhing this on my Dell G15 gaming rig while I get the links on my PC rig.

 

[amd64skater]

Hi guys, I think it's worth mentioning a few things that some might not be aware of, having a stronger CPU really boosts these cards I have built many with 6650, 6700 and 6750XT Variants and going from say 3600x to a 7900x or similar can push these cards by as much as 20% above card average performance scores not clocks, I will edit this post soon to prove it as I am writhing this on my Dell G15 gaming rig while I get the links on my PC rig.

Of course a better Cpu is going to make the Gpu have better performance. I currently have a 5700x. Would like to upgrade to a 58003d or 5900x someday.

 

[cheermate]

I didn't change the core or memory or voltage for mw2 it is factory tuned. Fan curve is usually the same on all games. I tend to lower the fans speeds and lower temps for noise but that is it. It's really not that loud beyond my case fan noise. I also have never heard it go up any rpms in that curve setting. Now I will set the minimum to like 90 percent across for benchmark purposes but that's it.
View attachment 309260View attachment 309261


That is awesome. See the 6700 series in general is a beast. Similar results on a non 50 version and they are about the same. The 6700 series in all is so underrated compared to the team green equivalent they hold their own.

ı appreciate it.Thank you

 

[Lew Zealand]

I'm new to this thread and browising through these 22 pages I thought; why the hell, lets give this benchmark thing a go on my XFX Speedster QICK 319 RX 6700XT Black a go.

Didn't play around much with the Radeon software yet, but never too late to start. From previous attemts, I noticed the VRAM didn't like user interferrence at all, so I've played around with the sliders a bit except for the VRAM. (sorry the language is in Ducth, but the layout is the same)


When I slide them full to the right, 3DMark Time Spy crashes and no results, except for a message: an error occured. But with the settings in the pic above it pulls through. PPT limit for this setting is 242.650 W, but it peaked at 221.607 W, GPU Core Voltage peaked at 1.120 V and the current peaked at 12.676 A. Frequency peaked @ 2,671 MHz, that's below the 2,753 of the slider.
Temps, GPU peak 43 C, hotspot 74 C peak, nothing to worry about, got a full cover waterblock.

Time Spy Score with this, see below:


Any suggestions how to spice these scores up? For the CPU, it's no problem atm it is undervolted and not running on steriods .

You could try an 1150 undervolt but if you're running at the card's upper power limits, this will have a small effect on overall speed, likely less than 1%. Undervolting is most noticeable lower in the efficiency curve, like at 24-2500 MHz, and you're running on water so temps and power are likely not a concern.

By the way, I noticed somewhere in this thread somebody mentioned his card had Micron IC's for VRAM, GPU-Z tells me I have Samsung. Does this mayby explain why I cannot tickle my VRAM?

That could be, I happened to get lucky with my VRAM as it goes to 2150 MHz with no problems but then my core is more of a dud sample while yours looks quite good at 150MHz higher than mine at max OC. IMO the 67xx series is core-shy and VRAM bandwidth-heavy (compared to the 6600 thru 6800 XT) so your good cores offer a better benefit than my VRAM.

And before commenting, I know I'm runnig @ PCIe x16 3.0 instead of PCIe x16 4.0, it's defenitly due to the riser cable that came with the Thermaltake Core P90. Have to look for a replacement, but the A-brands are still f*cking expensive and I'm reluctant to try an AliBaba one, unless somebody can advise a really good one with a reasonable pricetag.

I have no PCIe 4.0, only 3.0 so no shade here. At least I have ReBAR/SAM!

 

[amd64skater]

ı appreciate it.Thank you

No problem.

 

[cheermate]

No problem.

How much of your Vram does COD use at extreme settings 1080p? Mine uses almost 11.5 gb ı think something is off

 

[Noci]

You could try an 1150 undervolt but if you're running at the card's upper power limits, this will have a small effect on overall speed, likely less than 1%. Undervolting is most noticeable lower in the efficiency curve, like at 24-2500 MHz, and you're running on water so temps and power are likely not a concern.

less than 1% is in my opinion within the margin of error and not worth spending time on it.
You are talking about the efficiency curve, is that comparable with the curve as presented with the curve optimizer of Ryzen CPU's? I'm just trying to comprehend how this works with GPU's nowadays, I always thought it was just a matter finding the sweetspot combination of power, frequency and temps where it maxes out and is stable for both the GPU & VRAM. Is Adrenalin also using some algorithm with fail saves to complicate it for noobs like me?

IMO the 67xx series is core-shy and VRAM bandwidth-heavy (compared to the 6600 thru 6800 XT) so your good cores offer a better benefit than my VRAM.

I have no PCIe 4.0, only 3.0 so no shade here. At least I have ReBAR/SAM!

But does that not imply that using PCIe 4.0 is better than PCIe 3.0, as it doubles the bandwidth for transfering data? I know the RX 6700XT VRAM is limited to a 192 bit data bus, but have no idea if it can fill up the 16 PCIe lanes, especially with SAM enabled where the CPU can fully access all of the 12 GB VRAM. Probably I'm just mixing up things now, please enlighten me.

Just for the record, I'm very happy with the out of the box performance of this graphics card, it's merely curiosity to play with the software to see what the card really can do with some tuning. I don't have the illusion I won the sillicon lottery or want to break records, it's just fun and to learn from it.

 

[amd64skater]

How much of your Vram does COD use at extreme settings 1080p? Mine uses almost 11.5 gb ı think something is off

It's usually 82% This is the settings I use. For the AA I use normal anything higher really didn't see a difference. Also it makes the card work harder for really nothing. Also I'm using a 170 frame limit in game once again not really seeing anything past it and your limiting the frames to get better performance. Each system is different so tweak to what makes you happy.

 

[Lew Zealand]

less than 1% is in my opinion within the margin of error and not worth spending time on it.
You are talking about the efficiency curve, is that comparable with the curve as presented with the curve optimizer of Ryzen CPU's? I'm just trying to comprehend how this works with GPU's nowadays, I always thought it was just a matter finding the sweetspot combination of power, frequency and temps where it maxes out and is stable for both the GPU & VRAM. Is Adrenalin also using some algorithm with fail saves to complicate it for noobs like me?

Adrenaline's fail-safe is that if you crash from undervolting too much to make it unstable, it will reset all values to default. You don't get stuck but you do need to keep track of your values so you don't forget your last safe settings.

There are a great number of stable settings for minimum power/temps at any power setting you choose. One setting for 214W (my card's max as reported in GPU-Z) which is what it seems you're referring to. But there are many others for 200W, 176W, and so on. And different places in each game load the GPU differently so that power and voltage will vary anyway depending on your settings, VSync, loads of things. I choose not to run at max wattage very often and instead at a reasonable VSync level for each game and generally don't want to stress a video card to 100% during usage

For example, my current settings are 2451 MHz cores, 1125V, 2110 MHz VRAM. That takes up to 190W in Time Spy at ~2305 MHz but only 160-170W in Fire Strike. And running Ark: SE at 60fps VSync it'll vary from 100-160W depending on the scene. It'll be only 70W in Rocket League at 144Hz VSync as that's such a light load with the card clocking itself down to about 850MHz because it doesn't need any more for 144Hz.

But does that not imply that using PCIe 4.0 is better than PCIe 3.0, as it doubles the bandwidth for transfering data? I know the RX 6700XT VRAM is limited to a 192 bit data bus, but have no idea if it can fill up the 16 PCIe lanes, especially with SAM enabled where the CPU can fully access all of the 12 GB VRAM. Probably I'm just mixing up things now, please enlighten me.

The 192-bit bus is for the working VRAM in the card and as long as you're not streaming textures from regular memory (like 2 games do this so far), PCIe 3.0 will not bottleneck performance more than ~1% if even that, as the way more powerful 4090 was down all of 2% at PCIe 3.0.

Just for the record, I'm very happy with the out of the box performance of this graphics card, it's merely curiosity to play with the software to see what the card really can do with some tuning. I don't have the illusion I won the sillicon lottery or want to break records, it's just fun and to learn from it.

I agree 100%. Honestly that's what I've spent the past few weeks doing with my card. I know my silicon is not golden, rather it's rust but it still performs well and is not down from the best numbers online by more than a few %. Considering the price I paid, it's worked out very well.

 

[cheermate]

It's usually 82% This is the settings I use. For the AA I use normal anything higher really didn't see a difference. Also it makes the card work harder for really nothing. Also I'm using a 170 frame limit in game once again not really seeing anything past it and your limiting the frames to get better performance. Each system is different so tweak to what makes you happy.

View attachment 309450View attachment 309451View attachment 309452View attachment 309453View attachment 309454View attachment 309455View attachment 309456

I know this is a bother but could you do me a favor and run your game at the highest settings with no fps limiter and tell me the vram and ram usage of your system.I have been having virtual memory errors and I'm trying to pinpoint what's causing the issue

 

[amd64skater]

I know this is a bother but could you do me a favor and run your game at the highest settings with no fps limiter and tell me the vram and ram usage of your system.I have been having virtual memory errors and I'm trying to pinpoint what's causing the issue

Yeah thats easy at the highest setting the card runs past 65 degrees vram is 90 percent and I lose almost 40 or more fps in COD. Your basically slamming the gas pedal to the floor and not letting off. Thats why I have a 170 limit. I can go lower by say 5 or 10 fps or higher a bit. Your really pushing it with almost no gains and alot of fps loss. But I don't have issues. Do you have a big enough power supply?

 

[seoguru]

Hi guys, I would like to pass on another tip to get most out of these cards, please don't use each card vendor's drivers in conjunction with AMD adrenalin drivers, it's a waste of time.
ASUS and MSI seems to fair the worst, Asus seems to do absolutely nothing at all no matter what you try, which is funny. With Asus cards, I personally max out all AMD drivers options and don't touch the voltage settings.
Most builders tend to stay away from user bench with good reason (the intel/Nvidia fanboys) but if you are on AM5 and you switch from dedicated cards to IGPU most user bench tests then claim there is a missing GPU if that happens disable the IGPU IN BIOS, so you can all have a proper laugh at userbench stupid algorithm that proves how stupid its design flaw is.

 

[cheermate]

Yeah thats easy at the highest setting the card runs past 65 degrees vram is 90 percent and I lose almost 40 or more fps in COD. Your basically slamming the gas pedal to the floor and not letting off. Thats why I have a 170 limit. I can go lower by say 5 or 10 fps or higher a bit. Your really pushing it with almost no gains and alot of fps loss. But I don't have issues. Do you have a big enough power supply?

My PSU is 650w gold MSI .My main issue right now is whenever i play on maxed settings my game crashes with the error of " ran out of virtual memory"I got 16gb ram but for some reason hwinfo shows ı have 30gb virtual memory created while playing Cod.Do you have hwinfo downloaded?This issue has been bothering me so so much these past daya

 

[amd64skater]

My PSU is 650w gold MSI .My main issue right now is whenever i play on maxed settings my game crashes with the error of " ran out of virtual memory"I got 16gb ram but for some reason hwinfo shows ı have 30gb virtual memory created while playing Cod.Do you have hwinfo downloaded?This issue has been bothering me so so much these past daya

I would try and reinstall COD.

 

[AusWolf]

My PSU is 650w gold MSI .My main issue right now is whenever i play on maxed settings my game crashes with the error of " ran out of virtual memory"I got 16gb ram but for some reason hwinfo shows ı have 30gb virtual memory created while playing Cod.Do you have hwinfo downloaded?This issue has been bothering me so so much these past daya

How much virtual memory do you have allocated in Windows?

 

[Noci]

How much virtual memory do you have allocated in Windows?

To jump in on this subject, I never understood why Windows needed virtual memory if there is a surplus of RAM.
So it triggered me to look it up.

https://learn.microsoft.com/en-us/t...rce=recommendations#system-managed-page-files

Not 100% clear for me, but I think I now understand a little of it.
Still plenty of questions unanswered also due to mixed info:
Windows 11 System properties tells me I have 4096 MB virtual memory assigned,
HWiNFO64 tells me I have out of almost 52 GB available physical memory, 14 GB virtual memory assigned, almost 50 GB available virtual memory (I know these numbers are dynamic and tend to vary on usage). Nevertheless there is a big discrespancy between the Windows set value and what HWiNFO64 reports. Leaves me puzzled.

Back on topic, could this paragraph of that above mentioned webpage could have something to do with it?

The Automatic memory dump feature initially selects a small paging file size. It would accommodate the kernel memory most of the time. If the system crashes again within four weeks, the Automatic memory dump feature sets the page file size as either the RAM size or 32 GB, whichever is smaller.

Cheermate did have some crashes and as I understand, Windows then automatically increases the virtual memory (page file) , but than again in his case max 16 GB.
But of course I can be totally wrong, with my interpretation a limited knowledge on software, might as well be the game itself.

 

[AusWolf]

To jump in on this subject, I never understood why Windows needed virtual memory if there is a surplus of RAM.
So it triggered me to look it up.

https://learn.microsoft.com/en-us/t...rce=recommendations#system-managed-page-files

Not 100% clear for me, but I think I now understand a little of it.
Still plenty of questions unanswered also due to mixed info:
Windows 11 System properties tells me I have 4096 MB virtual memory assigned,
HWiNFO64 tells me I have out of almost 52 GB available physical memory, 14 GB virtual memory assigned, almost 50 GB available virtual memory (I know these numbers are dynamic and tend to vary on usage). Nevertheless there is a big discrespancy between the Windows set value and what HWiNFO64 reports. Leaves me puzzled.

Back on topic, could this paragraph of that above mentioned webpage could have something to do with it?

The Automatic memory dump feature initially selects a small paging file size. It would accommodate the kernel memory most of the time. If the system crashes again within four weeks, the Automatic memory dump feature sets the page file size as either the RAM size or 32 GB, whichever is smaller.

Cheermate did have some crashes and as I understand, Windows then automatically increases the virtual memory (page file) , but than again in his case max 16 GB.
But of course I can be totally wrong, with my interpretation a limited knowledge on software, might as well be the game itself.

Some programs use the virtual memory even when you could fit their entirety into the physical RAM because they're written like that.

If I remember right, Warhammer 40k: Dawn of War 2 won't even start if you have virtual memory disabled, regardless of how much RAM you have.

 

[Ruru]

After having an aftermarket cooler, I can't but laugh how shitty that Powercolor Figher original cooler was. My 6700 XT runs waaaaaaaaay cooler now and I can't even hear the fan noise on load (2x bequiet PureWings 120mm with a 2x PWM -> GPU adapter)

 

[Noci]

Tip for the RX 6750 XT (reference card) owners here, if you were ever considering to watercool your GPU, but found it too costly, Alphacool has a great offer atm., 51% off.
I"m no way connected to that company, just a happy customer who knows from experience that they have good products.
The full cover block I use; Alphacool Eisblock GPX is doing a great job on my card.

please see link below:

GPU Wasserkühler für RX 6750 XT Reference | Aquatuning