Wednesday, November 1st 2017
ZOTAC Avoids NVIDIA GTX 1070 Ti Factory Overclocking Restrictions via OC Profile
NVIDIA's decision to prohibit AIB partners from factory overclocking their GTX 1070 Ti graphics is a decision that may not sit well with hardware enthusiasts, but it's one that makes the most sense from a business perspective. As it is, it's expected that overclocked versions of the GTX 1070 Ti graphics cards, which have come to offer a more compelling NVIDIA option against AMD's Vega 56 graphics card, will offer performance that is by all accounts comparable to that offered by its more expensive (and older) GTX 1080 graphics cards. So NVIDIA is just ensuring that its introduction of the GTX 1070 Ti with its $459 MSRP doesn't cannibalize sales of their higher tier graphics card through higher out-of-the-box performance via factory overclocking.
This generates some issues for AIB partners, since now they have one less data point to help them differentiate their products from competitor's: performance. The only indirect claim to performance they now have is through efficiency of their cooling solutions, but that's a cloudy metric at best; as such, companies are finding ingenious ways to avoid going up against NVIDIA's restrictions, while offering some differentiating factors to their GPUs.
Case in point, ZOTAC will be including, via their FireStorm utility, the ability for GTX 1070 Ti users to select from a cadre of performance profiles that are pre-programmed on the software suite - as opposed to being automatically applied from factory. This means that ZOTAC can comply with NVIDIA's clock demands, while offering users an easily observable way of increasing performance. It's expected that more AIBs will follow suit with this implementation. One example profile that ZOTAC will make available for their GTX1070 Ti AMP! Extreme graphics card unlocks 140% power limit (252 W) and a +150 MHz overclocking on base core clocks, raising the boost clock to 1833 MHz.
Of course, there's always the question of whether or not this actually affects enthusiasts - or most TPU readers - at all. We generally tend towards acquiring the lowest priced option with the best cooling solution so as to allow us to achieve, through manual overclocking, performance levels that AIB partners would never be able to implement on volume production hardware.
Source:
Videocardz
Case in point, ZOTAC will be including, via their FireStorm utility, the ability for GTX 1070 Ti users to select from a cadre of performance profiles that are pre-programmed on the software suite - as opposed to being automatically applied from factory. This means that ZOTAC can comply with NVIDIA's clock demands, while offering users an easily observable way of increasing performance. It's expected that more AIBs will follow suit with this implementation. One example profile that ZOTAC will make available for their GTX1070 Ti AMP! Extreme graphics card unlocks 140% power limit (252 W) and a +150 MHz overclocking on base core clocks, raising the boost clock to 1833 MHz.
38 Comments on ZOTAC Avoids NVIDIA GTX 1070 Ti Factory Overclocking Restrictions via OC Profile
It's not 'magic' which allows for the extreme frequencies we see from the sub-ambient modders. ;)
Anyway, its not much different than AMD's limitations either. Aren't they are hard locked at "+50" or something? I haven't had a chance to play with Vega, just saw SS... But yeah, same thing.
BIOS modding does nothing , got it.
Here are some google links: www.google.com/search?q=GTX+1080+Ti+hard+mods&rlz=1C1CHBF_enUS767US767&oq=GTX+1080+Ti+hard+mods&aqs=chrome..69i57.3386j0j7&sourceid=chrome&ie=UTF-8
Boost or throttle - Nvidia doesn't sell you a 2100mhz card, they sell you a 16xx Mhz card with a stock performance that is clear to see, and with headroom to get out a bit more performance on your own risk. Intel: NO throttling until you hit critical temps. Throttling is a necessary mechanism to provide optimal performance, and it is only a negative thing when you are consistently seeing LOWER performance than what it said on the tin.
Last but not least, read this carefully, maybe you'll learn a thing or two
en.wikipedia.org/wiki/Leakage_(electronics)
"Leakage current is generally measured in microamperes. For a reverse-biased diode it is temperature sensitive. Leakage current must be carefully examined for applications that work in wide temperature ranges."
It's a sad reality that one can't mod pascal bioses. But I don't really think there's nothing that could give more MHz to core.
And why not let the cards run at anything higher than 70c while having the OC'ed GPU performance not be throttled like it used to be in earlier architectures? Having options matters, I'm frowning upon efficiency being forced on the GPU. But I guess the new architectures aren't able to handle high temps anymore like in the old days, the 14/16nm silicon tends to degrade pretty fast as evidenced by Ryzen chips going bad from having high temps on constantly.
Technically Nvidia could let Pascal run hotter but have even better performance, but they don't have to this time (I'm talking about how older GPU architectures used to get really hot, because they are pushing out all they have out of that particular architecture, but they do hit a wall at some point with thermals and performance, until for the sake of competition you have to change your strategy and opt for something new, now cards seem to have so much thermal headroom and yet it seems to be unused at this point) because AMD cards still have lower performance when compared in many games.
Looks like 70 Celsius max is becoming the norm of some sort for CPU/GPUs.
I'm just more impressed by a Fermi/Kepler card able to run at 100c and not thermal throttle the performance at all. Sure, you end up doing a pretty stupid thing, but it's still amazing nonetheless. Can Pascal do that? No.
Nobody will ever know what Pascal/Volta are truly capable of because of the thermal/voltage wall that has been put in place. Those cards could probably be stable at higher voltages, but I don't think there's a way to bypass the voltage limit. Nvidia is either hampering overclocking on purpose OR the silicon is just working as intended and there's simply no reason to go past the limits because you will kill the card (I would like to believe that, btw I am not savvy in electrical engineering, so feel free to correct me if I am completely wrong).
As Vya Domus said. Even modding the card won't get you what is potentially possible. They really hit the gold center with Pascal, more so than any GPU architecture in history I think. Now they have to milk it for all it is worth.
It's just Nvidia being themselves. Just that every time it happens I get irritated and start smashing the keyboard like it's 1999.
In the end it's the consumers that are voting with their wallets. And everyone just seems to love so far how things are going.
4GB total. 512MB of it is notably slower than the other 3.5GB. I said that already.