Thanks for your answer. But what is the function of acoustic limit? What happends when I change the values?
https://community.amd.com/thread/206281 Nice, in theory. Those are always good. Well hopefully it works as described.
I have acoustic limit set to 0Mhz in all my Profiles -> yes it is working Fine. Im on 0.2.7 Beta4 I've uploaded PP States for Vega XTX into Overdriven_Tool and Fans working correct.
Here from @tede: https://forums.guru3d.com/threads/overdriventool-tool-for-amd-gpus.416116/page-19#post-5558335
It shouldn't hit maximum fan speed until the GPU is at this frequency, used to control overall fan speed by allowing a lower noise level and then once the GPU is within this value it can ramp up. As such you don't want the value to be too high either as that's going to impair cooling by keeping the fan from going full speed. From how I'm reading it at least I'd probably keep it around the 5th or 6th clock speed state and the value for this so when the GPU hits the 3d states for core clocks then the fan can spin up and cool down the card.
I think that's the minimum temperature before the fan ramps up or where it tries to keep the GPU temperature but without spinning to max but it's not until it hits the max temperature specific that it ramps up to full so in effect the fan speed will be between the min speed and max speed determined by reported clock speed and GPU temperature. Plus bios settings and the thresholds therein but if it hits these temps it'll probably just shut down to avoid GPU damage. (The fan speed settings are fairly dynamic and also changes depending on GPU type and then third party models having their fan controllers and tricks such as Sapphire and a few others shutting the fans down completely until temperature hits 60 degrees I think it was and then you have bios specifics such as fan granularity which determines how fast it reacts and ramps up or down and then the various limits or thresholds from where it will ramp up to avoid damage to outright shutting down if temps get too high.) EDIT: Several settings doing similar things and having similar results but it's divided between clock speeds and temperature and then you have bios specifics and GPU vendor modifications for third party models basically, personally I find Wattman to be off by a few percent as well from what it's set to and what it actually reports although Polaris and Vega aren't as bad having been made for the new Wattman and it's Overdrive API instead getting uplifted to the new version in later drivers which worked to varying effect. (Bios editing being popular for Fury GPU's for example and they also differed more with a 5 - 6% offset compared to what you used in Wattman and what the GPU would report, not very nice when you're trying to dial in voltages and more sensitive adjustments.) EDIT: So if you set a 50 degree limit it'll gradually ramp up as this is exceeded and then try to bring it down again but if it then hits the max temperature it'll push fan speeds to the max allowed and get temperature back under control. I prefer to keep the max temp at a bit lower value since it's core and other components also get pretty hot but 90 should be fine, VRM's and such on Fury, Polaris and Vega are pretty highly rated although with air cooling that also means you'll be having hot air in the rest of the case until this gets handled by the fans and airflow. Guess it's around 95 - 100 degrees or so where Polaris will shut down but it shouldn't be too common for it to get that hot under regular usage situations. (Benchmarking or a very high ambient temperature maybe pushing it to the 90's at worst though both the 400 and 500 series should be far cooler than the 200 and 300 series were before it and then the Fury and Vega GPU's too.)
Here: My LC Fan will go up to 3300RPM by Default. I have 2nd one in front of GPU one in Push-Pull Max is set ~2900 (when tested in Prey >2h w/Mods I've noticed HotSpot Max is ~70deg, GPU & HBM is ~55-60deg, Fan Max is 2893RPM) Acoustic should be as Your Max GPU clocks or 0 ("0" means disabled, My default is 1750MHz) Max for any other Vega should be ~85-90deg. Target ~65-75deg. Note: Always use HWinfo64 or WattMan to see sensitive Temps: 1st. GPU/HBM 2nd. HotSpot Not needed when you set your GPU right where you want it Test for minimum 20-30min in Heavy Game or 3Dmark Stress Test or Loop. Note 2: Yes im Gaming at 1732 or 1717 Depends on Heavyness of the Game with UV Average in Heavy game for 2-3h is around 100-130tW including ZEN 70-80GPU 30-50CPU So much for the statement that Vega is Power Hog
The GPU is 68C and fan is 54%. Here are the settings for the card. I though the fan would speed up more since target temp is 50C. Any ideas? Should I lower acoustic limit? There is no way I can pump up memory clock more :/ I get an error message.
Try acoustic at 0 Also Look & check is everything OK inside PC. Good to have DIY Fans for better cooling. e.g. Blowing onto the Back of the GPU or >>===>>===>> Good Fans: Fander Roxo FRX2-14025-PWM 600-1300rpm 140mm Gelid Wing UV 140mm Cooler Master JetFlo Air Flow 140m Also Good are Noctuas but those are priced notch higher (Good but IMO Fander do the same job for half the price 3-5€ vs 10-15€ lol)
Hi Guys, Overdriventool saves the pci vendor for enumerating the gpus? someting like afterburner.. For example i have 6 gpus, 1 vega and 5 rx580, if one goes off for something reason, the overdriventool will not overwrite the setting of one gpu to other ? Thanks
Hello. I have a clean wina installation, with Vega cards, I've tried adrenaline drivers 18.7.1 and 18.6.1, but with the "Failed to load ADL error 126" error, I've already tried all the possible guides on the net and nothing is the same. The only case I'm going to load over is with the blockchain drivers, but I do not want to use thank you if anyone knew the advice
Try 18.5.1 and see if that works. 18.6.1 and later seems like they're changing stuff around so I'd start with a older driver and see how that works. I need to look up that error though and I need to check how bios editing affects Vega just in case it's running on a altered bios file though if so TechPowerUp and their bios database should work to get it back to defaults if needed. Just be aware that Vega 56 has either Hynix or Samsung HBM2 memory chips and the configuration between these differs so do not flash a Samsung bios on a Hynix GPU. (Even the Nano variants such as my own Pulse GPU come with Hynix although the initial batches had Samsung but it's not as common anymore though not entirely replaced by Hynix either so GPU-Z and checking the reported memory type and bios config would help verify what it's using.)
Hi Guys, For example i have 3 gpus, 1 vega and 2 rx580, if one goes off for something reason, the overdriventool will not overwrite the setting of one gpu to other ? Example: 0 - Integrated VGA 1 - Vega 64 2 - Rx580 3 - rx580 Now, i have an .bat to apply all the setting for gpu 1, 2 , 3... Now if for some reason vega 64 goes off, my vega 64 setting will go to rx580 right? Because overdriventoll will enumerate the offline Vega 64 with Rx 580, and the setting will not apply.. Im just asking, not a problem yet Thanks
I keep getting 3 different ADL error 8's when I try to apply any settings with 0.2.7b4. Using a VegaFE. I can save the bios and create a custom registry PPtable that lets me change HBM clock speed but that seems to be all that is affected.
Can you execute "Reset all" from command line ? Would be awesome since Wattman doesn't allow changes if computer resumes from Sleep ... it thinks it crashed. So far I've been able to do it with: http://www.gandalph3000.com/cast_xmr/switch-radeon-gpu-compute-mode-hbcc-largepages/ "switch-radeon-gpu.exe -G 0 restart --admin" and "switch-radeon-gpu.exe" set to run as admin. This helped also: https://sumtips.com/how-to/run-program-windows-wakes-up-sleep-hibernate/
