How TDP power limits (wattage) respond to values (VDDC) as card seems hotter than before flash. GPU-Z reports 1.125v constant before was 1.0xx dynamic but other programs reports 1.25v, now 1.13v and card is hotter about 7-8C above. I am sure that lower VDDC will be better as this is just a "test bios mod" to test overclock potential and I need to decide which values must be set for 24h stability usage without risking card health.
Basically voltage can be set to max if your voltage regulators have cooling and your sure of it. This will maintain stability at any clock you choose as long as it's within that voltage threshold and can only be found with testing and continuous monitoring.
I find the tdp is basically a peak allowable usable watt ouput the card can see, for instance at 1920x1080 at 1250Mhz and 1.200v requires like somewhere between 170 and 200 watt tdp to minimize
atikmdag.sys crash and bsod, though you must consider what your card could see by it's PCIe power spec and PCIe slot power.
a X16 slot will suck down 75 watts max, each 6-pin PCIe could see more than 150 watts, so that means my single 6 Pin PCIe power and 75 watt slot could see a max 225 watts in extreme cases. An extreme case is that of running beyond 1920x1080, like 2720x1530.
At 2720x1530 nothing higher than a 225 watt tdp with 1.200v at 1250Mhz will be stable since 2720x1530 with those settings will not allow 1250Mhz to be stable. At 2720x1530 only at 1240Mhz can 1.200v and 225 watt tdp be the absolute max stable for this card being XFX AMD Radeon R7 370 4GB.
It also depends on what resolution do you use most? 1920x1080, 2720x1530, 3840x2160? Each one needs equally as much power percentage as the increase of resolution requires. Gaming at 1280x720? What isn't stable at high res is perfectly stable at 720p, easily 1300Mhz no issue at 720p.
What one should know is it's not Too Hot until it's eating past 85C at 100% fan... That's when you will either buy Fujipoly Ultra Extreme from
www.fozencpu.com and bitch slap those high temps or Bend Over and take less volts/ watts/ clocks to keep the card in-line.
Just to analyze this a bit further my XFX R7 370 with no cooling on voltage regulators by stock saw VREG temps go sky high on stock 995Mhz/ 1.163v/ 105 watt tdp though among higher than I want core temps of 75C and well beyond.
It wasn't until I tore the card down that I modded the threaded mounting pegs so I could increase core pressure using different screws without "Stops". My screws stop when the peg meets the PCB. I added heatsinks to the voltage regulators but not the ram (not enough room and can't find 3 to 4mm tall heatsinks). By the way I'm not running Fujipoly Ultra Extreme on my XFX R7 370, I use Arctic Silver Ceramique 2 on the core diode, since I used all my Ultra Extreme on other cards... Which by the way my Fujpoly Ultra Extreme modded cards all see World Record Breaking performance. Thermal performance is directly related to fps and scores, so you can clock it all the way out and see LESS FPS. Once you use the material it'll allow the card to finally get out of it's "Cacoon" and "Stretch It's F*ing Legs Already".
If your out of cash and want damn good and decent performance for literally $6.60 shipped you get the best paste performance around and don't let anyone tell you otherwise this stuff is the bomb only if applied to the core diode only EXACTLY like this... (Use Only Fujipoly on Ram and Voltage Regulators).
http://www.ebay.com/itm/Arctic-Silv...016153?hash=item28113ebd19:g:26kAAOSwFMZWssYv
So if your running stock pads on voltage regulators and ram please don't touch the voltage or ram clocks because you'll kill it. All manufacturers use the crappiest thermal material seeing like maybe 2 to 3 watts per meter kelvin but the thicker the pad the higher the thermal resistance reducing that 2 to 3 to 0.1 to 0.75 watts per meter kelvin. Only 0.5 to 1.0mm pads should be used and Fujipoly ultra extreme 17 watts per meter kelvin and which when tightened decreases thermal resistance therefore making that 17 watts 100% all there all the time.
2 sheets should cover everything on the card... cut out to precisely the size of the core diode and apply to core diode, among ram modules, and one long large piece for the voltage regulator heatsink (add material to the heatsink surface and then install and tighten).
Core diode and Ram
http://www.frozencpu.com/products/1...rmal_Conductivity_170_WmK.html?tl=g8c487s1797
Voltage regulators
http://www.frozencpu.com/products/1...rmal_Conductivity_170_WmK.html?tl=g8c487s1797
Now I can run it at peak potential 24/7 365 without ever worrying about it burning up, because it will not.
Finding your right bios settings would be to open HWinfo64 and let it run for hours/ days/ weeks/ months to know what clocks/ volts/ temps/ fan speeds it sees on a regular basis Min/ Avg/ Max so you can determine what you feel needs to be done or what your concerned about.
Post a HWInfo64 screenshot of your gpu monitoring here after a hell of an intense gaming session so I can tell you about your card.
. It also booted without a problem (BIOS was already set to UEFI and legacy mode). What I've noticed though is MSI afterburner always shows the correct max voltage on load e.g. 1.25v, 1.3v, but GPUZ never shows 1.3v on max load, it always shows less e.g. 1.22v-1.27v, but mainly only shows 1.2v. So I am not sure wich reading is correct or is it maybe some vdroop in action or something else. But anyway, thanks for the feedback.
