AMD Software: Adrenalin Edition 22.11.1 - Driver download and discussion

Discussion in 'Videocards - AMD Radeon Drivers Section' started by justdoge, Nov 15, 2022.

  1. shm0

    shm0 Member Guru

    Messages:
    142
    Likes Received:
    29
    GPU:
    6800XT
    Yes it's B Die, with active cooling, corsair fan holder + noctua fan modded onto it.
    It does 3800@ CL14 with 1.5v+ but I didn't want it to push this far :p
    Well could be ~48-52 degress, i can't remember. Also comes down which test is used.
    Could try higher trfc in such cases?
    I have the theory that a lower trfc actually hinders performance even when it's results in lower latency because the controller spends more time refreshing?
    So I just the default which always seems to be ~350ns regardless of speed.
     
    HD64G likes this.
  2. DimkaTsv

    DimkaTsv Active Member

    Messages:
    89
    Likes Received:
    22
    GPU:
    AORUS RX 6750XT 12G
    Don't mitake RFC for REFI. tRFC is time that bank spends in "out of work" state while refreshing. tREFI is wait period between 2 refreshes.
    upload_2022-11-24_16-10-13.png

    And as, unlike on Intel CPU (they have tREFI registry, that people usually set to around 65535 when overclocking RAM iirc), AMD doesn't have tREFI registry, tRFC is one of the main way to increase performance of RAM by reducing time it spends in refresh state.

    It is like this.
    456/14829 *100% = 3.075% of time spend refreshing}
    250/14829 *100% = 1.685%, which is almost 2 times less.
    And for Intel
    456/65535 *100 = 0.695%, which is even less, despite having same (quite bad) tRFC. . .

    That's why B-die matter more for AMD CPUs than to Intel CPU (they can work with Hynix or Micron bit better than AMD), as they are also more tolerant to higher RAM frequency unlike AMD systems

    P.S, i guess i should look into active cooling for my H8D 2R RAM . . . But it will be tough to do here, in Russia, as our choises are really limited in system modding. I also don't have enough mobo fan ports, so i would need a divider.
     
    Last edited: Nov 24, 2022 at 2:21 PM
    Horus-Anhur likes this.
  3. shm0

    shm0 Member Guru

    Messages:
    142
    Likes Received:
    29
    GPU:
    6800XT
    Yes, I mean trfc.
    How can a lower trfc be "less time spending refreshing" when it does more refreshing ?
    Are you sure your calculation is right?

    //edit
    Oh I see trefi is somewhat a "max average refresh" and trfc is the delay in between these refreshes?

    Also when lower refresh is meant to give better performance.
    Why are lower trfc values used at higher ram temperatures > 85? (trfc2 and trfc4)
    Because with higher temperature ram tends to lose data faster and needs more refreshing (which reduces performance?)
    So why don't use a lower trfc by default?

    And I have the feeling that lower trfc (other settings like faw,rrds,rrdl) results in more stuttering.
    I found a reddit post from some user stating the same that lower values actually resulting in worse performance.
     
    Last edited: Nov 24, 2022 at 2:36 PM
    HD64G likes this.
  4. DimkaTsv

    DimkaTsv Active Member

    Messages:
    89
    Likes Received:
    22
    GPU:
    AORUS RX 6750XT 12G
    Yes, i am sure my calculations are right.
    Too low tRFC causes data corruption as it begins refreshing not long enough to save data bits from switching.
    As if you don't charge battery long enough and phone powers off midday.
    tRFC tells controller amount of time bank should stay on recharge mode. Higher value = more time. Recharging means it gets constant current to fill up data cells with power
    tREFI tells controller amount of time it should wait before doing next recharge.

    If you reduce tRFC, it won't affect tREFI, as this is different register. So RAM will spend less time recharging overall.

    No... Because tRFC is basically like plugging phone to a wall, it actually emits heat.
    So if RAM is overheating, it is safety measure so it will create less heat. I suppose it also means that RAM heavily downclocks at this point to both reduce heat generation and gain ability to use these timings without penalties

    Also... All RAM timings are measured in nanoseconds, but all timings that are written in registry are relative to current RAM frequency
    Meaning tRFC = 456 T will be different for 3200 RAM and 3800 RAM
    1/(3200/2)*456*1000 = 285 ns
    1/(3800/2)*456*1000 = 240 ns
     
    Last edited: Nov 24, 2022 at 2:41 PM

  5. shm0

    shm0 Member Guru

    Messages:
    142
    Likes Received:
    29
    GPU:
    6800XT
    well,
    i'm not so sure about this but ty.
     
  6. Cp0

    Cp0 Member Guru

    Messages:
    112
    Likes Received:
    22
    GPU:
    AMD 6900 XT H2O
    I have triple radiator setup custom loop i don't even get errors when i let ram get as hot as 60c almost
    I just had system freeze with a cheap memory kit either my 4e slot is faulty mainboard or psu or GPU is really on its way out altho it passes 30 minute furmark 20 min vram test no errors neither 20 minute 3d test on occt has errors
     
  7. Blisk

    Blisk Member Guru

    Messages:
    136
    Likes Received:
    80
    GPU:
    RX 6900XT
    Anecdotes aren't data, but FWIW, I'm using a 6900XT and it's been fine.
    I have a friend with a 6800 and another with a 6800XT, each of them have 2+ displays (mismatched configs) on W10 and they've all been fine with the latest drivers.

    Kinda scary to see how config dependant stuff like this can be though.

    The WHQL status is shown for each release now if you scroll down on the driver release notes page.

    I would recommend Karhu (paid software for Windows) or Memtester (FOSS linux package) to verify your timings within your OS. If you want to rule out and actual hardware fault with your modules, a bootable tester like memtest86+ (FOSS image, updated very recently for newer platform support) or Passmark's Memtest86 would be a good shout. I'd run first at JEDEC, and then at XMP (or custom timings if you're so inclined).

    Testing time will depend on your thread count and memory allocation. When verifying timings in Windows, I run Karhu to a test scope of 10000% with no errors with 128 GiB. On my AM4 systems with about 32GiB, I go for 20000%
     
    Last edited: Nov 24, 2022 at 4:58 PM
  8. GamerNerves

    GamerNerves Master Guru

    Messages:
    306
    Likes Received:
    72
    GPU:
    RX 5700 XT Nitro+
    My friend has had issues on 22.10.3 and RX 6700 XT when simply turning Youtube videos to fullscreen on Chrome and I had two black screens total, which resulted to rebooting PC via the reset button. Now I have installed this 22.11.1 driver, but I'm already worried since on Firefox there was a sudden reconstruction of web page elements on display, which points to instability. I hope for the best, but jeez, what a mess Radeon Group has created. I'm surprised how Nvidia folks are relatively silent about how great Nvidia's drivers are, since now is the best time in a long while to bash AMD's, though of course immature bashing is pointless.

    PS. On a positive note, I'd like to point out that this is surely a big transition period for the Radeon Group, building from the ground up almost. After this we will eventually get better drivers than ever before!
     
    Last edited: Nov 24, 2022 at 6:04 PM
  9. Cp0

    Cp0 Member Guru

    Messages:
    112
    Likes Received:
    22
    GPU:
    AMD 6900 XT H2O
    I just wanna confirm as soon as possible right now that the freezes happen exclusively in wow, then find out why it happens, its incredibly frustating cos i had tdr this morning on 22.5.1 as well i think last 2 updates broke the game, seems flickering has returned to Nvidia users as well may indicate shader cache is handled badly or mpo related problems returned to Nvidia users as well, altho if currently disabled MPO again, also run enchanced sync on again i don't think it matters if its on or not.

    Noticed minor freeze while alt tabbing exist on 22.5.1 as well if enchanced sync is on.
     
  10. BlackFireHawk

    BlackFireHawk New Member

    Messages:
    8
    Likes Received:
    2
    GPU:
    6800xt
    I have a 240hz Display..
    Didnt See the Difference between freesync and full-hz at this refreshrates anymore..
    So i Just disable any sync.. No vsync.. No freesync.. No enhanced sync.. and No MPO.. and anything feels Fluid and responsive to me
     

  11. shm0

    shm0 Member Guru

    Messages:
    142
    Likes Received:
    29
    GPU:
    6800XT
    WoW runs fine for me with MPO disabled.
    But performance is really bad since the prepatch.
     
  12. Cp0

    Cp0 Member Guru

    Messages:
    112
    Likes Received:
    22
    GPU:
    AMD 6900 XT H2O
    Whats better for transients loads on asus board phase control optimised or standard ? if had it on standard before i know extreme turns on all phases at all times
     
  13. DimkaTsv

    DimkaTsv Active Member

    Messages:
    89
    Likes Received:
    22
    GPU:
    AORUS RX 6750XT 12G
    No idea... If you about LLC calibration, then usually Normal, Motherboard controlled and Standard are same thing from what i know. But it is hard to know for sure because not everyone have oscilloscope lying around.
     
  14. Cp0

    Cp0 Member Guru

    Messages:
    112
    Likes Received:
    22
    GPU:
    AMD 6900 XT H2O
    If read that optimised is mix of extreme and standard so better then standard while extreme uses all phases at once but i am not sure if also heard using optimised is worse then standard so if been using standard ever since but if never really figured out whats better, im guessing it i figure out if it matters or not vrm phase control does help with transients and i am suspecting im getting freezes as result , im about to have mental breakdown if i don't figure out whats causing it cos the only clue i get is GPU is getting either a tdr or whole system freezes, these tdr starting to happen on old drivers to, its easier to fix a hardware problem then a driver issue then i am at the mercy of AMD and we all known AMD drivers aren't to great right now.
     
  15. shm0

    shm0 Member Guru

    Messages:
    142
    Likes Received:
    29
    GPU:
    6800XT
    I'm also using standard for both cpu and soc.
    And soc calibration on highest level which results in flat 1.1v and but sometimes drops one voltage step.
    I think most asus boards have only 2 phases for the soc.
    So optimized is maybe not ideal when pushing higher clock + voltage...
     

  16. Espionage724

    Espionage724 Master Guru

    Messages:
    879
    Likes Received:
    398
    GPU:
    XFX RX 6600 XT
    I figure Extreme would be best for spreading smaller load across all VRMs instead of large load on specific ones, but I'm guessing at how that works and don't know what's really happening :p

    I've not noticed a difference at all with a 2700X with any of the settings, but right now I have it on Optimized.
     
  17. Cp0

    Cp0 Member Guru

    Messages:
    112
    Likes Received:
    22
    GPU:
    AMD 6900 XT H2O
    Still freezes this really frustating its almost as if this happens exclusively to wow i hear there is a memory leak some say it also effects dx12 some say it does not, but i see no proof of a memory leak as memory leak would mean memory usage keeps going up and it clearly does not unless i miss understand what a memory leak is that is not visible on task manager if so well crap.
     
  18. illusiveman

    illusiveman Member Guru

    Messages:
    198
    Likes Received:
    145
    GPU:
    XFX 6700XT Ultra
    Did you try setting everything to factory default ( CPU overclock, PBO, Curve optimizer, mem OC, xmp,GPU vcore, VRAM clocks etc) ?
    There are too many things that can cause TDR...you must eliminate at least some of them so that only the drivers, windows or some hw fault remains.
     
  19. Cp0

    Cp0 Member Guru

    Messages:
    112
    Likes Received:
    22
    GPU:
    AMD 6900 XT H2O
    Tried that before still causes system freezes im seriously starting to suspect its just the game if not had any time playing anything else cos i want to level all 53 characters to level 60 before next expansion, i only need to level 2 more characters i think i am gonna take a break from wow for the next 2 or 3 days and play something else in hopes to have no problems at all whatsoever, i never had problems on desktop either it never been more stable on desktop and desktop is where i used to get blackscreens before disabling MPO

    It seems whenever game is trying to load something it freezes most of my freezes have happened after loading, a new memory kit cant possibly faulty either im leaning towards memory leak in wow, altho i am not sure if that would be visible since usually i only sit at 16 gb memory usage in wow also when it freezes unless old memory does not clear and windows trying to acces ram that is clear but technically not clear.
     
  20. illusiveman

    illusiveman Member Guru

    Messages:
    198
    Likes Received:
    145
    GPU:
    XFX 6700XT Ultra
    If it was a memory leak the blizzard support forums would crash and burn lol. Any other people in the tech support forums complaining about this ?
    If this issue is only in wow it could be the game ... Idk I didn't play wow since wrath of the Lich King.
    Did you try running the game in windowed/borderless/Fullscreen mode?
    In any case I suggest writing a support ticket to blizzard.
     

Share This Page