NVIDIA locks overclocking in GeForce GTX 1070 Ti

Introduction To GeForce GTX 1070 Ti

GTX 1070 Ti: Coming Soon

NVIDIA is rumored to be mere weeks away from launching its new GeForce GTX 1070 Ti graphics card, something that will completely bury AMD's luke-warm-but-runs-so-hot Radeon RX Vega 56, but the new rumors are very interesting.

Our friends at eTeknix are reporting that the GTX 1070 Ti "can't be overclocked", with the site saying: "An industry insider told eTeknix that the GTX 1070 Ti can't be overclocked. In addition, we've been informed that NVIDIA will not release any reference cards and all AIB partner cards will be locked to the same speeds. However, price and aesthetics are expected to differ across partner cards". If this rumor is true, it would be a big change in direction for NVIDIA and its board partners.

EXPreview also reported similar news, where they said: "NVIDIA will be fully unified market GTX 1070 Ti frequency, that is, regardless of which you buy a GTX 1070 Ti their frequency will be fixed at 1607MHz-1683MHz , if you want to be stronger performance only Can be manual overclocking, for non-hardware enthusiasts are not friendly, or graphics card manufacturers have to surprise, such as get a software overclocking to avoid restrictions on NVIDIA".

NVIDIA is reportedly launching the GeForce GTX 1070 Ti with 2432 CUDA cores, up from the 1920 CUDA cores on GTX 1070, while both cards will have 8GB of GDDR5 RAM and a 256-bit memory bus. The GTX 1070 Ti however, will bump up the TMU count to 152, up from 120 TMUs on the GTX 1070, while both cards have 64 ROPs. GTX 1070 Ti will push 7.8 TFLOPs of compute performance, up significantly from the 5.8 TFLOPs on the standard GTX 1070. GTX 1070 Ti will also require 180W, while just 150W is required on the GTX 1070.

Sneaky Moves By NVIDIA?

Thoughts: NVIDIA has been much more aggressive lately, with the company unveiling their new Founders Edition cards with the launch of the GTX 10 series in May 2016, a launch that included the GTX 1080 and GTX 1070 and then saw the release of the GTX 1080 Ti earlier this year. The Founders Edition cards were exclusives for months, before AIB partners began pushing out custom GTX 1080 and GTX 1070 designs.

With the release of the GTX 1080 Ti, NVIDIA pushed their 'GEFORCE GTX' branding onto ALL partner cards. This isn't something the partners have an option with, they need to put the GEFORCE GTX branding on it, which sometimes looks really tacky.

I'm personally not a fan of it, but there are one or two custom GTX 1080 Ti cards with smaller GEFORCE GTX branding on it - like the ASUS cards with stickers that can be peeled off and thrown away. But the ones with HUGE branding of GEFORCE GTX are annoying, so it'll be interesting to see if we see GEFORCE GTX branding on the NEW GTX 1070 Ti cards.

Now with the launch of the GTX 1070 Ti, a no-overclocking placeholder on these cards is strange. NVIDIA first went exclusive with Founders Edition with GTX 10 series launch, then GEFORCE GTX branding with the GTX 1080 Ti, and now no overclocking on the GTX 1070 Ti. These are the sorts of things NVIDIA can do when their main competitor is DOWN AND OUT. AMD is in dire trouble with its Radeon division, with Raja Koduri stepping away from the company for a while, and AMD's entire Radeon marketing team reportedly being rolled into AMD's control, like their CPU division.

If AMD had kicked NVIDIA in the shins with their Radeon RX Vega graphics cards, NVIDIA would have to think twice about pissing consumers and companies off with Founders Edition exclusivity, hardcore branding of GEFORCE GTX on all enthusiasts cards (like we didn't know they were GeForce cards in the first place, guys), and now overclocking restrictions on GTX 1070 Ti. Ugh. Where will this end? Just give me good cards and let companies do WHATEVER the hell they want with them. That would be nice.

Overclocking Blocked On GTX 1070 Ti - Part 1

NVIDIA has reportedly locked overclocking the GPU clocks on their upcoming GeForce GTX 1070 Ti, and while it definitely doesn't sit well with me as an enthusiast (NVIDIA locking their AIB partners from overclocking the GPU, making it nearly irrelevant which GTX 1070 Ti you buy doesn't incentive AIB partners to make GTX 1070 Tis), I understand it from a business perspective.

First, the business perspective. NVIDIA doesn't want to cull its GeForce GTX 1080, which would happen in a few ways. First, NVIDIA launched the GeForce GTX 1080 as its flagship GTX 10 series card at the time, so launching a GTX 1070 Ti that is already nearly as good as the GTX 1080 would hurt GTX 1080 sales. Even in stock form, the GTX 1070 Ti could take close blows to the performance of the GTX 1080, so overclocking it by 10-15% would definitely begin not just matching, but beating the GTX 1080.

NVIDIA can't allow a GTX 1070 Ti to beat the GTX 1080 as it means the lower number card beats the higher number card, 1070 beating 1080 doesn't sound good. This is the type of confusion it would create, and it would from a business perspective make NVIDIA look weak by releasing a tweaked GTX 1070 that beats the GTX 1080. This is a direct response to AMD's surprise with the Radeon RX Vega 56, which just so happens to beat the GTX 1060 with 9Gbps of GDDR5, and the GTX 1070. So... GTX 1070 Ti is built to completely murder the RX Vega 56, and it will.

I've spoken to a few industry sources that seem to think AMD knows its Radeon RX Vega 56 is no longer a threat, at all, to NVIDIA once their GTX 1070 Ti is released. Secondly, AMD has its hands tied behind its back with production of Vega GPUs in general. HBM2 yields aren't great right now, which is hurting AMD, but production of Vega is hurting the company more. Instability from within RTG is also worrying, with AMD CEO Lisa Su having to come in and clean up the various messes left behind by the Radeon team. There's no bias here, this is just what I'm hearing, seeing, and reporting after years of being involved in these industries.

AMD is now a sitting duck until a Vega refresh... something that can't happen until at least Feb/March 2018. NVIDIA doesn't need to release anything between now and then, all they need from here is a soft refresh of Pascal - nothing drastic, just voltage increases and maybe some tweaks to the memory used - higher clocks on GDDR5X and the introduction of GDDR6. AMD, even with the help of the super-advanced HBM2 memory, couldn't touch NVIDIA's flagship card from nearly 18 months ago, the GTX 1080.

Now consider a GTX 1070 Ti is on the horizon, and AMD couldn't beat the GTX 1080. NVIDIA is continuing to add cards to their GTX 10 series pile, which now includes:

• GTX 1050
• GTX 1050 Ti
• GTX 1060
• GTX 1060 9Gbps
• GTX 1070
• GTX 1070 Ti
• GTX 1080
• GTX 1080 Ti

AMD has the following Radeon RX Vega cards:

• RX Vega 56
• RX Vega 64
• RX Vega 64 LCE

See the difference? AMD won't have another card in that stack for a while, and even then - it'll be competing with what, the GTX 1050? No one cares about that in the real-world, although they do represent a large chunk of sales in large markets with internet cafes, etc. But, as a company striving for the very best in GPU technology... AMD aren't in the position to fight GTX 1070 Ti. So NVIDIA can get away with locking the overclocking on GTX 1070 Ti without a word from AMD, who doesn't have enough cards on the shelves or the money to burn by lowering the price of RX Vega 56 right now. Rock, hard place.

Overclocking Blocked On GTX 1070 Ti - Part 2

Final Thoughts

Now, the enthusiast part of me is less mad, more disappointed. NVIDIA has been a company that, in my eyes, has always had a strong leading position. NVIDIA continues to innovate and push into markets that it might not be all that successful in, but when it does be successful, it doubles down. The company did it with their original GeForce 256 graphics card (which is where my online handle 'anthony256' came from, a little TIL for you), and they did it again most recently with the Pascal GPU architecture.

The Pascal GPU architecture was a tour de force in how to build an incredibly powerful, but also incredibly powerful graphics card. The GTX 1080 at the time offered some great performance that rivaled the already powerful GTX 980 Ti, and did so using less power. But even then, NVIDIA had restrictions on voltage running through Pascal, with volt-modding and sub-zero cooling breaking those limits, the cards were utter BEASTS.

We saw even harder restrictions placed on the GP102 with the GTX 1080 Ti, because when that card is truly unleashed (volt modding and LN2) it is an absolute technological freaking marvel. I can't imagine what a fully unleashed GP100 GPU @ 3GHz would do with GDDR6 @ 16Gbps...compared to the GPU @ 2.2GHz max, and GDDR5X maxing out at 12Gbps or so. This is just Pascal, too. No refresh, no Volta.