It seems that no one is suffering from this as much I do, I have tried everything, and I mean everything. I have the 5650 "68C1" with the last WHQL from Microsoft driver for windows 10 x64 DriverVer=11/04/2015, 15.201.1151.1008 that support my card which is the same as the AMD beta 15.11.1 DriverVer=11/04/2015, 15.201.1151.1010 same files just added a few cards in the list, and it seems the best regarding this problem, but still very bad. I tried all the Crimson out of desperation although they include OpebCL 2,0 which is not supported by my card, but they seem to be worse regarding this problem. Now knowing that AMD is no longer supporting these card, it seems there is no way to fix it, and even reading in the AMD forum seems new cards with newer drivers still having this problem. In Linux Ubuntu when installing AMD driver I will have this problem, even it is worse in Linux cause there is no way to disable ULPS if exist. If anyone have a solution I will be more than grateful, otherwise we have only this option http://www.amdsurveys.com/se/5A1E27D277CE17F4 Thanks in advance
I was about to ask if AMD really hasn't fixed this issue after all this time, but my experience with this issue was fixed entirely on Linux (had no issues with both the open-source driver and fglrx when it still somewhat existed back then), so perhaps it's not the same issue that I've experienced. In Linux, perhaps it could be related to ASPM. I'd try the following kernel options: Code: pcie_aspm=off pcie_port_pm=off acpi_osi='Linux' For acpi_osi='Linux', there's also a bunch of other options for this too. Basically it spoofs what OS your BIOS thinks you're running and can change certain hardware configurations. Can fix issues with Bumblebee and GPU initialization, and more commonly, can fix brightness controls on certain laptops. Other combinations for it to try (not limited to) are: Code: acpi_osi="!Windows 2015" Code: acpi_osi=! acpi_osi="Windows 2009" Code: acpi_os_name=Linux acpi_osi=
Sleep wasn't resuming correctly on my desktop system until I enabled a feature in the BIOS which would force refresh the video cards on resume. This was off by default. Normally upon awaking, the screens would remain black.
hey bro i have this problem too so i wanna ask you about ASPM its supposed that when turn it off in win 10 for example give your more stability and less problems and black screens so i have ASPM in windows not in bios it was off disabled i noticed a lot of black screen from sleep sometimes from restart too than i put it as windows default moderate or something i noticed less black screens so its this making a sense for 7970
Thanks all for your reply, I am afraid it is not a power issue or bios nor kernel, it is a driver issue, here is the first time I got it http://forums.guru3d.com/showthread.php?p=4471456#post4471456 After AMD claiming of addressing and fixing this issue, it changed from a loop of waking up and sleeping to restart. The only thing will help is disabling ULPS in windows, but it will not completely fix the issue. Before disabling ULPS I will get almost 95% fail resuming from sleeping, but after disabling ULPS I will get almost 10% fails, but still very annoying when you leave your laptop most of the time to sleep and there is unfinished work.
Sorry for this late reply, I didn't think that someone will reply I disabled windows error report by adding a registry entry. Last time it happened yesterday, I am not sure there will be a report even if it windows error reporting system Is enable, cause there is no message just a restart. I will be more than happy to send you one, but how? although I am not sure it will not help, it is a driver issue, it happened since 1 update and kept happened, the only solution is to go back to old driver, or uninstall driver in linux
On some computers (as with my 790FX) W10 will only resume from S3 standby if HPET is forced on. This can be set via admin prompt: bcdedit /set useplatformclock true To return to default: bcdedit /deletevalue useplatformclock In previous TH1/2 it'd be on by default on this board that requires it (while it wouldn't on my other 890GX), but since RS1 it seems disabled by default.