Overclock.net banner
Cancel
Create thread New Forums

[Official] AMD Radeon RX 6800 XT Owner's Club

Tags
6800 xt amd radeon rdna2
1 reading
Jump to Latest
1,361 - 1,380 of 1,702 Posts
#1,361 ·


This guy got his 6800XT to score higher than a 3090ti on 3D Mark. I don't know how to use MorePowerTool. You need to load your bios/MPT file, where do you find that?

MorePowerTool - Navi and Radeon VII Tweaking and Overclocking Software | Download (Update 1.1.1) | igor'sLAB

I have a Gigabyte Gaming OC 6800XT. I don't know if using the MorePowerTool will make a difference but I overclocked it to what he had on 3D Mark and played a game but my driver crashes within 20 minutes-1 hour of playing a game after an overclock that high.
 
#1,362 ·
MagnusFarseeR said:


This guy got his 6800XT to score higher than a 3090ti on 3D Mark. I don't know how to use MorePowerTool. You need to load your bios/MPT file, where do you find that?

MorePowerTool - Navi and Radeon VII Tweaking and Overclocking Software | Download (Update 1.1.1) | igor'sLAB

I have a Gigabyte Gaming OC 6800XT. I don't know if using the MorePowerTool will make a difference but I overclocked it to what he had on 3D Mark and played a game but my driver crashes within 20 minutes-1 hour of playing a game after an overclock that high.
Click to expand...
Hmm interesting, they say that reducing GDDR6 voltage will improve performance by reducing temperatures. But does it make any sense as these are very cold? 70°C max for me, that's not high. Lower voltage = Lower stable clocks
 
#1,365 ·
1.2
 
Save
Reply Quote Rep+ Rep+
#1,367 ·
Of course not ) for 24/7 the stock power and voltage limits are more than enough for me.
 
Save
Reply Quote Rep+ Rep+
#1,369 ·
Astawowski said:
View attachment 2605909 Hmm, that's weird... HWInfo says that GPU Voltage under load is ~1.1V, however both GPU-Z and AMD Adrenalin show 1.15V. Which one of them is telling the truth?
Click to expand...
The load line on these parts has considerable droop. 1.15v sounds like the set voltage, and I'd expect that to equal ~1.1v, or less in, gaming loads.
 
Save
Reply Quote Rep+ Rep+
#1,370 · (Edited)
Yeah, we can only guess what is the source for the SMU GfxVdd, reported by Adrenalin, GPU-Z, etc.

VDDCR_GfxVdd in HWInfo is SVI2 telemetry Vdd provided by the Gfx loop of the VRM controller, and as @Blameless said it's affected by load line settings in vbios at boot and on-the-fly.

Hard to say which one is better for practical use since none of them seem to tell us about the actual Vdd supplied to the Gfx core. Although, with the SVI2 I can at least see the expected V/F correlation in different workloads.
 
Save
Reply Quote Rep+ Rep+
#1,372 ·
  • Rep+
Reactions: Insidious Supra
Save
Reply Quote Rep+ Rep+
#1,374 · (Edited)
The simple answer is the reasons behind lowering voltage have little to do with cooling.
 
Save
Reply Quote Rep+ Rep+
#1,376 ·
Astawowski said:
View attachment 2606380 Do you guys have any experience on FCLK memory controller clock or Socket clock? Does increasing it make any performance improvement?
Click to expand...
Both FCLK and SoC clock can improve performance, but it's very subtle, and best tuned after finding stable core and memory clocks. Both are also heavily dependent on SoC voltage. 1300 SoC and 2100 FCLK are probably beyond what's completely stable without throwing extra SoC voltage at things.

I haven't been able to measure any advantage from increasing FclkBoostFreq.

RR.Bear said:
what else can they be хуu connected with?
Click to expand...
Keeping power within certain constraints while still maximizing performance.
 
Save
Reply Quote Rep+ Rep+
#1,377 ·
I picked up a used MSI Gaming X Trio 6800XT after so many years of being stuck with a GTX 1080 which felt like just barely enough for 1440p. Excited to get to overclock stuff again since pascal overclocking was so amazingly boring. Really disappointed I can't just RBE my overclock settings after I have them dialed in on RDNA2. Oh well, at least I don't have to deal with afterburner adding +100 to my GTX 1080 silently after being installed and requiring like -70 to get back to stable 2ghz boost clocks until you uninstall it.

The 6800XT was overheating and sold by the previous owner who had done a repaste(kryonaut I assume) and replaced the thermal pads in the past which didn't fix the problem. The memory thermal pads were less than half the size of stock and the ones on the pci-e slot side were not fully making contact, the cooler would also lift away from the vrm chokes and the card would hit 111c on the hotspot in some games under stock settings. So it got taken apart fairly quickly. I repasted it with MX-6 and replaced all of the pads with assorted thermalright odyssey ii pads where afterwards it would stay in the mid 100's at 265w/stock voltage but going from "way too hot" to "better but still hot" wasn't good enough. Now I am currently using a 20x30 Honeywell PTM7950 pad which I figured might just give this poor die a shot under the cavernous direct heatpipe coldplate that the MSI cards have. At the worst it would be something interesting to try out.

The PTM7950 seems to be working fairly well, under higher loads at 265w it's usually around 66-71c on the gpu and 88-93c on the hot spot. 2500/2600 at -160mV. I seem to average about 100mhz higher running under linux though, the performance seems about right and it will frequency cap on nearly everything. Under really extreme workloads it can touch 101c or so before dipping back down. Adrenaline seems to tie the fan curve to the hotspot where Corectrl follows the gpu temp which is a little strange. Linux RDNA2 memory overclocking is also apparently an absolute chore bordering on straight broken, maybe I'll poke that after some more fine wine happens or I'm really bored.

I'm really thinking this thing needs more pressure on the die but this cooler kinda frightens me, the springs don't even sit flat and one side always seems to stick up higher above the backplate. I'd happily spend a few hours lapping the cooler but I'd honestly be scared about going straight through the heatpipes or further ruining the piss poor mounting pressure that MSI has designed. Here's the front side and the back side of my standard mx-6 paste mount using 0.5mm washers under the spring screws, you can visibly see one of the standoffs at different heights. There's no x-bracket behind the die either as MSI didn't use one. When I used the PTM pad, directly after install before any heat cycles you could visibly see the mount looked more even but there was a slightly larger gap. Here is Front side and back side with the PTM pad. I assume it has evened out some as the temps have came down further and are far more in line with reported temps for a 6800xt.

It's pretty hard to get it to power cap in most games, hogwarts seems to do it but it's hard as 2600mhz seems to often not be enough to hit 265w.



Here's some timespy results, hwinfo/adrenaline are grabbed from the very end of graphics test 2 in these.




Does anyone have any input on how my mount looked, unfortunately I'm lacking on the before images aside from the one linked at the top of the post. I'm also not overly familiar with thermal pad material and what good contact should look like. I dropped 10-15c on the memory, they look like they had decent pressure on my MX-6 paste mount which was installed for about 2-3 weeks. The pads aren't the same size as stock but supposedly they're on the harder side for pads. I'm the sides are 1.6mm pads and the two that touch the heatpipes are 1mm. The PTM7950 wasn't particularly cheap and I highly doubt it will survive cooler removal so I'd like to avoid that until it needs to happen. I could probably stack the .5mm washers or switch to some wider 1mm washers still.



I'm not really sure if I'm missing anything here with these cards, it makes decent numbers and stays in the realm of what some other 6800xt's seem to do. It is a card with this exact known problem of it having a low end cooler, I'm not sure if I should be expecting more than I'm already getting out of it.

 
Save
Reply Quote Rep+ Rep+
#1,378 ·
Slayer3032 said:
I picked up a used MSI Gaming X Trio 6800XT after so many years of being stuck with a GTX 1080 which felt like just barely enough for 1440p. Excited to get to overclock stuff again since pascal overclocking was so amazingly boring. Really disappointed I can't just RBE my overclock settings after I have them dialed in on RDNA2. Oh well, at least I don't have to deal with afterburner adding +100 to my GTX 1080 silently after being installed and requiring like -70 to get back to stable 2ghz boost clocks until you uninstall it.

The 6800XT was overheating and sold by the previous owner who had done a repaste(kryonaut I assume) and replaced the thermal pads in the past which didn't fix the problem. The memory thermal pads were less than half the size of stock and the ones on the pci-e slot side were not fully making contact, the cooler would also lift away from the vrm chokes and the card would hit 111c on the hotspot in some games under stock settings. So it got taken apart fairly quickly. I repasted it with MX-6 and replaced all of the pads with assorted thermalright odyssey ii pads where afterwards it would stay in the mid 100's at 265w/stock voltage but going from "way too hot" to "better but still hot" wasn't good enough. Now I am currently using a 20x30 Honeywell PTM7950 pad which I figured might just give this poor die a shot under the cavernous direct heatpipe coldplate that the MSI cards have. At the worst it would be something interesting to try out.

The PTM7950 seems to be working fairly well, under higher loads at 265w it's usually around 66-71c on the gpu and 88-93c on the hot spot. 2500/2600 at -160mV. I seem to average about 100mhz higher running under linux though, the performance seems about right and it will frequency cap on nearly everything. Under really extreme workloads it can touch 101c or so before dipping back down. Adrenaline seems to tie the fan curve to the hotspot where Corectrl follows the gpu temp which is a little strange. Linux RDNA2 memory overclocking is also apparently an absolute chore bordering on straight broken, maybe I'll poke that after some more fine wine happens or I'm really bored.

I'm really thinking this thing needs more pressure on the die but this cooler kinda frightens me, the springs don't even sit flat and one side always seems to stick up higher above the backplate. I'd happily spend a few hours lapping the cooler but I'd honestly be scared about going straight through the heatpipes or further ruining the piss poor mounting pressure that MSI has designed. Here's the front side and the back side of my standard mx-6 paste mount using 0.5mm washers under the spring screws, you can visibly see one of the standoffs at different heights. There's no x-bracket behind the die either as MSI didn't use one. When I used the PTM pad, directly after install before any heat cycles you could visibly see the mount looked more even but there was a slightly larger gap. Here is Front side and back side with the PTM pad. I assume it has evened out some as the temps have came down further and are far more in line with reported temps for a 6800xt.

It's pretty hard to get it to power cap in most games, hogwarts seems to do it but it's hard as 2600mhz seems to often not be enough to hit 265w.



Here's some timespy results, hwinfo/adrenaline are grabbed from the very end of graphics test 2 in these.




Does anyone have any input on how my mount looked, unfortunately I'm lacking on the before images aside from the one linked at the top of the post. I'm also not overly familiar with thermal pad material and what good contact should look like. I dropped 10-15c on the memory, they look like they had decent pressure on my MX-6 paste mount which was installed for about 2-3 weeks. The pads aren't the same size as stock but supposedly they're on the harder side for pads. I'm the sides are 1.6mm pads and the two that touch the heatpipes are 1mm. The PTM7950 wasn't particularly cheap and I highly doubt it will survive cooler removal so I'd like to avoid that until it needs to happen. I could probably stack the .5mm washers or switch to some wider 1mm washers still.



I'm not really sure if I'm missing anything here with these cards, it makes decent numbers and stays in the realm of what some other 6800xt's seem to do. It is a card with this exact known problem of it having a low end cooler, I'm not sure if I should be expecting more than I'm already getting out of it.

Click to expand...
Overheating rx 6800xt HotSpot - super common issue. I know that it's a bit frightening to do, but putting more pressure on the die really does help it. I switched to custom AIO water cooling because above ~280Wats power draw, hotspot began to climb up to 110C. On the first try I was super careful, super low pressure and temps were very high, same or worse then on air, then I slowly added more pressure two times by tightening the screws, and now max gpu temp is 48C, and Hotspot 90C. (~370Wats power draw, unlocked via MPT) Average usage (~320Wats PD) ~45C edge/ ~78C Hotspot.
 
#1,380 ·
Slayer3032 said:
I picked up a used MSI Gaming X Trio 6800XT after so many years of being stuck with a GTX 1080 which felt like just barely enough for 1440p. Excited to get to overclock stuff again since pascal overclocking was so amazingly boring. Really disappointed I can't just RBE my overclock settings after I have them dialed in on RDNA2. Oh well, at least I don't have to deal with afterburner adding +100 to my GTX 1080 silently after being installed and requiring like -70 to get back to stable 2ghz boost clocks until you uninstall it.

The 6800XT was overheating and sold by the previous owner who had done a repaste(kryonaut I assume) and replaced the thermal pads in the past which didn't fix the problem. The memory thermal pads were less than half the size of stock and the ones on the pci-e slot side were not fully making contact, the cooler would also lift away from the vrm chokes and the card would hit 111c on the hotspot in some games under stock settings. So it got taken apart fairly quickly. I repasted it with MX-6 and replaced all of the pads with assorted thermalright odyssey ii pads where afterwards it would stay in the mid 100's at 265w/stock voltage but going from "way too hot" to "better but still hot" wasn't good enough. Now I am currently using a 20x30 Honeywell PTM7950 pad which I figured might just give this poor die a shot under the cavernous direct heatpipe coldplate that the MSI cards have. At the worst it would be something interesting to try out.

The PTM7950 seems to be working fairly well, under higher loads at 265w it's usually around 66-71c on the gpu and 88-93c on the hot spot. 2500/2600 at -160mV. I seem to average about 100mhz higher running under linux though, the performance seems about right and it will frequency cap on nearly everything. Under really extreme workloads it can touch 101c or so before dipping back down. Adrenaline seems to tie the fan curve to the hotspot where Corectrl follows the gpu temp which is a little strange. Linux RDNA2 memory overclocking is also apparently an absolute chore bordering on straight broken, maybe I'll poke that after some more fine wine happens or I'm really bored.

I'm really thinking this thing needs more pressure on the die but this cooler kinda frightens me, the springs don't even sit flat and one side always seems to stick up higher above the backplate. I'd happily spend a few hours lapping the cooler but I'd honestly be scared about going straight through the heatpipes or further ruining the piss poor mounting pressure that MSI has designed. Here's the front side and the back side of my standard mx-6 paste mount using 0.5mm washers under the spring screws, you can visibly see one of the standoffs at different heights. There's no x-bracket behind the die either as MSI didn't use one. When I used the PTM pad, directly after install before any heat cycles you could visibly see the mount looked more even but there was a slightly larger gap. Here is Front side and back side with the PTM pad. I assume it has evened out some as the temps have came down further and are far more in line with reported temps for a 6800xt.

It's pretty hard to get it to power cap in most games, hogwarts seems to do it but it's hard as 2600mhz seems to often not be enough to hit 265w.



Here's some timespy results, hwinfo/adrenaline are grabbed from the very end of graphics test 2 in these.




Does anyone have any input on how my mount looked, unfortunately I'm lacking on the before images aside from the one linked at the top of the post. I'm also not overly familiar with thermal pad material and what good contact should look like. I dropped 10-15c on the memory, they look like they had decent pressure on my MX-6 paste mount which was installed for about 2-3 weeks. The pads aren't the same size as stock but supposedly they're on the harder side for pads. I'm the sides are 1.6mm pads and the two that touch the heatpipes are 1mm. The PTM7950 wasn't particularly cheap and I highly doubt it will survive cooler removal so I'd like to avoid that until it needs to happen. I could probably stack the .5mm washers or switch to some wider 1mm washers still.



I'm not really sure if I'm missing anything here with these cards, it makes decent numbers and stays in the realm of what some other 6800xt's seem to do. It is a card with this exact known problem of it having a low end cooler, I'm not sure if I should be expecting more than I'm already getting out of it.

Click to expand...
Slayer3032 said:
I picked up a used MSI Gaming X Trio 6800XT after so many years of being stuck with a GTX 1080 which felt like just barely enough for 1440p. Excited to get to overclock stuff again since pascal overclocking was so amazingly boring. Really disappointed I can't just RBE my overclock settings after I have them dialed in on RDNA2. Oh well, at least I don't have to deal with afterburner adding +100 to my GTX 1080 silently after being installed and requiring like -70 to get back to stable 2ghz boost clocks until you uninstall it.

The 6800XT was overheating and sold by the previous owner who had done a repaste(kryonaut I assume) and replaced the thermal pads in the past which didn't fix the problem. The memory thermal pads were less than half the size of stock and the ones on the pci-e slot side were not fully making contact, the cooler would also lift away from the vrm chokes and the card would hit 111c on the hotspot in some games under stock settings. So it got taken apart fairly quickly. I repasted it with MX-6 and replaced all of the pads with assorted thermalright odyssey ii pads where afterwards it would stay in the mid 100's at 265w/stock voltage but going from "way too hot" to "better but still hot" wasn't good enough. Now I am currently using a 20x30 Honeywell PTM7950 pad which I figured might just give this poor die a shot under the cavernous direct heatpipe coldplate that the MSI cards have. At the worst it would be something interesting to try out.

The PTM7950 seems to be working fairly well, under higher loads at 265w it's usually around 66-71c on the gpu and 88-93c on the hot spot. 2500/2600 at -160mV. I seem to average about 100mhz higher running under linux though, the performance seems about right and it will frequency cap on nearly everything. Under really extreme workloads it can touch 101c or so before dipping back down. Adrenaline seems to tie the fan curve to the hotspot where Corectrl follows the gpu temp which is a little strange. Linux RDNA2 memory overclocking is also apparently an absolute chore bordering on straight broken, maybe I'll poke that after some more fine wine happens or I'm really bored.

I'm really thinking this thing needs more pressure on the die but this cooler kinda frightens me, the springs don't even sit flat and one side always seems to stick up higher above the backplate. I'd happily spend a few hours lapping the cooler but I'd honestly be scared about going straight through the heatpipes or further ruining the piss poor mounting pressure that MSI has designed. Here's the front side and the back side of my standard mx-6 paste mount using 0.5mm washers under the spring screws, you can visibly see one of the standoffs at different heights. There's no x-bracket behind the die either as MSI didn't use one. When I used the PTM pad, directly after install before any heat cycles you could visibly see the mount looked more even but there was a slightly larger gap. Here is Front side and back side with the PTM pad. I assume it has evened out some as the temps have came down further and are far more in line with reported temps for a 6800xt.

It's pretty hard to get it to power cap in most games, hogwarts seems to do it but it's hard as 2600mhz seems to often not be enough to hit 265w.



Here's some timespy results, hwinfo/adrenaline are grabbed from the very end of graphics test 2 in these.




Does anyone have any input on how my mount looked, unfortunately I'm lacking on the before images aside from the one linked at the top of the post. I'm also not overly familiar with thermal pad material and what good contact should look like. I dropped 10-15c on the memory, they look like they had decent pressure on my MX-6 paste mount which was installed for about 2-3 weeks. The pads aren't the same size as stock but supposedly they're on the harder side for pads. I'm the sides are 1.6mm pads and the two that touch the heatpipes are 1mm. The PTM7950 wasn't particularly cheap and I highly doubt it will survive cooler removal so I'd like to avoid that until it needs to happen. I could probably stack the .5mm washers or switch to some wider 1mm washers still.



I'm not really sure if I'm missing anything here with these cards, it makes decent numbers and stays in the realm of what some other 6800xt's seem to do. It is a card with this exact known problem of it having a low end cooler, I'm not sure if I should be expecting more than I'm already getting out of it.

Click to expand...
I bought the same MSI 6800XT Gaming X Trio card at the height of the shortages for close to MSRP (otherwise a no go) when I RMA'd my reference MSI 5700XT not knowing if I would see it again.

When playing AAA games, the card will stay in the 60's C and Hot Spot will hit 92-93C. Usually, the span between the two can reach 35C which is crap.

I have good thermal paste, even liquid metal. I have expensive thermal pads from water cooling. Did I crack the card open? Nah. Not wasting my time or putting another dime on the card. Many did just that and the ordeal backfired and made matters worse.

What I did do is in MSI Afterburner (most guides use Radeon software) put Power Limit to max, overclocked the Memory by 112MHz, undervolted to 1130v and watched those temps/watts fall without any loss to performance!
 
  • Rep+
Reactions: Slayer3032
Save
Reply Quote Rep+ Rep+
1,361 - 1,380 of 1,702 Posts
About this Discussion
1.7K Replies
218 Participants
Last post:
Travis Scott
Overclock.net
posts
28M
members
562K
Since
2004
The premier forum for overclocking experts and enthusiasts. Discuss hardware optimization, custom builds, benchmarking, cooling solutions, and pushing the boundaries of computing performance. From beginner guides to extreme overclocking, join our technical community to master system tuning.
Full Forum Listing
Explore Our Forums
Intel CPUs Hardware News NVIDIA PC Gaming Water Cooling