The RX 6000 series Owners' Club

[ChristTheGreat]

At first glance, the voltage you use for 1800mhz is way to low (0.891), try it with something like 1030mv, see how it goes, if its not stable, raise it a bit, if its stable, lower it a bit until you get stable settings.

I use radeon tuner, set at 1150, but it wont change :/

 

[Felix123BU]

I use radeon tuner, set at 1150, but it wont change :/

Not sure what radeon tuner is, I would just use wattman, something like the below. Also, somethin came to mind, do not set both minimal and maximal frequency at the same value, that will misfire, use a 100 Mhz difference between the 2, with the desired frequency being between the 2, see below:

 

[ChristTheGreat]

Not sure what radeon tuner is, I would just use wattman, something like the below. Also, somethin came to mind, do not set both minimal and maximal frequency at the same value, that will misfire, use a 100 Mhz difference between the 2, with the desired frequency being between the 2, see below:

View attachment 196456

Looks like if I play and lock FPS, the voltage will stay low. When I run 3dmark time spy, goes to 1100 no problem. Seems I can't force it to keep high voltage and high clock, even if the card is not push to the max.

 

[Space Lynx]

I really have no idea how so many of you have problems. these have been my settings for two months now and 0 issues, so I don't know

 

[sepheronx]

Had issues again with mining on this GPU. It will auto ramp up speed of fans for a few seconds, then goes back to normal but then starts mining at 20 - 30 MHs from its initial 62MHs. This card is f'd.

I cant figure it out. So I am just gonna let it continue and monitor it. May have to send it in for RMA. I'll have to put it through rigorous gaming tests to see if there are any artifacts or anything screwing up the GPU causing it to act so wierd. Could also be the drivers and a possible format of the system and trying again may fix it.

Edit: Tested Witcher 3. No issues there.

I am wondering if my memory OC was too high and it was causing it to fail at mining? But usually when that happens, the miner will have a red lettering over the GPU meaning there is an issue. Mine never gave that. Actually, I would rarely have failed hashes either.

 

[Felix123BU]

Had issues again with mining on this GPU. It will auto ramp up speed of fans for a few seconds, then goes back to normal but then starts mining at 20 - 30 MHs from its initial 62MHs. This card is f'd.

I cant figure it out. So I am just gonna let it continue and monitor it. May have to send it in for RMA. I'll have to put it through rigorous gaming tests to see if there are any artifacts or anything screwing up the GPU causing it to act so wierd. Could also be the drivers and a possible format of the system and trying again may fix it.

Edit: Tested Witcher 3. No issues there.

I am wondering if my memory OC was too high and it was causing it to fail at mining? But usually when that happens, the miner will have a red lettering over the GPU meaning there is an issue. Mine never gave that. Actually, I would rarely have failed hashes either.

Wait, you are using those clocks for mining? No, that's wrong if its the case, if you mine ETH, set the clock to 1200mhz Core Min, 1300 Core Max, Voltage 881 or the lowest you can set, and memory to a max of 2100 or 10-20mhz lower if its unstable. That gives you around 61-62 Mhs at 150w. I let my 6800XT mine ETH whenever I can, it stays at 61.5 Mhs all the time without a hitch.
The fan auto ramp could be because it gets quite hot, using the settings above would help a lot with temps while mining.

 

[sepheronx]

Wait, you are using those clocks for mining? No, that's wrong if its the case, if you mine ETH, set the clock to 1200mhz Core Min, 1300 Core Max, Voltage 881 or the lowest you can set, and memory to a max of 2100 or 10-20mhz lower if its unstable. That gives you around 61-62 Mhs at 150w. I let my 6800XT mine ETH whenever I can, it stays at 61.5 Mhs all the time without a hitch.
The fan auto ramp could be because it gets quite hot, using the settings above would help a lot with temps while mining.

what?

No.

I had it 500 - 1500 core, lowest for voltage and mem at 2150. Also set it to fast timings on vram

I now switched it to 2100 for memory. It may have been the issue.

 

[Felix123BU]

what?

No.

I had it 500 - 1500 core, lowest for voltage and mem at 2150. Also set it to fast timings on vram

I now switched it to 2100 for memory. It may have been the issue.

Most likely, if mem is not stable it will drop the hash a lot over time, try to find a mem speed that remains stable, if it still drops over time, most likely the mem is not stable at a set speed. At 2150 and fast timing, mem operations will crap out giving low hash speed. From some forums I gathered info, 2100 is sort of the max with fast timings (depending on how good the mem chips are), and the difference between fast and normal timings is negligible. With normal timings you might get a few extra mem Mhz stable over 2100, but its not worth it really.

 

[sepheronx]

I am getting to around the same amount that I got before anyway at current 2100. I may lower it to just default timings for optimal stability then.

 

[turbogear]

I am getting to around the same amount that I got before anyway at current 2100. I may lower it to just default timings for optimal stability then.

It could very well be your memory is unstable. You should go in smaller steps towards 2150MHz to see what is highest without performance drop.
My experience with multiple RDNA2 cards shows that there are very few cards that can operate much higher than 2100MHz on memory.
Increasing it further than stable point drops the performance.

 

[Space Lynx]

It could very well be your memory is unstable. You should go in smaller steps towards 2150MHz to see what is highest without performance drop.
My experience with multiple RDNA2 cards shows that there are very few cards that can operate much higher than 2100MHz on memory.
Increasing it further than stable point drops the performance.

I get errors in game if I oc my vram at all. i can oc the gpu core hardcore with 0 issues though and that's all that really matters for fps increases in gaming

 

[sepheronx]

It could very well be your memory is unstable. You should go in smaller steps towards 2150MHz to see what is highest without performance drop.
My experience with multiple RDNA2 cards shows that there are very few cards that can operate much higher than 2100MHz on memory.
Increasing it further than stable point drops the performance.

Yeah, I figure this is the case. It's at 2100 and so far so good.

It runs fine in games now as well. I really got to upgrade cooling for this card. I refraine from watercooling as it's a messy expense. But I wonder, are there any AIO's planned for these GPUs besides the specialized models of GPUs?

 

[Garlic]

Upgraded from a 5700xt to this, so far performs great and junction temp never go above 65 when gaming. also surprised how far this can OC, 2920-2930 is when it starts to crash but 2910 is very stable.

Upgraded from a 5700xt to this, so far performs great and junction temp never go above 65 when gaming. also surprised how far this can OC, 2920-2930 is when it starts to crash but 2910 is very stable.

Meant to say junction temp never goes higher than 75 when gaming

 

[turbogear]

Upgraded from a 5700xt to this, so far performs great and junction temp never go above 65 when gaming. also surprised how far this can OC, 2920-2930 is when it starts to crash but 2910 is very stable.


Meant to say junction temp never goes higher than 75 when gaming

Welcome to the club.

 

[GamerGuy]

Upgraded from a 5700xt to this, so far performs great and junction temp never go above 65 when gaming. also surprised how far this can OC, 2920-2930 is when it starts to crash but 2910 is very stable.


Meant to say junction temp never goes higher than 75 when gaming

Awesome, welcome to the club!!!

 

[basco]

Mr. Garlic welcome and would ya plz upload a screenshot of gpu-z and is this stable in benchmarks?
the device id ?
thanks i am just interested because that would be an enormous overclock.

 

[Garlic]

Mr. Garlic welcome and would ya plz upload a screenshot of gpu-z and is this stable in benchmarks?
the device id ?
thanks i am just interested because that would be an enormous overclock.

Here, in games it sits around 2800-2830 and stable, but as soon as i go over 2915 it crashes. In furmark its quite a bit lower for some reason, prob because draws a lot more current than most things.

 

[Raendor]

May I now enter the club too?

received my upgrade yesterday and spent some time in the evening to re-assemble the rig, since I put a new mobo and cpu too.

now to find out what makes sense to do in terms of clocks and voltage. Much more heat coming out of it compared to old 1080.

 

[turbogear]

May I now enter the club too? View attachment 196703View attachment 196704
received my upgrade yesterday and spent some time in the evening to re-assemble the rig, since I put a new mobo and cpu too.

now to find out what makes sense to do in terms of clocks and voltage. Much more heat coming out of it compared to old 1080.

Wow that's really compact.
I hope you would be able to cool it well in this small case.

 

[Raendor]

Wow that's really compact.
I hope you would be able to cool it well in this small case.

Ran RDR2 a bit and when it’s at 254w mark (guess it’s tdp limit as I haven’t seen it higher) it was no higher than 78 C at default fan curve. Didn’t tweak anything yet as I’ve got no time. So it’ll be a slow and gradual process. Case is fine. Mesh everywhere and 2 120 intake fans under gpu and 120 exhaust on side panel doing well. But I’d definitely undervolt before summertime.

 

[Felix123BU]

May I now enter the club too? View attachment 196703View attachment 196704
received my upgrade yesterday and spent some time in the evening to re-assemble the rig, since I put a new mobo and cpu too.

now to find out what makes sense to do in terms of clocks and voltage. Much more heat coming out of it compared to old 1080.

"May I now enter the club too?" - Absolutely, 1000%, welcome mate

 

[Garlic]

Ran RDR2 a bit and when it’s at 254w mark (guess it’s tdp limit as I haven’t seen it higher) it was no higher than 78 C at default fan curve. Didn’t tweak anything yet as I’ve got no time. So it’ll be a slow and gradual process. Case is fine. Mesh everywhere and 2 120 intake fans under gpu and 120 exhaust on side panel doing well. But I’d definitely undervolt before summertime.

78 Junction temp? if so thats really good, if not you might want to look at the junction temp since its almost always 10-20C hotter

 

[Raendor]

78 Junction temp? if so thats really good, if not you might want to look at the junction temp since its almost always 10-20C hotter

Not sure now which temp I was looking at, but I guess general one rather than junction. Will take a closer look, but some undervolt won’t hurt for sure.

"May I now enter the club too?" - Absolutely, 1000%, welcome mate

Honestly I‘m wondering if I should keep 6800 xt or just put 6800 in my system instead (which I ordered earlier, but is coming later lol). Of course I like more power and looks of midnight black but 6800 is a bit smaller and less power-hungr. On the other hand having 12 more CUs is great and nothing stops me from tweaking the card closer to 6800 consumption levels, just need to study a bit as I haven’t used radeon since 2004
my only real complaint is coil whine, which apparently is quite often forthcoming on these reference models.

How’s it for you guys? My psu is sf750 and less than year old, so I don’t think that has anything to do with it.

 

[GamerGuy]

78 Junction temp? if so thats really good, if not you might want to look at the junction temp since its almost always 10-20C hotter

Wow, 78C is pretty awesome, my Nitro+ RX 6900 XT Junction Temp hangs around 83-85C, and the GPU temp itself is around mid 70, so their temp delta between them is about 20C.

To all those guys who've recently gotten their RX 6000 cards....WELCOME!!!

 

[Garlic]

After playing around with wattman and games for a few hours I was able to push the gpu a bit more while still having junction temp under 80C, this actually looks like a pretty good silicon. BUT I noticed that despite me taking the voltage down to 1125 it still goes to 1200 for some reason, if anyone know why this is the case please let me know