1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

AMD Fury X Owners' Thread

Discussion in 'Videocards - AMD Radeon' started by Fox2232, Jun 25, 2015.

  1. Synoxia

    Synoxia Member Guru

    Messages:
    113
    Likes Received:
    11
    GPU:
    Rog Vega 64
    Your Bioses are for Fury Nitro, not for Fury X unfortunately
    The last hope is Fox XD.
    So strange btw, my card is 100% stable in all games but in 2d game like LoL i get artifacts and crashes with FX5 and FX6... FML
     
  2. Agonist

    Agonist Ancient Guru

    Messages:
    2,726
    Likes Received:
    156
    GPU:
    2x RX 480 Nitro 4GB
    I have the Nitro OC+ 1050/500 1.225v. And I can undervolt it to around 1.152 and its stable on stock clocks. Stays under 60c even in a warm room.
    The OC+ do not unlock sadly to Fury X. Tried that already.

    Id love to get as much out of this card though. Its starting to show its age in new games.
     
  3. Synoxia

    Synoxia Member Guru

    Messages:
    113
    Likes Received:
    11
    GPU:
    Rog Vega 64
    @Fox2232 @OnnA can you help? Does the FX6 bios already have offset? Must change FA to 01,02 etc and save?
    [​IMG]
     
    Last edited: Mar 8, 2018
  4. Synoxia

    Synoxia Member Guru

    Messages:
    113
    Likes Received:
    11
    GPU:
    Rog Vega 64
    Anyone has a clue?
     

  5. Fox2232

    Fox2232 Ancient Guru

    Messages:
    9,523
    Likes Received:
    2,104
    GPU:
    5700XT+AW@240Hz
    Yes, FA = -6 * 6.25mV , that's effect it has on base 0.9V.
     
  6. OnnA

    OnnA Ancient Guru

    Messages:
    8,914
    Likes Received:
    1,742
    GPU:
    Vega 64 XTX LiQuiD
    I have value of 09 (1.356v)
    9*6.25=56
     
  7. JonasBeckman

    JonasBeckman Ancient Guru

    Messages:
    15,419
    Likes Received:
    1,451
    GPU:
    Sapphire Vega Pulse
    Wonder if it's similar to Vega and there's several different memory manufacturers thus the voltage differences or if it's just just the usual that some are better binned (Possibly depending on GPU model.) and can just handle higher (Or lower.) amounts without running into stability issues. :)

    Probably not quite the same as Vega though where the initial memory modules were the better ones and then the current models can't manage quite the same overclock for whatever reason.
    (Availability perhaps or just cost related, it still operates within the cards specified speed at least though doesn't overclock as well.)
     
  8. OnnA

    OnnA Ancient Guru

    Messages:
    8,914
    Likes Received:
    1,742
    GPU:
    Vega 64 XTX LiQuiD
    Today i this day ;)
    look at this.


    [​IMG]
     
  9. gupsterg

    gupsterg Member Guru

    Messages:
    177
    Likes Received:
    1
    GPU:
    RX VEGA 64
    Putting aside HBM clocks for a moment the purpose of that table was to show how on Fiji increased voltage can lead to performance scaling gains not to occur. I tested it also with increased PL ROMs, one time I did 250+ 3DM FS benches on Fiji. I owned 1x Fury Tri-X, 8x Fury X and 2x Fury Nitro over the course of my foray with Fiji. I gained various leakage GPUs (ASIC Quality), regardless of this aspect increased voltage would lead to performance scaling loss. The Stilt also created ~3 ROMs for me with a modification which some Polaris bios modder are using these days to tweak GPU voltage, even those mods had no effect.

    Register 8E is loop 2 of IR3567B. Changing data value offsets HBM voltage. Data value is bold test 8E 00 00 00. 01h = +6.25mV (1x 6.25mV), results in HBM @ ~1.306V. 0Ah = +62.5mV (10x 6.25mV), results in HBM @ ~1.362V.
     
    Last edited: Mar 30, 2018
    OnnA likes this.
  10. M300

    M300 Member

    Messages:
    22
    Likes Received:
    13
    GPU:
    R9 Fury
    Furry Nitro OC+ owner here, overclocks bad, undervolts okish.
     

  11. M300

    M300 Member

    Messages:
    22
    Likes Received:
    13
    GPU:
    R9 Fury
  12. z8373767

    z8373767 Member Guru

    Messages:
    171
    Likes Received:
    47
    GPU:
    Fury Nitro 1070/600
    Welcome. Sounds familiar. Max stable core clock for my Fury is 1070 on 1.281V(stock), PL 0%.
    Forget about any undervolts xD.
    If you want some extra performance take a look on Onna's modded Bios and do some HBM OC. Of course change voltages for your Fury, if i remember Onna's bios have 1.212V which is unstable on my GPU. Use Fiji BIOS Editor for it. Before all, make backup of your original BIOS. And flash Blue-One BIOS, push button on PCB.
     
    Last edited: Apr 1, 2018
    M300 likes this.
  13. Agonist

    Agonist Ancient Guru

    Messages:
    2,726
    Likes Received:
    156
    GPU:
    2x RX 480 Nitro 4GB
    Exactly how mine is.
    Still performs good though.
     
  14. JonasBeckman

    JonasBeckman Ancient Guru

    Messages:
    15,419
    Likes Received:
    1,451
    GPU:
    Sapphire Vega Pulse
    Interesting, was comparing bioses via the TechPowerUp bios repository and discovered that the Tri-X secondary bios which has a higher GPU temperature threshold also has a difference in watt or rather the allowed power draw I think is what corresponds too which is not present in the newer Nitro models even for their alternate bios switches or the OC models.

    So instead of 300 this runs at 350 so if I'm doing this right taking the 300w regular draw limit and applying a 50% in Wattman would place it at 450w max allowed power draw however for the alternate bios this instead becomes 525w and going by the Fiji bios editing thread on Overclock .net and their forums that might be too much for the power connectors.

    However from comparing reported power draw and over and under volting and clocking the GPU the actual usage is generally around 300w so actually exceeding these limits might not be possible thus there's no issue with having a higher allowed draw limit since the card could never reach it before becoming unstable either from voltage (1.3v core clock being max aside from the Nitro model.) or overclocking. (1100Mhz or so depending on chip.)

    Probably a non issue in the end, guessing the newer chips in the Nitro have better overall ASIC or how it's calculated and thus reduced leakage and power usage or seeing how these cards overclock (Not very well at all.) the alternate bios for the initial Tri-X model might have already been above what the card could ever reach so it was removed from the newer Tri-X model which just has the higher temperature threshold. :)
    (Card will draw what it needs after all, lowering the threshold could be a way to throttle it down a bit but increasing it might not really have much of a effect as a result.)


    EDIT: And yeah it's a good card for undervolting, quite a bit of headroom actually. Though it can take a lot of testing to find a fully stable overclock or undervolt since every software is a little bit different and pushes the card in different ways, guess that's something else to look into one day. :D

    Especially as to why some lesser demanding games can really push up the fan curve but some of the more demanding titles and the fan is all quiet and the card is cooled without issue, reported temps can be the same too but some games really have a tendency to ramp up the fan speed and I think there's a bug or glitch that can also temporarily raise the fan curve to where it ramps up to near max for a little while before the driver scales it down again, not really sure but it can be quite a surprise when it happens ha ha.

    Guessing that's on the driver software though and maybe related to Wattman's functionality depending card model and fan settings and hardware.


    EDIT: Actually a bit surprising that the Tri-X has this difference since it's the reference PCB with a custom cooler and it's the Nitro that uses a custom PCB allowing for higher overclock results, depending on chip and even then hitting above 1100Mhz might be difficult.
    (Then again the reference PCB much like Polaris and Vega is very over-engineered for the GPU chip and quite capable already although for Fury the stock air model ran into limits.)
     
    Last edited: Apr 21, 2018
  15. z8373767

    z8373767 Member Guru

    Messages:
    171
    Likes Received:
    47
    GPU:
    Fury Nitro 1070/600
    I really appreciate Nitro Cooling system. I mean at all, because this IFC system is a joke xD
    Originally my Fury easy reach 80°C and clock down.
    I investigated that, only one fan spins. Centre-one.
    Made custom curved-map for fans, which force all fans to work.
    1400RPM = 74°C on 300W GPU!
    I had friend's PowerColor RX 480 Red Devil for two weeks.
    To get 74°C fan reached 2400RPM :eek:
    150W GPU.

    I think i never mentioned that.
    Your Fury coil whine?
    My previosly card, R9 280X had it too :D
     

  16. JonasBeckman

    JonasBeckman Ancient Guru

    Messages:
    15,419
    Likes Received:
    1,451
    GPU:
    Sapphire Vega Pulse
    Yeah the current GPU is thankfully not very whiny, 290X I had years ago had a tendency to whine though but it also ran at a target temp of 95 degrees though it was engineered to last but that's also some additional 30 degrees or so generated by the GPU which even for a blower design (Which was also quite loud.) it still leaks into the case itself.

    Both the Tri-X and Nitro have a thing about trying to keep a 20% fan speed until it hits the thermal target but even at 20% they're not bad, zero-fan tech is intended to ensure only one fan is spinning and the rest can be shut off but I prefer to keep a permanent 30 to 40% constant speed since the noise doesn't bother me and this particular model only starts getting audible after the 50% mark or so though it then quickly becomes very loud ha ha.
    (Good day without too much heat from outside I can manage around 28 - 30 degrees idle, hot summer days that's going to be upwards of 40 degrees though and then around double during full load but it varies depending on game, shader workload seem to generate more heat.)

    Hardware wise I believe the GPU and HBM modules are cooled very efficiently with either the tri-x or nitro (And it's lengthy metal heat sink.) but the custom PCB on the nitro affords less tight VRM placement and these are cooler for that instead of being +20 to 30 degrees above what the GPU core reports. (Still rated to handle upwards of 150 degrees however so it's unlikely to be a problem.)
     
    z8373767 likes this.
  17. Synoxia

    Synoxia Member Guru

    Messages:
    113
    Likes Received:
    11
    GPU:
    Rog Vega 64
    Guys i am afraid to tinker with those hex values... how does this look? I simply wrote FA to 01, can i load this into my Fury and not blow things up? :D
    I think fx6 bios is unstable with my card because its not strong enough with tight hbm timings
    [​IMG]
     
  18. OnnA

    OnnA Ancient Guru

    Messages:
    8,914
    Likes Received:
    1,742
    GPU:
    Vega 64 XTX LiQuiD
    Hmm, i can say that 'Blow up' is impossible ;)
    Just try, test in HWinfo64 -> if OK then leave it.
     
  19. G13Homi

    G13Homi Active Member

    Messages:
    72
    Likes Received:
    23
    GPU:
    Vega 64 2x xfx
    http://www.3dmark.com/fs/15493125

    still cruising after all these years on stock clocks, ever since 18.1.1 ive been enjoying xfire in pubg without any issue and AMD drivers have really matured for me, i cannot justify an upgrade yet.
    ( i just imagine what i could do with this machine if i had half of the knowledge you folks have)
     
  20. user1

    user1 Maha Guru

    Messages:
    1,340
    Likes Received:
    417
    GPU:
    hd 6870
    curious, has anyone tried running a furyx with dpm disabled (via grub) on linux, and test whether the voltage scaling problem goes away? came to mind while i was messing around with the powerplay tables.
     

Share This Page