The Radeon VII Owners Thread

Jamin280672 · 5 Jul 2020 at 15:19

Illuminist said:
You're very welcome, you are the second person in this thread with an r7 from me lol

LOL did you sell your new one ? why did you do that, I thought you wanted to go back to the R7 ?

Illuminist · 5 Jul 2020 at 16:18

Jamin280672 said:
LOL did you sell your new one ? why did you do that, I thought you wanted to go back to the R7 ?

Lol I'm saving for something sexy and most likely stupidly expensivein the next few months rocking a vega now

phatboy · 5 Jul 2020 at 19:17

Mines absolutely flying!

Have used the settings that @Illuminist gave me and running undervolted with the memory bumped to 1050 and that's all I really need!

I have a freesync 75 32" screen and everything plays happily at that level with all the pretties cranked up.

Always had a soft spot for these cards and it hasn't disappointed

Dave2150 · 5 Jul 2020 at 20:01

Illuminist said:
Lol I'm saving for something sexy and most likely stupidly expensivein the next few months rocking a vega now

Keeping my Radeon VII until I get a next gen HDMI 2.1 card later this year. Hoping it's big Navi, but this time I'm only getting it if it's within 5-10% of the NV priced equivalent, with NO driver issues reported on release. Navi's driver issues, coupled with the amount of time AMD took/is taking to fix them, are not a good sign.

I've had my Radeon VII since release day (Feb 8th 2019). Think that's the longest I've kept a single card in recent memory! Fair play it's been a very good card.

Mechorpheus · 7 Jul 2020 at 12:48

Hey guys, I recently joined the R7 club myself and I have a couple of weird driver questions.

One, I've been playing with the manual OC settings in the driver and for some reason if I touch the core frequency (set it to anything other than 1801), it'll apply fine if done while running a game, but if I close and restart that game it drops back to 1720(ish) core and 965mV regardless what I set the core clock to. If I leave the core clock alone, custom mV values are set as expected, and custom values for memory OC also work as expected. I see this behaviour when trying to use Afterburner too. Anyone have any idea why this happens or how I might fix it?

Also, and I posted this in a couple of places but no one replied, I'm having issues getting games launched at 1440p to run at high refresh on my monitor. It's not freesync, but 120hz works fine if I force it on the desktop, and some games which allow specifications of the refresh rate (e.g. DiRT Rally 2.0) work fine too, but games which don't have such an option (e.g. DOOM Eternal) refuse to load at 120Hz. The Nvidia driver had an option to set 'Highest Available' refresh rate, but there doesn't seem to be any equivalent in the AMD Driver.

Thanks in advance!

phatboy · 10 Jul 2020 at 11:22

Finally had a chance to play about with settings a bit more

Based on people's experiences does this look on par? Core is undervolted to 980mv and bumped the memory to 1050

https://www.3dmark.com/fs/23024230

phatboy · 10 Jul 2020 at 11:24

If I set VSYNC on games and ram everything to max at 1440p the card is running at like 50 degrees and 70 at the junction. Unbelievable, it's like it's not even having to work for it

Gerard · 10 Jul 2020 at 22:21

phatboy said:
If I set VSYNC on games and ram everything to max at 1440p the card is running at like 50 degrees and 70 at the junction. Unbelievable, it's like it's not even having to work for it

About the easiest way to keep it quiet, though depending on the game it can ramp up a bit in certain scenarios. BF5 loading screen for instance or when using the sniper scope zoom can make it ramp up more.

phatboy · 11 Jul 2020 at 09:05

yeah you aren't wrong, F1 2019 has been the biggest offender for me. The game at 75fps freesync is stting card around 60/70 degrees but the menus and the little interview spots and driving it up to 80 and 100 on junction, how on earth they code that so badly!

Chud_Jenny · 12 Jul 2020 at 16:11

Mechorpheus said:
Hey guys, I recently joined the R7 club myself and I have a couple of weird driver questions.

One, I've been playing with the manual OC settings in the driver and for some reason if I touch the core frequency (set it to anything other than 1801), it'll apply fine if done while running a game, but if I close and restart that game it drops back to 1720(ish) core and 965mV regardless what I set the core clock to. If I leave the core clock alone, custom mV values are set as expected, and custom values for memory OC also work as expected. I see this behaviour when trying to use Afterburner too. Anyone have any idea why this happens or how I might fix it?

Also, and I posted this in a couple of places but no one replied, I'm having issues getting games launched at 1440p to run at high refresh on my monitor. It's not freesync, but 120hz works fine if I force it on the desktop, and some games which allow specifications of the refresh rate (e.g. DiRT Rally 2.0) work fine too, but games which don't have such an option (e.g. DOOM Eternal) refuse to load at 120Hz. The Nvidia driver had an option to set 'Highest Available' refresh rate, but there doesn't seem to be any equivalent in the AMD Driver.

Thanks in advance!

I've written several posts on core OC. The bad news is it was seemingly killed off never to work again after 19.7.5 driver!
https://www.amd.com/en/support/kb/release-notes/rn-rad-win-19-7-5
Might want to give that a try if you want to run some 3DMark.
I run 1801@950mv, 1140MHz HBM, on an EK block. That gives me very low temps. After my loop warms up, 32c edge and 65c (max) junction. Lower voltage really helps to keep the junction temp very low.

Mechorpheus · 12 Jul 2020 at 22:07

That's a shame about the OC in the driver. It's very strange in that it does actually apply it if I make changes while running something (e.g. Unigine Heaven), and I do see framerate incrases so it's not just indicating a higher speed. I'll grab that driver and give it a shot though, thanks for the tip!

I've been running mine stock core clock (1801MHz but w/ 1200MHz memory), can UV to 1006mV and it's totally stable (can run as low as 980mV but I see the occasional artifact and figured it was better to up it as AMD driver crashes seem much more severe to windows and NVidia onces did, often just BSODing). I guess I lost the silicon lottery a bit though as the stock voltage of my card is 1106mV which seems on the high side from what I've been reading. Even at 1006mV though unless I run the fans at 100% it'll hit the 110'c throttle point, and sounds like it's about to take off! I tried the washer mod and it only made things worse (hot-spot temp was more than 35'c above edge temp, which seemed really bad). I may look at water-cooling but I havn't ever done custom loops, and while I've done AIO GPU watercooling before (had great success on a 980 ti back in the day) It seems like a major faff on the Radeon VII as you have to drill new holes on the Kraken bracket to get it to fit. I've ordered some sand-paper to have a go at lapping the cooler which I've read some people have had success with, maybe go water later if it's still a furnace.

Chud_Jenny · 13 Jul 2020 at 08:39

Mechorpheus said:
That's a shame about the OC in the driver. It's very strange in that it does actually apply it if I make changes while running something (e.g. Unigine Heaven), and I do see framerate incrases so it's not just indicating a higher speed. I'll grab that driver and give it a shot though, thanks for the tip!

I've been running mine stock core clock (1801MHz but w/ 1200MHz memory), can UV to 1006mV and it's totally stable (can run as low as 980mV but I see the occasional artifact and figured it was better to up it as AMD driver crashes seem much more severe to windows and NVidia onces did, often just BSODing). I guess I lost the silicon lottery a bit though as the stock voltage of my card is 1106mV which seems on the high side from what I've been reading. Even at 1006mV though unless I run the fans at 100% it'll hit the 110'c throttle point, and sounds like it's about to take off! I tried the washer mod and it only made things worse (hot-spot temp was more than 35'c above edge temp, which seemed really bad). I may look at water-cooling but I havn't ever done custom loops, and while I've done AIO GPU watercooling before (had great success on a 980 ti back in the day) It seems like a major faff on the Radeon VII as you have to drill new holes on the Kraken bracket to get it to fit. I've ordered some sand-paper to have a go at lapping the cooler which I've read some people have had success with, maybe go water later if it's still a furnace.

https://www.ekwb.com/shop/ek-quantum-vector-radeon-vii-rgb-nickel-acetal
They are out of stock buy quoting "Expected date of availability will be added ASAP" Might be worth emailing them to ask if they plan a production run as they did on the Vega 56/64 blocks recently. OCUK no longer have stock.

Dave2150 · 13 Jul 2020 at 10:20

Mechorpheus said:
That's a shame about the OC in the driver. It's very strange in that it does actually apply it if I make changes while running something (e.g. Unigine Heaven), and I do see framerate incrases so it's not just indicating a higher speed. I'll grab that driver and give it a shot though, thanks for the tip!

I've been running mine stock core clock (1801MHz but w/ 1200MHz memory), can UV to 1006mV and it's totally stable (can run as low as 980mV but I see the occasional artifact and figured it was better to up it as AMD driver crashes seem much more severe to windows and NVidia onces did, often just BSODing). I guess I lost the silicon lottery a bit though as the stock voltage of my card is 1106mV which seems on the high side from what I've been reading. Even at 1006mV though unless I run the fans at 100% it'll hit the 110'c throttle point, and sounds like it's about to take off! I tried the washer mod and it only made things worse (hot-spot temp was more than 35'c above edge temp, which seemed really bad). I may look at water-cooling but I havn't ever done custom loops, and while I've done AIO GPU watercooling before (had great success on a 980 ti back in the day) It seems like a major faff on the Radeon VII as you have to drill new holes on the Kraken bracket to get it to fit. I've ordered some sand-paper to have a go at lapping the cooler which I've read some people have had success with, maybe go water later if it's still a furnace.

Personally, I just run mine day to day with the fans limited to 44% (this is less than 2000 RPM) and the GPU undervolted to 969mv. The noise at full load is much quieter, though of course I lose some performance vs running the fan faster. I feel I gain enough from the undervolt and HBM2 overclock (running at 1200Mhz) to get away with it.

My junction temp is regularly over 110DC, though this is normal and expected. My normal GPU is temp is low 80's, that's the nvidia "equivalent" temperature to monitor. My card's been running like this since February 2019 and hasn't missed a beat, they are designed for temperatures like this and there's no need to worry

Gerard · 26 Jul 2020 at 18:15

Anyone else noticed that there seems to be a number of "non functional" radeon vii's on ebay? They all seem to have the same issue with no video output and the device manager showing a code43 with an exclamation mark beside it (via booting with another gpu obviously). Others just point blank not displaying a video output or being shown in device manager, but fans and logo illuminates.

This a common issue? Over the last couple of months I've seen quite a few listings with this, there's currently 2 on it at the moment.

Chud_Jenny · 26 Jul 2020 at 18:22

Gerard said:
Anyone else noticed that there seems to be a number of "non functional" radeon vii's on ebay? They all seem to have the same issue with no video output and the device manager showing a code43 with an exclamation mark beside it (via booting with another gpu obviously). Others just point blank not displaying a video output or being shown in device manager, but fans and logo illuminates.

This a common issue? Over the last couple of months I've seen quite a few listings with this, there's currently 2 on it at the moment.

Might be worth sending them a message and see what the background of the cards are. Power mods over time could be the issue?

Gerard · 26 Jul 2020 at 19:31

Chud_Jenny said:
Might be worth sending them a message and see what the background of the cards are. Power mods over time could be the issue?

Not sure, some of them in a few listings still have had the warranty sticker intact so hard to say. Code-43 ive seen fixed with a bios flash.

Chud_Jenny · 26 Jul 2020 at 19:35

Gerard said:
Not sure, some of them in a few listings still have had the warranty sticker intact so hard to say. Code-43 ive seen fixed with a bios flash.

Ye, could have been from the first batch without the uefi bios.

Gerard · 26 Jul 2020 at 22:08

Chud_Jenny said:
Ye, could have been from the first batch without the uefi bios.

Possible i suppose, think that's what people are gambling on when buying them, that it's a simple fix.

Chud_Jenny · 26 Jul 2020 at 22:27

Gerard said:
Possible i suppose, think that's what people are gambling on when buying them, that it's a simple fix.

Just had a look at sold listings. Several this month oddly. One quoting no display in windows. Sounds like it's down to a list of things all together to cause this. Possible outdated bios, outdated chipset drivers for AMD platforms, outdated driver, and the May 2004 windows build. Some mix of those would do the trick I'd think.

LtMatt · 28 Jul 2020 at 15:55

Just procured another Radeon VII and this one appears to be a very good sample. Have been able to under volt to -0.950mv without any obvious stability issues, memory at 120Mhz. Will try and lower voltage a bit further to see at which point I get crashes or artifacts.