I updated this post too many times so I decided to write it from start after testing the AMD OMEGA Catalyst driver. I have spent about 3 weeks on this problem as I am an owner of an r9 290 that I can’t rma and I am too tired to do so. I have managed to make the black screens disappear 1 week now, which never happened with any other driver version and any possible set of tweaks.
In general I will say that thις problem isn’t related to PSUs or older PCI-E version slots or gpu temperatures directly or gpu memory speeds directly. As I read at some forums, even with many tweaks even ROM changes, the problem persisted even though at less frequent appearances . It is definitely, for some users, a hardware fault and a reason to RMA your card if you can. If you can’t anymore or you got bored RMAing or you just plain cant then read this post and try to implement all the tweaks I gathered the last 3 weeks of my effort.
Keep in mind that, at least for me, with Omega driver you just need the tick on "force contant voltage" on msi afterburner and adding some voltage on System Agent voltage at motherboard BIOS. With just those two tweaks I haven’t seen a crash all day now, even at games that were crashing the least often every 2 hours.
First of all, do the proper thing and clean your system of previous AMD drivers before installing Omega driver
Now for the tweaks
1) At bios, make sure you add some voltage to System Agent Voltage. A 1.055 is the max I would suggest, I am at 1.005, keep in mind that every motherboard is different in terms of sensitivity at voltage tweaking.
2) Uninstall your amd radeon drivers
Ι had issues properly uninstalling 14.9 and every driver i installed afterwards was installing with some report logs and ! marks.
Download Display Driver Uninstaller (or DDU) 13.4.1.0 and extract it in a folder
Boot to safe mode (various ways to do that, that differ from OS to OS google it)
Go to device manager, find your gpu, uninstall it and tick also the "uninstall all associated drivers" or smth like that
Run DDU and select AMD then uninstall and restart pc, you can also select to delete the c:\amd folder
it should reboot to normal mode, check device manager, should have as display controller a "microsoft basic display adapter" or smth like that.
install your Omega amd driver. Although i was avoiding AMD Gaming Evolved software, some games optimized by it, seem to have become 100% stable! So I know, strange but whatever works right?
reboot
3) Install, Start MSI afterburner
1) Settings - > Force constant voltage ticked
2) Settings-> Unlock Voltage Control ticked (MSI )
3) Custom Fan settings, I prefer setups like 30% fan speed up to 60Celsius, then 100% at 80% Celsius and adjust it depending your room temps, as long as you never see 75+ Celsius. I do that because I believe that graphic cards that their temperature changes never go over 20-30 celsius last longer. Extremely true on laptop graphic chips..A Simple contraction expansion rule really.
4) Make 2 profiles (3rd is optional but it proves something is weird and not related to memory speeds 100% )
1st Profile similar to the one MrToast72 suggested : Core Voltage +25, Core Clock 875, Memory Clock 950, Power limit +5%, let's call it Safety profile (if you still get some black screens take power limit to -10%, it worked for some!)
2nd Profile, the average profile (that proves that the other settings in bios , drivers are helping) Core Clock Default, Memory Clock Default, core voltage 30, power limit 10%, let's call it Default profile
3rd Profile (and what proves memories could be not at fault but a controller or something like that on the gpu or an unstable auto-voltage rule somewhere)
Core Voltage +40, Power Limit +20, Core Clock 1000!, Memory Clock 1300! Yes an overclock profile even for my VTX3D r9 290 x-edition V2,
That’s it! Hope your black screen errors have minimized or even disappeared.
I have gathered a lot of information around the web for this frustrating problem, in case you wanna test more.
For some users disabling adobe flash acceleration in browsers helped
For some others, installing only the drivers of the gpu and not the whole catalyst packet helped
Overclocked CPUs were affecting negatively the whole effort
Some others said that older PCI-E versions were affecting it negatively
Some others blamed the PSUs but I have a 1000watt 80% psu and have it.
Some others blamed the memory of the gpu. Although I can’t confirm otherwise, the overclocking profile should be crashing continuously if that was the case.
UPDATE 11 April 2015
After more and more and more tinkering i have concluded that something doesn't always work and overheats without apparent reason. In the end the most stable of all setups (1 game freeze every week max!) is by having
a core clock of 825, memory clock of 875, Core Voltage -19mV !! and power limit -15%!!!!
This setup is extremely stable even at dragon age inquisition and all very heavy games (Which seem to be more stable even at overclocking settings! compared to world of warcraft...)
Update 20 May 2015.
At some points, once every few days, even when PC was idle or just desktop work, the graphic card fans would go crazy 100% speed, with VRM1&2 and GPU CORE temps being ok (40-50Celsius). This made me think that there is definitely something not working ok as a component on this graphic card, that will give a red alert temperature wise and fans will go crazy just to help. Also mentioning that after 1-2 minutes of 100% fan speed gpu would crash, with black screen. I have tried extending the intervals in registry of windows checking if the graphic driver works or not . Except that i found the time to add Geils Enhancement R9 290/x kit. There was no difference on the VRM heatsinks from stock to geils but what was different was the thermal pad quality (of course..). The only difference i see is that VRM temp at idle is like 15celsius less but when you push the card in games VRM1 and 2 Temps are almost identical after a while (makes sense). There were 12 days (i keep calendar for the crashes) straight of heavy GPU use (from adobe premiere pro cc mercury engine, to video converting and gaming) without any strange incident, then things started to happen again on a one crash/black screen every week. (My pc is like 18 hours on per day or even 2-3 days straight). So i can't say that what the other poster said , about geil enhancement kit works as a charm, is true but there is obvious temp difference on the GPU VRMs and i did see over a week of no crashes. What i believe is happening is that the VRM heatsinks are of the same quality as of the manufacturer's ones but what changes is the quality of the thermal paste/pad/sticker helping the heat transfer easier to the heatsinks. After all in my gpu the geil vrm heatsinks are smaller in size and slightly lighter, i think they are of the same material.
Also, i implemented my fan settings that i was using on my ATI HD5870 card. The settings were set fan speed to be really low up to 65celsius, and then increasing steadily to 100% fan speed at 80celsius. This made the card slightly more warm at idle but the idle temperature to the high cap temperature is way less and perhaps this also helps with the simple rule "expansion contraction". Meaning that if a component on the card is behaving erraticaly due to changes of temperature being to harsh, perhaps if it is hotter all the time might mean that it won't crash if you just add 20celsius up, down from 40..)
The end result as i speak is 4 days no crash, with mantle enabled games going on (Battlefield 4, PvZ garden warfare) , world of warcraft , dragon age inquisition, World of tanks (its stable only if i remove post processing), Robocraft, a full clear on wolfenstein old blood, and some hours of gpu stress test from msi kombustor.
I want to add that Unity engine games (like wasteland 2 ?) are pretty buggy? or very unstable on my R9 290, i don't know why, they create a crash, the monitor display seems to tear (like on heavy no vsync visuals) and then crashes, havent found a solution for that on any setting.
--- UPDATE windows 10 x64 15.7.1 AMD drivers---
I upgraded my windows 81 to windows 10 but didn't realize that the 15.7.1 drivers i had were of windows 81, so after a DDU uninstall and then installing the windows 10 x64 15.7.1 driver guess what happened. As a rule of thumb i reset msi afterburner settings before an amd driver uninstall and a new amd driver install. After the whole process i forgot my R9 290 at default settings (+0 power, +0 core, 975 / 1250) and i was playing for few days without incidents.
Then one afternoon my daughter told me that while she was playing Garden Warfare PvZ she had a sudden black screen but the game didn't crash afterwards. I checked and realized my gpu was at default settings , and started playing wow on heavy graphic or lots of pupulated areas to see what will happen. Few minutes later on demanding gpu areas, i had a crash. However it was just for 2-3 seconds, the driver recovered and the game kept playing!
Something with windows 10 and the new driver makes the black screen temporary, the driver recovers without the game freezing or dieing. I have not tested many games, Garden Warfare for example is a 50-50 situation, WoW always recovers, same with need for speed rivals and most wanted.
I now play on my safemode gpu settings (-20% core clock, memory clocks, power limit and -19mV voltage) and i have no crash for days, even though i did stress the card with crashing games like WoW on heavy areas, need for speed most wanted and some others.
So keep that in mind if you get your windows 10 and the new amd driver. Test the games with default gpu settings on msi afterburner, if the crashes are temporary and games recover without them freezing/crashing report it here
UPDATE 05 March
I found the arctic cooling accelero extreme iv and i made a small customization and added the arctic cooling kit too(didnt use screw at one side and used a tightener) . There is a link in here somewhere how to do it, i have done it similar. Also, updated the drivers to the lastet from amd 1-2 days ago.
After this the heat on the gpu, rams and vrms are like super low. going just 60-65 at overclocked!!! (1075gpu 1375vram) and the black screens are as often as one every 10 days, actually i posted tonight after the first black screen in 10 days.
It happened at overclocked profile in unreal tournament 4 prealpha which i playtest.
The problem didn't disappear, but as i already said, every bit helps to reduce how often it appears. It is definitely a problem on a chip and it is affected by temperature but the temperature is not the culprit.
Update 28th March
With the new beta amd driver update (Crimson Edition 16.3.1 Hotfix AND 16.3.2 just yesterday will keep informing for that), the accelero extreme iv, the accelero vrm kit, (i promise to add a picture in few days where i plan to clean the PC again from fiber dust etc) and the following settings i have gotten no black screens on all games. I dont know if this has fixed the problem but it sure doesnt show.
Core Voltage -12mV
Power limit 0
Core clock and Memory clock default
Also suprisingly a 10% overclock on my gpu and gpu ram is also super stable, no crashes yet.
Core Voltage +25
Power Limit +20%
Core Clock 1073MHz
Memory Clock 1375MHz
I tested on varius games non stop, managed to take a screenshot while playing need for speed most wanted with 4x Antialiasing at 1920x1200 max details. I have 2 monitors so its easy for you to see the data and the game loaded
I think this effort of 1.5 year is covered and complete with partial if not complete success. As i said, you can spend some money and a very specialized electronic engineer can find out what vrm or ram has some failure and fix it..or do this
http://i67.tinypic.com/rvk8kx.jpg