Hello Tede, I have issue with OverdriveNtool. When I am restarting PC normally it works great. However when I will cut off electric and switched that on again all profiles will remain empty. It will stay empty even when i will try to run bat file manually. I have to go through the all GPUs select profile, apply, save and it works again till the first power shortage again. I have 3xsapphire Pulse 8GB and 3xsapphier Nitro+ 8GB special edition and 1200w HP server PSU and corrsaire 860w PSU for risers, motherboard H81 Pro BTC and SSD OverdriveNTool.exe -consoleonly -r1 -cm1″P01RX580-SP″ -r2 -cm2″P02RX580″ -r3 -cm3″P03RX580″ -r4 -cm4″P04RX580-SP″ -r5 -cm5″P05RX580″ -r6 -cm6″P06RX580-SP″
Currently, it doesn't seem that Overdriventool will work with the Vega Frontier Edition when using Adrenaline drivers. I get the error message "Function not supported by driver" no matter what setting I choose. Is it possible to add support? Thanks
It was working some drivers ago in pro mode but seems amd has blocked it on latest drivers. Switch to game mode or try to use PPTable editor. Maybe I should add an option to ignore this error -8 (ADL_ERR_NOT_SUPPORTED - Function not supported by the driver) because also Fiji users seem to have this error on memory apply. This won't fix anything but at least get rid of annoying messages
After testing for several hours, over the course of days when able, these are the settings that appear to offer the best performance while being most mindful of the GPU hotspot: https://www.dropbox.com/s/qvg8gdkwayagdg6/Vega_Frontier_Settings.PNG?dl=0 This is in a system using a Corsair H115i water radiator setup for the CPU. The only major heat dissipator that is purely air-cooled then is the AMD Vega Frontier. To keep the GPU hotspot around 85 degrees Celsius the fan must engage at its maximum RPM during load. To ensure that happens the Target Temp was set to 60 degrees celsius. The target temp in the fan section appears to only affect when the fan will engage its maximum RPM setting, and the gradient leading to that point. Note the minimum RPM for this Frontier was set to match the minimum RPM consistently observed, and the maximum RPM was set to the maximum RPM the fan ever obtained when higher settings were set. The following are graphs resulting from 1 hour of running the GPU synthetic benchmark built into AIDA64: https://www.dropbox.com/s/4p8n4vwjzevp41n/20180211_GPUsynthBench_1hour.PNG?dl=0 https://www.dropbox.com/s/l47vjk5itsrhjze/20180211_GPUsynthBench_1hour_fanRPMs.PNG?dl=0 Note the voltage levels were either left the same as the original settings or, in the case of the Memory P3 setting, under-volted. Only the clock settings for the GPU core were boosted in some cases to more closely match the ratio of MHz:mV in the P7 stage. Lower voltage stages have lower ratio values as seen with the default P0 stage values of 852:800. My two cents for settings on this otherwise factory default Vega Frontier edition. Note I'm running the standard Radeon Pro Enterprise 17.Q4.1 drivers (no wattman needed) and playing games in UHD resolution with 10-bit color mode enabled for the monitor and GPU. Skyrim Special Edition is amazing with these settings . . . Absolutely ZERO driver issues. The key for stability is proper cooling of the GPU, and overall card (like the hotspot). With the above settings the AMD Vega Frontier has been ROCK SOLID; adamantium even. * One further note: Rarely does the VFE push past the P5 stage. Not sure what all the factors are for that.
Hello guys, i have a rather bizarre problem. Once i unplug my monitor and try to access the computer via teamviewer i get an error message and my cards do no receive any tuning. This is an error coming from the .bat file i have on windows start up. i also get the same error for all GPUs, the strange thing is that once i plug my monitor in, everything goes back to normal and the bat file is working properly. I am unsure what to do, maybe just create "1gpu1" profile. So please guys, take a look and tell me is my code wrong, am i missing something else ? Additionally, the OverdriveNTool gui is not working with teamviewer, when i select it i either get a black screen or nothing. However, i am not sure whether this is a Teamviewer or Overdrive issue. Thanks in advance
@AirStyle, the problem is that you currently have less than 11 GPUs and use two digits as gpu_id, which is correct only when there is more than 10 GPUs. By plugging your monitor you probably get an additional GPU with gpu_id=0 and GPU count raise to 11, and above commandline is correct. When monitor is unplugged you have only 10 GPUs and gpu_id's are shifted by -1 When this additional GPU is not supported best solution for you is to go to "Settings" and check "Do not list unsupported GPU's", then change above commandline to: OverdriveNTool.exe -r0 -p0gpu1 -r1 -p1gpu2 -r2 -p2gpu2 -r3 -p3gpu4 ..... -r9 -p9gpu10 Note that your profile "gpu1" is now applied to gpu_id 0 instead of 1 and so on About TeamViewer no idea, I don't use it.
Thank you, that fixed it. I originally though i should use the 01,02,03 even if i have 10 cards, obviously that is wrong. Still i have a massive hah rate drop, from 30 to 1, on GPU0 when i turn on team viewer and still cant operate Overdrive under it. However, those are problems for another day. Thank you ! Edit: I had to edit hash to hah so it doesn't say hash rate
Dear Tede, My question is about P1 memory Profile and voltage. Is it possible to edit the P1 mem voltage field directly in the bios file with a hex editor? It currently seems none of the polaris bios editors support this feature correctly. It should be very possible but the main thing is to find the correct line and I was unable to.
Hello Tede, After testing it still causing the same issue. I am sorry. I tried to do it several times.
How I can prevent closing window at the end of start up script. I tried pause and several other commands and nothing happend.
Tried Radeon Pro Software Enterprise 18.Q1 drivers. They disabled the ability to modify all of these settings, including the fan settings. The AMD Vega Frontier is flat out unstable with a maximum RPM of 2000, or at least mine is. Absolutely ridiculous they would disable something that was not broken. Those that actually want to make use of the card are not supposed to use the Enterprise drivers? WTF? You have my money, why would you now prevent me from using the product the way I want? Fire the marketing non-genius that came up with the plan to burn all of the enthusiasts on non-oem systems. How dare end users attempt to get the most performance they can for the money they spent . . . Disgusting. So, I'm back on 17.Q4.1, where the utility is still able to make the changes. Looks like I'll be out of a driver update for at least 3 months. Someone at AMD needs to be let go . . .
From my testings problem was fixed. In 0.2.5 ini file write caching/buffering is disabled and I can reproduce ini corruption in 0.2.4 after electricity turn off, but not in 0.2.5. I don't know what else I could do, maybe automatic ini backup? Did you edit ini file manually? Btw. file corruption after system crash is common problem not only for OverdriveNTool ini file but for all recently saved/opened files. I suggest to never turn off electricity when windows is running.
I have a Dell RX580 8 GB Hynix and a Sapphire Nitro+ RX570 OC 8GB Micron installed in my machine and I am using this tool version 2.4. After getting the RX570 all installed and drivers patched and bios modded and all that I was trying to overclock and undervolt the RX570. I was watching in GPUz and when I would make changes the settings would show an update but in my miner XMR-Stak no H/S changes even if I put my memory clock down to 300 MHz. Not sure what I did to correct this but I went into the Setting in OverDriveNTool by right clicking the title Bar. I selected rescan on restart and turned on some of the options in the general area. I restarted OverDriveNTool and loaded my profiles and BAMM RX580 and RX570 now both hashing about 900 H/S on XMR. I am not crazy I spent all day trying to figure this out. I am using the block chain drivers version 17.14 I believe. Thanks for the tool Tede
Hello to everyone from Greece, I would like to thank the developer Tede for this very handy tool. However the tool does not seem to work with vega frontier edition and latest 18.Q1 drivers as posted above. Can we please have an update . Also it would be very nice if we can use this tool with a working gaming driver in the future, although.....i have not found any working gaming driver so far for vega frontier.... Whatever i try to install, AMD driver installs drivers automatically only in pro mode when it sees the frontier vega, and not only that but also locks down the fan speed to 2000 rpm making vega cards practically unusable. Also switch driver option does not function as well no matter if i have one vega or more installed... But all those are driver bugs not a utility issue and we have to report to AMD.... For the time being as far as i can say, vega frontier owners, we have to use only this utility and 17.Q4.1 drivers in order to use the frontier cards..so straight out of the box with only AMD software the cards are practically unusable....
I have R9 Nano with latest Adrenalin driver and I can confirm that HBM overclocking isn't working. I get the same "Failed to set memory clocks(...)" error message.
Yes Fidelis, the factor is your power limit, set this to 50% and it will lock to 1600 P7. Also it is highly recommended to lower P7 voltage to 1120 and P6 to 1110, (values may vary from card to card but those seem to work 90% of cases) also max out the fan speed and you are good to go