• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

The Radeon RX 6900/6950 XT Owners Thread.

Soldato
Joined
8 Jan 2003
Posts
3,702
Location
Scotland
Ye, its fitted with a Bykski block.
If you go to page 87 #1721, there's a pic on my old board.

I did my brothers Asus tuf 6800xt with a Bykski block the other month too, massive temp drops for him too. 25-30c down on the massive air cooler!

Nice! I used a Bykski block in my first water cooled rig attempt last year. Good blocks for the price. I'm currently considering WC my Red Devil 6900 XT along with the rest of the rig. Just got to plan it out and consider if it'll work with a single 280 rad in a Meshlicious case.
 
Caporegime
Joined
12 Jul 2007
Posts
40,520
Location
United Kingdom
I repasted my Toxic again with Silver King Liquid Metal and I got my delta between Edge and Junction temperature down to 5-7c peak under 350-400W load. Absolutely delighted with that outcome. :cool:
Look at anbd admire those deltas between edge and junction under heavy load with the Toxic using liquid metal.
LOOK AT THEM AND UNDERSTAND THEM JACKIE WEAVER! :cool::cool::cool:
Those are deltas people dream of. Or should dream of. IMO.
 
Last edited:
Associate
Joined
21 Jun 2018
Posts
1,099
Location
Ashton
@LtMatt Just re-pasted the GPU with Kryonaut since I dont have LM at hand. The stock paste was not covering the entire die and looked like it they barely put anything in there. I'll test in a few mins.

Edit: So Watch Dogs Legion was taking it to 62-63C and over 70C on the junction before the re paste, now it was maximum running at 56C, and junction was moving between 64-66C. It seems I shaved a few degrees off.
 
Last edited:
Associate
Joined
4 Jun 2021
Posts
77
Location
Rocky Mountains USA
I will be interested to hear how you get on with your 6900xt vs the 3080Ti with the Quest 2. (two nice computer setups!!)

In fairness to Airlink it is still in the experimental stage and it is getting better with each update.

Before Airlink came out this year, you really had nothing else only ALVR or AMD Relive for the Quest. Both were and still are terrible compared to Virtual desktop. The Dev of Virtual desktop is still active. He is still bringing new features and fixes to Virtual desktop. He has setup a discord channel and is always willing to help out...

...Yeah, I was always in the AMD camp until I got into VR when it became clear that Nvidia gave a much better VR experience. AMD have mostly caught up now though with the 6000 series apart from their H.264 encoder :(


So, tonight I played Dirt Rally 2.0 and Moss VR with SteamVR on my Quest 2, with my overclocked Sapphire Nitro+ 6900XT in the Ryzen 5800X rig. I also tried Virtual Desktop to watch a YouTube video. My Sapphire runs close to 2400 Mhz GPU/2124 Mhz VRAM in Time Spy with a 22100 GPU score. My 6900XT had NO trouble doing wired VR play, and playback was pretty smooth and detailed with zero glitches. I'm using an inexpensive braided no-name USB-C to C high-speed "Oculus compatible" 16-foot cable from Amazon with no issues (90 degree plug on headset end, straight USB-C on the other, came with a C to A adapter as well). I plugged the USB-C into my front panel USB 3.2 Gen 2 port.

The PC case fans really spun up fast though, sounding like full blast during Dirt Rally 2.0; but the fans did not spin up very much with Moss in SteamVR. I always get a little motion sickness playing any racing games in VR, but it was no different than it has been with the 3080Ti in the 5950X rig. Moss played very well - smooth and detailed with no glitches either, and I didn't get any motion sickness with that game. The virtual desktop with YouTube video played flawlessly as well.

However, I also tried Airlink wireless play with these games through SteamVR, and my PC is connected via WiFi6/AX to my home network (internet download speed was 345+ Mbps when I checked). I forget which network the Oculus is set up for, either my legacy Apple TimeCapsule AC system or my Asus AiMesh. I only got 73 Mbps on my old 802.11AC Apple network when I checked on my PC just now. Racing is still unplayable with Airlink, jumpy and glitchy. But Moss VR was smooth enough to play wireless in terms of minimal lag or stutter, although the visual quality was pretty grainy and poor with wireless play. I forgot to test YouTube in virtual desktop with Airlink before shutting it down to go to bed.

Bottom line is that I could not see or feel any problems in Quest 2 VR performance via wired SteamVR on the 6900X in my 5800X rig.


PS: I forgot I had a more aggressive case fan curve set up because I was using a 3070Ti while my Sapphire went back for warranty on a broken RGB pin, and the 3070Ti needed a more aggressive fan curve than the 6900XT to stay cool.
 
Caporegime
Joined
12 Jul 2007
Posts
40,520
Location
United Kingdom
@LtMatt Just re-pasted the GPU with Kryonaut since I dont have LM at hand. The stock paste was not covering the entire die and looked like it they barely put anything in there. I'll test in a few mins.

Edit: So Watch Dogs Legion was taking it to 62-63C and over 70C on the junction before the re paste, now it was maximum running at 56C, and junction was moving between 64-66C. It seems I shaved a few degrees off.
Not bad at all. What power limit is that using and what is the card drawing (wattage wise) under heavy load?
 
Associate
Joined
5 Aug 2017
Posts
1,057
Location
Cornwall
21.12.1 drivers word of warning memory does not downclock. Filed bug report.
Scratch that the factory reset option ticked solved it.
Ah cool, just on with the Forza update 15.43GB and then I'll give that a bash. See how the clocks are on this driver vs the last 2 as it was loving 2700MHz in forza on 21.11.2 and then seemed to be happier at 2550/2650 on 21.11.3 set vs 2650/2750.
 
Caporegime
Joined
12 Jul 2007
Posts
40,520
Location
United Kingdom
21.12.1 drivers word of warning memory does not downclock. Filed bug report.
Scratch that the factory reset option ticked solved it.
No problems here.
69RVx4W.png

If I open up Edge and put YouTube on it climbs to 1400Mhz or so, but it's always done that. Hardware acceleration enabled in Edge.

EDIT - Just saw your edit. :p
 
Soldato
Joined
31 Dec 2007
Posts
13,616
Location
The TARDIS, Wakefield, UK
Just had a near panic attack this afternoon repasting with Liquid Metal. I bought myself a Corsair HX1000 PSU to give me a bit more headroom and decided at same time to LM the Ultimate (and wave my warranty good bye).
Power supply fitting went fine. Although the switch on the PSU has me a bit confused as its the 12v SINGLE or MULTIPLE setting. Single all 12v rails are free to grab as much as they can, MULTIPLE limits each to 40A max. Googled this but could find
anything conclusive or recently (2021) so left it on default.
Already had experience with LM when I delidded my 8700k. This time I was extra careful with all precautions taken. Went well it seems assembled back power on PC, no display. :eek:
Panic creeping in it tried various things nothing. Calming down took it out and disassembled. The spread on the heatsink didnt look right, not enough contact then I thought first time I
put the screws back in the wrong sequence (GPU 4 last) so could be that. Respread existing LM on both heatsink and GPU, assembled it again, made doubly sure the PCIE plugs were
in securely, panic starting to build, turned it on and..................... YES a display. Phew. Panic over.

Set my Radeon Software to 2500-2700 on core, 2120 mem, +15% power and ran Timespy. Where my hotspot would shoot to 105-106c on the 2nd test it now maxes at 89c! (GPU temp 63c down from 73c)
Tried Superposition 1080p Extreme and hotspot maxed at 89c with GPU at 69c, previously I think GPU was 79c. So far its been a result.
 
Caporegime
Joined
12 Jul 2007
Posts
40,520
Location
United Kingdom
Just had a near panic attack this afternoon repasting with Liquid Metal. I bought myself a Corsair HX1000 PSU to give me a bit more headroom and decided at same time to LM the Ultimate (and wave my warranty good bye).
Power supply fitting went fine. Although the switch on the PSU has me a bit confused as its the 12v SINGLE or MULTIPLE setting. Single all 12v rails are free to grab as much as they can, MULTIPLE limits each to 40A max. Googled this but could find
anything conclusive or recently (2021) so left it on default.
Already had experience with LM when I delidded my 8700k. This time I was extra careful with all precautions taken. Went well it seems assembled back power on PC, no display. :eek:
Panic creeping in it tried various things nothing. Calming down took it out and disassembled. The spread on the heatsink didnt look right, not enough contact then I thought first time I
put the screws back in the wrong sequence (GPU 4 last) so could be that. Respread existing LM on both heatsink and GPU, assembled it again, made doubly sure the PCIE plugs were
in securely, panic starting to build, turned it on and..................... YES a display. Phew. Panic over.

Set my Radeon Software to 2500-2700 on core, 2120 mem, +15% power and ran Timespy. Where my hotspot would shoot to 105-106c on the 2nd test it now maxes at 89c! (GPU temp 63c down from 73c)
Tried Superposition 1080p Extreme and hotspot maxed at 89c with GPU at 69c, previously I think GPU was 79c. So far its been a result.
Great result well done.

I can relate to the panic when it doesn’t boot up correctly. Had something similar myself first time I did it.
 
Associate
Joined
26 Dec 2020
Posts
156
So, Is anyone here messing around with the new voltage unlock using MPT? I am experiencing some weird behavior with it. The voltage shown in HwInfo and GPU-Z (Wattman) are not the same? I thought they were supposed to be the same if you are using MPT TEMP_DEPENDENT_VMIN to unlock voltages?

ghY9kR9.png
 
Caporegime
Joined
12 Jul 2007
Posts
40,520
Location
United Kingdom
So, Is anyone here messing around with the new voltage unlock using MPT? I am experiencing some weird behavior with it. The voltage shown in HwInfo and GPU-Z (Wattman) are not the same? I thought they were supposed to be the same if you are using MPT TEMP_DEPENDENT_VMIN to unlock voltages?

ghY9kR9.png
Are you sure you've done it correctly? Under feature control you need to enable Temp Dependent VMIN option.
 
Associate
Joined
26 Dec 2020
Posts
156
Are you sure you've done it correctly? Under feature control you need to enable Temp Dependent VMIN option.
Yes, That is the process that I have followed. I can see the voltage going up to 1.206 mv in HWinfo. Also, with default voltages, HWInfo never reports anything above 1.175 which is not the case after applying voltages and GPUz is reporting 1.262mv as well. I am puzzled ?
 
Back
Top Bottom