Nvidia has announced today that they will be refreshing their Turing line-up with the new 2060, 2070 and 2080 Super GPUs that will be hitting store shelves soon.

While I myself was super (a-ha) wrong about the teaser that was shown a few short months back in Computex on what it could be, this announcement came hot off the heels of AMD launching their own competitor to the Turing line-up that will also be available sometime this month.

More importantly, Nvidia hosted a short announcement session detailing what most of the rumour mill had churned out, although there was no mention of the gaming bundle in our conversation.

Still, we had a chance to speak to Jeff Yen, one of Nvidia’s representatives about the new products and software that will be available in the near future.

Future Turing product stack

As the rumors indicated, the 2070 and 2080 will be phased out and eventually replaced by the super variants announced today. Like the move we saw with Pascal vs Vega with the 1070ti replacing the 1070 at the price segment, Nvidia looks to do the same with a half-generation refresh of their GPUs (more on that later).

One interesting thing to note is that Nvidia plans to keep the RTX 2060 in the line-up despite only being $50 cheaper than the next tier in the stack; it remains to be seen if there was any reason for Nvidia to do so as any advantage or disadvantage in terms of performance will certainly cannibalize sales of the other. Perhaps there is a price drop for the RTX 2060 in the works but for now, it just seems like an odd choice to say the least.

 Below is the product stack that Nvidia will have the second half of this year (prices in USD):

  • RTX2080Ti – $999
  • RTX 2080 Super – $699
  • RTX 2080 [EOL] – $699
  • RTX 2070 Super – $499
  • RTX 2070 [EOL] – $499
  • RTX 2060 Super – $399
  • RTX 2060 – $349

Pricing and release announcement

The pricing of the cards can be seen below along with their release dates that will be available, at least in the US.

Name Release date Price (USD)
GeForce RTX 2080 SUPER    7/23    $699 
GeForce RTX 2070 SUPER    7/9 $499 
GeForce RTX 2060 SUPER    7/9    $399 

As for local availability, we of course didn’t really get a concrete answer due to the variability between regions but for Singapore, it is usually about 2 weeks after the official launch before the shipment arrives and the cards are sold.

Since these prices reflect the same kind of release prices we see for Pascal cards, expect them to cost about $1100, $800 and $600 respectively locally.

Performance

While the presenter didn’t give out much info detailing exact performance figures, there was a ballpark figure quoted when one of the attendees asked this particular question:

Q: Can you give us a performance comparison between the 2070 and 2070 super?

“Performance figures are from 10-22%. If you average it out it would be [a difference] in the teens”

Jeff Yen, Nvidia Representitive

Now this definitely puts AMD in a tough situation for the value proposition that the RX5700 XT is going for, being about just as fast as the current RTX 2070 (and a little more) going by AMD’s numbers. If the super variant were even at least 10% faster than the current 2070 by conservative measures, $50 doesn’t seem that unjustified for the relative performance increase.

It seems that AMD’s new mid-range flagship is now sandwiched as the lesser performing card without the meaningful corresponding decrease in price. Of course, all these speculations are not very meaningful until we can get concrete results from independent parties or reviewers.

Given the historically higher margins by AIBs on Nvidia products, we might once again see custom variants coming in very close at MSRP for certain territories which will perhaps offset the $50 pricing advantage on the RX5700 XT as far as custom cards are concerned.

Short Q&A

Nvidia opened a short Q&A session in our conference call following the announcement and the media had the floor to ask any questions regarding the products. I, of course, asked the best questions undoubtedly.

Frameview software

For the uninitiated like myself, Nvidia has apparently been at work at developing a new software suite for monitoring statistics ala MSI afterburner. This was also news to me as I haven’t heard of this till now.

Q:Will frameview work with all APIs? Will there be an overlay (to show statistics)?

“Yes, from all that I have tested. I’m not so sure about DX11 [In regards to having the overlay]. So on DX9 and 10 you won’t [see it]. Vulkan will not, yet. The idea is that we are adding more support to the overlays and [along] with the functionalities; stuff like that.”

Jeff Yen, Nvidia Representitive

A commentator also commented that overlay support works for DX11 so that’s nice. Hopefully they can bring support for all APIs so that you can see all the stats “live” as with MSIAfterburner. Although he is quick to point out that all statistics are still logged into the log file.

Q:Will we see Frameview out of beta soon? Will there be a public beta? Will there be any performance impact using the software?

“we basically don’t see any performance impact [when using the software] Feel free to validate this with your own testing. Basically, if you use our old FCAT, it’s really complicated stuff.

…There will be a [public] beta soon.”

Jeff Yen, Nvidia Representitive

Q:Will frameview also work with AMD GPUs?

“Yes, but certain [metrics] will not be reported such as total board power (TBP) as it is not being reported in the API”

Jeff Yen, Nvidia Representitive

Looks like Nvidia isn’t tying their software down on a vendor level which is good to hear.

Technical and Misc. questions

Q: Are the super versions considered a half generation refresh or more akin to the “Ti” versions of these cards? Are there any architectural improvements for the Super cards?

“Kind of, it not unlike what we have in the previous generation(s). No “Ti” is not going away, its just how we decide to name it this time. [Its] still the same architecture. It’s our new [and] improved product

Jeff Yen, Nvidia Representitive

[On whether there is any architectural improvement] we didn’t really make any specific changes in the architecture itself, more of the configuration of the chip. It’s more of the [making the changes from] lessons we learned in the past 10 months and finetuning our craft. In some instances, we were able to get more out of the chip.”

Jeff Yen, Nvidia Representitive

Seems evident to me that these Super cards will probably not have a “Ti” variant as they are basically the equivalent for this generation of refreshes.

He also pointed out that these should be considered different products from the RTX line-up and its not just Nvidia “one upping” their own customers with a better product months after release. Although to that point I can see where both sides are coming from; We’ve seen Nvidia always pull out the better priced and sometimes better performing card with regards to the x80Ti series basically being on par (or beating) the Titan variants of the cards just a few months after release.

Q: Any comments on why the launch coincides with the launch of AMD’s RX5700 series?

“Quite frankly, it is just [our own] timing. If you look at our own history, our product [refresh] generally come about a year after launch traditionally (its been 10 months). I wish we were able to kneejerk a product out [this soon]. This product has been in development for months, so I think It was just a coincidence.

…[the] E3 [period] is just a good timeframe for both our companies [to announce products]”

Jeff Yen, Nvidia Representitive

It seems to me that releasing them in the same month is too much of a coincidence as far as I’m concerned. If Nvidia not revealing this announcement either at Computex or E3 is anything to go by, I’d say that Nvidia was working overtime to get this sooner through the door to deliver the first strike like we saw with the 1070ti.

Q:Will you be working with Asrock on Nvidia cards?

“Good question. I’m not sure about individual AIBs. We worked with them in the past and they are now largely working with AMD so…”

Jeff Yen, Nvidia Representitive

Seems like a really large maybe if we do see something coming from the new AIB. Given that Asrock’s latest “Phantom Gaming” cards are more targeted towards the mainstream market, perhaps Asrock would be interested in picking up at least up till the RTX 2070 super level of dies.

Thoughts

While it slipped my mind at the time, I’d love to hear what he has to say about the comments on FidelityFX, which is the post-processing competitor to the more intensive DLSS processing done for Nvidia partnered games.

 It is clear to me that both companies have taken a different approach to next-gen processing with Nvidia once again favouring proprietary technology and high-cost (in terms of resources) implementations while AMD is sticking to their open-source, low-cost implementations yet again.

With Crytek’s own version of Raytracing being demo’d on AMD hardware, Nvidia’s RTX may not hold that technology advantage for long. Unlike the tressFX vs Hairworks implementation back in 2012/2013 with neither company’s technology garnering much use from developers themselves, this time AMD has the gaming console advantage with both Sony and Microsoft’s console stated to be able to use Raytracing in some form on AMD hardware.

If there were any things that were missed following the official media release, I’ll be sure to update this article.

Conclusion

I was actually surprised that this teaser or rather, tidbit of information about these new products didn’t make it past the presentation at Computex because following this announcement, it will certainly put a damper on AMD’s pricing strategy at (again) slightly undercutting the competition while providing a small, but sizable performance lead.

Today was just a glimpse at the battle that will ensue as these Super cards take on Navi in a Hell-in-a-cell 1v1 ladder match this WRESTLEMANIAAAAAAA  battle for the mid-range crown with the RX5700 going up against the 2060 Super while the RX5700XT clashes it up with the 2070 Super.

I think that 2020 is shaping up to be even better than expected as we turn the clock on the decade to usher in the new consoles, high-end Navi and perhaps Turing on 7nm. I think the best has yet to come.

Stay tuned.

Chia is the horse-author from the far flung year of 2153. While not grazing on grass pastures or reviewing old time-y games and technology from the early 21st century pretending to not know what comes next (as to not disturb the space-time continuum), he can be seen exchanging vast quantities of Earth currency for parts needed to fix his damaged space ship.

Chia is the horse-author from the far flung year of 2153. While not grazing on grass pastures or reviewing old time-y games and technology from the early 21st century pretending to not know what comes next (as to not disturb the space-time continuum), he can be seen exchanging vast quantities of Earth currency for parts needed to fix his damaged space ship.