Thanks!
Seems the Steam binary was built on Dec 10th, scratch buffer commit went in on Dec 22nd. From the discussions on forums and github, it looks like the new commits alleviated the issues for some -- not for others (unclear), yet I'll try compiling a build with them and see if I get any further.
Problem is I have blocked nothing I am aware off
Returned to 20.12.1 and all is well again...
What happens if you uninstall the video card driver > restart into safe mode > use DDU to clean old traces of drivers > restart > install new driver > restart
All up and running here, although haven't tested any games yet.
https://i.imgur.com/sqNf139.png
DayZ and Escape From Tarkov are only using a small amount of the GPU. Sometimes utilization is 40%. Never had this issue with a 2080 super before in this same rig, using the RX 6800 now. 3700x cpu. I used ddu to remove nvidia drivers. I've fiddled with so many settings and cant get it to use the whole gpu in the game. CPU usage sits between 10 and 30% usage. In DayZ looking one direction the gpu will be at 99% and I turn and it will drop to 40%.
In Tarkov at the beginning of the match the gpu will be at 100% and running perfect frames, like 30 seconds into a raid loading usage drops down and so do the frames. Games like warzone and fortnite run perfectly fine 99% usage but those are not the types of games I like to play.
Asus x570 tuf wifi with latest bios and chipset drivers.
Looking for any advice I can get, thank you.
Warning!
For me a troublesome install and on top caught in a safe mode loop - an early restore point brought it back to normal, phew!
Cannot confirm....
All went fine, installed on top of older drivers.
Oh and I wanted to report that the issues with not only the Samsung G9 are fixed but also the C49RG90.... Seem these two monitors share a little bit of similarities.
Nah, I'll be patient and await fixes -- the burst in testing effort was due to the change log stating support, just like when Vulkan RT support was added. I would like to some-day be able to play through the game though, being that raytracing is an advertised feature of the hardware.
Effectively this is just bug-reporting, not protesting that raytracing is non-functional at this time.
The game isn't that lengthy and if we're talking the earlier base maps here that's less than an hour into the game something that's easily repeatable so it's odd that AMD listed it when their own testing should have been hitting the same problem plus the Vanguard testers too assuming AMD has something of a program here at least so if Q2 RTX support is going to be a thing at least a part of the V-G team is going to be testing that game across a range of hardware.
EDIT: Not that Vanguard is bad or anything loosely testing and doing a bit of whatever getting metrics and other telemetry data, crash reports and such info helps but it's a bit random and who knows how these issues get filed and prioritized then.
Think that's how some issues like the Samsung G9 one was identified it's tough to have all the hardware variants and such at hand so it helps with that.
Although for upcoming titles since the Vanguard team aren't AMD testers they can't have access to unreleased software and additional NDA's and what not that goes into that either.
(Doesn't quite apply to Q2 RTX here but most early support of games would be internal stuff.)
The game isn't that lengthy and if we're talking the earlier base maps here that's less than an hour into the game something that's easily repeatable so it's odd that AMD listed it when their own testing should have been hitting the same problem plus the Vanguard testers too assuming AMD has something of a program here at least so if Q2 RTX support is going to be a thing at least a part of the V-G team is going to be testing that game across a range of hardware.
EDIT: Not that Vanguard is bad or anything loosely testing and doing a bit of whatever getting metrics and other telemetry data, crash reports and such info helps but it's a bit random and who knows how these issues get filed and prioritized then.
Think that's how some issues like the Samsung G9 one was identified it's tough to have all the hardware variants and such at hand so it helps with that.
Although for upcoming titles since the Vanguard team aren't AMD testers they can't have access to unreleased software and additional NDA's and what not that goes into that either.
(Doesn't quite apply to Q2 RTX here but most early support of games would be internal stuff.)
Just did a self build of Q2RTX from the git master branch through Dec 23rd for further testing, which I have uploaded here. [for my fellow AMD Gurus]
SHA1 Size Name
---------------------------------------- ------------- ------------
78289422D11206E1474174342B1163CB8010DA02 1001067851 build_0a16c73_Dec23_2020.zip
This one contains the scratch buffer adjustment, the pull from 'amd-fixes' on Dec 19th, etc. q2rtx_media and other required assets are already in there, just need the game-files. The Steam build is from Dec 10th, as is the build off NVidia's website, so this is slightly newer.
-- Sadly, even with the latest commits it still crashes at exactly the same spot.
EDIT: 21.1.1 -- still crashes 100% of the time. 20.12.1 -- passes combined with the commits.
Though thankfully we had SpajdrEX (Vanguard) confirm that AMD is aware .. yet yes, it's odd that they would sign off the game as supported {patch notes wise} when aware of such an issue.
Something similar happened to me yesterday; after giving the 21.1.1 a try, I went back to 20.4.2. Then I went to WU and saw, among other Microsoft updates, an AMD driver update (never seen that before). After installing them, Win10 asked to reboot, so I did and upon logging in I got a Blue Screen with a "reference by pointer" error. Rebooted, Win10 did its thing to self repair and I was able to log back in, but when I clicked the Radeon SW icon it gave me an error about a mismatch between driver and software; at that point I did a clean install of the 21.1.1 and everything now is back to normal.
can someone please report if the memory downclocking issue is fixed for 6800/6900 users with 144hz monitors yet? my mem clocks are stuck at full speed at idle while @ 144hz.
Spoke too soon on my assumption that building Quake II RTX had no impact on the crash. Check this out-
Driver version 20.12.1 {tried downgrading again}, combined with the latest github commits {build used posted up higher in this thread}. Consistently pass the crash in the COMM CENTER. Yahoo!
-- Though, something is going on in 21.1.1 for sure, since there's graphical artifacts {missing polygons in reflections, and other weird stuff}, and it does still crash in the same spot. If I run build 1.4 {without the latest commits}, as distributed by Steam, I will also crash the moment that I kill that enemy as seen on the left {on any driver, of course}.
can someone please report if the memory downclocking issue is fixed for 6800/6900 users with 144hz monitors yet? my mem clocks are stuck at full speed at idle while @ 144hz.
can someone please report if the memory downclocking issue is fixed for 6800/6900 users with 144hz monitors yet? my mem clocks are stuck at full speed at idle while @ 144hz.
I have the same issue with my 75Hz monitor and it's not fixed! For now I use the workaround with CRU to set "the proper" timings that AMD drivers are ok with. Look at my comment HERE
Can't install these. Installer crashes to black screen and mouse stops working did full ddu and reg clean then installed only to find cold war would not load tried all sorts to fix and give up rolled back to 20.11.1 also these drivers did same as others windows yet again will not shut down ,
UPD.
Yes it is still with us 😉
(Note: Is somehow better now, most of the time (hammer in hand) You can see what to build and where).
For now is the fastest driver for Vega.
No other bugs to report.
Tested so far:
BFV & Forza H4
Undying:
What is the vega bfv bug?
When building You sometimes can't see what can be build o_O
Also includes AP_Mines for medic, can't see sometimes where to place it.
can someone please report if the memory downclocking issue is fixed for 6800/6900 users with 144hz monitors yet? my mem clocks are stuck at full speed at idle while @ 144hz.
Nope. I've just tried to switch from my 142Hz (custom resolution with CVT reduced blank created with AMD Drivers) to 144Hz (original sandard resolution) and GPU PWR went up from 8W to 30W.
This is the (almost only) and biggest issue I have with AMD drivers.
This driver has a problem with "BLACK MESA - Definitive Edition".
I play the game with vsync on. With previous 20.12.1 driver I had 75 fps constantly (because my monitor is 75Hz and my 6800XT is very fast). Now with 21.1.1 I have fps varying - from very low like 3, 45, 50 to 75 max at some moments. It's not normally working vsync! With vsinc off I have 120-200 fps (constantly over 100).
Workaround is to set vsync off in the game and to use "Advanced Sync" and FRTC to 75 fps in drivers. The problem is that I can't set FRTC for one game only but it's a global setting.
I have tried other DX9 games - they have no problem with vsync (including HL2).
Spoke too soon on my assumption that building Quake II RTX had no impact on the crash. Check this out- Driver version 20.12.1 {tried downgrading again}, combined with the latest github commits {build used posted up higher in this thread}. Consistently pass the crash in the COMM CENTER. Yahoo! -- Though, something is going on in 21.1.1 for sure, since there's graphical artifacts {missing polygons in reflections, and other weird stuff}, and it does still crash in the same spot. If I run build 1.4 {without the latest commits}, as distributed by Steam, I will also crash the moment that I kill that enemy as seen on the left {on any driver, of course}.