Just doing some more testing tonight..I am thinking now that we have extreme cards, factory water blocks and custom water blocks on 980Ti's now, we should start to think about BIOS branches.... We've never needed a separate BIOS for AIR vs H2O with GM204 but there's a very different story with GM200!
980Ti is showing us that it likes to get REALLY hot, even with the BEST coolers.. They are screaming for a waterblock
. 1.237v in my opinion is the MOST voltage you should be sending to GM200 on AIR.
Yes I know some GPUs with a STOCK BIOS can hit 1.243v but with extreme tests @ 4K max settings we start to see that little bit of voltage make a HUGE difference in max temps.
I just did some tests on AIR and I'll give you the summary. This is (2) GPUs in SLI, and I am giving you the TOP card which is the HOTTEST:1.262v = 90C max temps, test aborted by me because I wasn't comfortable with this much heat!
1.243v = 88C max temps, test aborted by me because I wasn't comfortable with this much heat!
1.237v = 86C max temps, test was allowed to complete
So this doesn't mean everyone will get these temps, it means that with that voltage, and @4K with max settings, if you push the GPU hard enough you CAN get those temps on AIR. I will also add that I have my fans maxxed out, supplemental fans installed and my case is open with a room that is kept cool so I'm really surprised how hot these GPUs can get when pushed hard.
The other point to highlight is that never once did I get any crashes or artifacts despite the high temps... there is clearly a THERMAL RESTRICTION
preventing us from pushing GM200 further (where this was NOT the case with GM204, both AIR and WATER could run 1.281v
I think it is clear that for those that have a water block on GM200, it isn't just for show anymore
H2O has a clear advantage over those on AIR with regard to higher voltages with GM200 and therefore should benefit from a BIOS optimized to take advantage of the extra cooling.
We also have some learning to do with these new Gigabyte extreme cards. The LN2 mode and the extra 6-pin, are they bound together? Can the "non-LN2" BIOS access/utilize that additional 6-pin PCI-e also? For those on water that have no intention of using LN2, we might be able to use the LN2 BIOS and change it for H2O so that water blocks can use the additional 6-pin PCI-e also.. of course we'll need to see if it is of really any benefit to do so as well..
Some ideas floating
Flash at your own risk of course!
If anyone want to give this a try.. The first H2O BIOS specific to the GTX 980Ti G1? (not the extreme, that will come later)
So how many out there on H20
have already found your max overclock using 1.237v and want to see if additional voltage will allow more headroom on 980Ti?
H2O_V1.1BETAF4_DP.zip 152k .zip file
☞ This is a DP
BIOS! Don't flash if you are using DD! (But you can copy over all the settings to your DD BIOS
(1.275v in BIOS and will report 1.274v under load)
☞ Default BOOST speed increased to 1506.5Mhz
(vs 1455Mhz with V4BETAF4 [AIR] BIOS).
As always, Voltage and Power % sliders maxxed in your O/C app = 1506Mhz boost @ +0Mhz GPU.
☞ Disabled fan settings H2O ONLY!!!
☞ Power and everything else is the same as V4BETAF4, 450W (700W+ PSU recommended), No perfcaps, etcSo the main thing here
is that when GM200 is on AIR
, we can sometimes run into thermal limitations without have run into crashes or artifacts
. Purely high temps only
resulting in overclocks having to be lowered. Anyone with this GPU on H2O should gain an additional 20C or so of thermal headroom which should allow additional voltage to be compensated... but .. There's NO GUARANTEE this BIOS would result in any additional overclocking headroom
, it is a TEST to see if removing thermal limitations on GM200 will allow higher voltage which can theoretically allow additional overclocking headroom..
We won't know unless we try right