Originally Posted by DisposableHero7
here is my first impression of the AORUS GTX 1080 TI Xtreme.
I read alot about this card on several gaming/hardware sites and it seemd like a realy good performer. High clockrates and a good cooling solution, keeping the card at low temps and still silent at this. Finaly a friend of mine got the very same card and was realy happy with it, so the decision was made.
So i ordered mine 2 days ago via express delivery, got it today and was very stoked to try it out. Installation was easy, even though it is bigger than my already bulky EVGA GTX 1080 FTW. Luckily space is not an issue in my Coolermaster HAF-X case, lots of it in there
After a clean, fresh install of the drivers (removed the old ones with DDU and got the recent ones from NVIDIA - 381.89), it was time to start up some benchmarks to look what i would be able to achieve from this new beast. For reference i ran through several benchmarks with my old card just the day before.
The first benchmark i tried was "Superposition", the new fancy looking one from UNIGINE, the guys who made the well known "Heaven" and "Valley" benchmarks. And i was immediatly amazed: Out of the box the card ran @2015 MHz, rocking about 46 FPS average on the "1080p Extreme" preset of the Superposition benchmark. That is 50% more fps than my GTX 1080 FTW could push.
This was very exciting ... for a few moments. After about 30 seconds, the clocks went down to about 1900 MHz. Temps where at ~62°C by then, when the benchmark started to suddenly go into slow motion and studdering. When the temps hit 69°C, the benchmak crashed and the grafics driver was reset. Ok, this can happen, so i fired it up again. Now it already started with 65°C and only ~1900MHz, frames where down to ~40fps at start and started to drop furher already. Again studdering occured and when the GPU reached 69°C it crashed again
To make sure it wasn't a problem with the relativly new benchmark, i tried Heaven, Valley and 3DMark11 ... all with the same result. Starting off cold, the card pushed some serious frames, but in a matter of about 1 minute, clocks dropped to ~1900, temps rose to ~70°C and the applications crashed.
Opening my case, i could see the fans spinning up when the card got warmer, so it was not a total failure of the fans per se. Besides that, temperatures of 70°C shouldn't bother a Pascal GP102 chip at all...
At this point i started googleing if others might have similar problems. Was i missing something etc.? And i found quite alot of people on different forums reporting similar problems ... great. I then installed the Gigabyte AORUS tool that enables the OC mode and also offers monitoring of the card. Starting it up, it was in "Gaming" mode, which is supposed to be the standard mode for this card. It offers two other presets and a manual OC setting that lets you individualy set different parameters of the card. I didn't care about the manual mode and stuck to the offered presets, OC-mode beeing the one boasting the impressive clock speeds that Gigabyte is advertising this card with. For testers, i switched the presets back and forth and then set it to "gaming"(normal) and tried the benchmarks again.
Same result as before, the benchmark starts off impressive, but very quickly the temps rose and the frames dropped until the bench crashed at 69°C again. When i set the card into OC-mode, any 3D-apllication, be it benchmarks or games, crash immediatly
The only mode that keeps the card from crashing is the "Silent" (underclocked) mode that has the Founders Edition clock speeds. Strangly though, the card goes up to 79°C in this mode, besides having considerably lower clock speeds. After reading some more in the Gigabyte forums, i was guided to a bios update for the card - the newest F4 Bios supposedly adresses stability & cooling issues.
I got the F4 bios update (quite comfortable .exe) and applied it my card, without much luck. Somehow i could run the Superposition benchmark once on "gaming" mode, but other benchmarks and games still crashed after 1-2 minutes. Another run of Superposition benchmark then didn't make it through again.
When trying different games in the "silent" mode with FE clocks, the card performed as or sometimes even slower than my 1080 FTW and with temps between 75-80°C, depending on the game. This is very odd in itself, because even at FE clocks, the TI should be faster than a slighty overclocked regular 1080 like the EVGA FTW. This is not quite what i was expecting from a 880€ card.
Then i decided to do what had to be done ... RMA the card. I had already idled a bit, chatting with friends and filling out the RMA form of my retailer, then powered down my PC and was about to pull out the card from my case, when i seriously burned the tips of my index-, middle- and ringfinger on the backplate. It still was boiling hot after about 15 minutes idling in windows with just discord and a webformular opened. The card was WAY beyond the 45°C the Gigabyte monitoring tool showed just before i powered down the PC.
There is something seriously wrong with the cards bios, tempsensors or the cooling presets in general and it seems to affect alot of AORUS 1080 TI & Xtreme users - but than again not all, since my friend doesn't have any of the issues i am having. QA seems to be on vacation @ Gigabyte ....
I am now back on my old 1080 FTW writing this. Since monday is a national holiday here, i'll send the card back in to my retailer on tuesday and just hope that i'll get a working replacement without this issues. But this ofcourse means another 1-2 weeks of waiting. If the 2nd card behaves like that as well and doesn't run the "OC"-Setting reasonably cool and 100% stable, i'll go for a refund and switch the manufacturer ...