Nvidia Turing Product Reviews and Previews: (Super, TI, 2080, 2070, 2060, 1660, etc)

More detailed rumors on the Nvidia Super lineup:

NVIDIA GeForce RTX 20 ‘SUPER’ Lineup Allegedly Launching in July – Specs For RTX 2080 SUPER, RTX 2070 SUPER & RTX 2060 SUPER Leak Out

https://wccftech.com/nvidia-geforce-rtx-20-super-graphics-cards-specs-launch-leak

The top of the stack in the ‘SUPER’ lineup seems to be the GeForce RTX 2080 SUPER which is going to feature the full TU104 GPU die known as TU104-450. The chip will feature 3072 CUDA cores, 8GB of GDDR6 VRAM running at 16 Gbps and a 256-bit bus. The pricing is said to replace what the RTX 2080 currently costs while the original RTX 2080 would take a price drop to around 4000 (Chinese) Yen which is about $549-$600 US which should be a $100 US drop in cost and what we have been hearing for a while.

The GeForce RTX 2070 SUPER is said to feature pretty much the same specifications that we posted a few days ago which would be 2560 CUDA Cores and 8 GB (14 Gbps) memory. The specific chip is said to be TU104-410 GPU. The most interesting part is probably the GeForce RTX 2060 SUPER that is said to feature the full TU106-410 GPU.

The difference between the current RTX 2060 & the Super one will not only be the higher core count of 2176 CUDA Cores but it is also rumored to feature a higher 8 GB GDDR6 VRAM along a 256-bit bus instead of 6 GB GDDR6 VRAM along a 192-bit bus. It is possible with TU106 since the full GPU die does allow for 256-bit wide bus interface and 36 SMs. The RTX 2060 SUPER still won’t utilize the full 36 SMs that the TU106 die but rather utilize two less SM units (34).
 
Do you think they can propose that now because of better yields, or just because they really didn't need to before, and they do that now just to be talked about when Navi is coming (and possibly beating it perf/price wise)
 
The 5700XT by AMD own number is less than 6% faster, so IRL I'm expecting to be equal but with this:

D83mD3fX4AI7z-U.jpg:large

If this is true well...Not very appealing for AMD.
 
The 5700XT by AMD own number is less than 6% faster, so IRL I'm expecting to be equal but with this:

D83mD3fX4AI7z-U.jpg:large

If this is true well...Not very appealing for AMD.

This is super confusing. Why didn't NVIDIA just call these RTX 30x0 (Ti)?
 
Don't we get pissed off when they do rebrands with the same chips? Now you're asking for that?

Pure rebrands are annoying, but if they're new products with (even slightly) different specs, then yeah, I prefer it that way. If mean, is Super better than Ti? This is clear as mud. Not that big of a deal in the grand scheme of things, of course, but it's very strange, and very different from NVIDIA's traditional naming schemes.
 
The 5700XT by AMD own number is less than 6% faster, so IRL I'm expecting to be equal but with this:

D83mD3fX4AI7z-U.jpg:large

If this is true well...Not very appealing for AMD.


This is a better article showing what the new Super Lineup is. Also the RTX 2080 Super has only 8GB not 11GB of Vram:

https://wccftech.com/exclusive-nvidias-super-gpus-unleashing-monsters

Rumors had stated that only the 2060, 2070 and 2080 were going to get SUPER versions and they were going to be refreshes – let’s just say that’s not even half of the story. Ladies and gents, I can confirm that SUPER series is very much a new lineup in its own right and contains a new flagship: The RTX 2080 Ti SUPER (yes, even the 2080 Ti is getting a SUPER version). Oh and launch date? sooner than you might think, it’s going to be a SUPER summer after all.

NVIDIA’s SUPER Summer starts on 6/21

New chip: the NVIDIA RTX 2080 Ti SUPER exists and will be an absolute beast of a card
Turns out, there is an RTX 2080 Ti SUPER variant and its a brand new chip. It’s not an old part and it’s not a repurposed Quadro part (trust me, I asked). Oh, and its completely unlocked, meaning the AIBs can have 300 watts TDP on this thing if they wanted to – and knowing the insane custom designs that come out of AIBs, I wouldn’t be surprised to see this. I don’t have the exact specifications of the chip or the exact pricing yet but I am fairly confident this will have a higher core count as well as faster GDDR6 memory.

This is not the PN-A for OC version chip or the non-OC chip, its a completely new chip as I mentioned earlier and will have a new PN. Since it’s unlocked and will have higher TDP as well as higher core count we expect this to be much faster than the existing RTX 2080 Ti card. Everything we know so far points to this slotting in at roughly the same price point as the existing RTX 2080 Ti or slightly above. In other words, you are going to get a card that is much more powerful but for roughly the same price or slightly more. Unfortunately, however, you are going to have to wait a bit for this card as per my info, it will have a delayed launch.

The NVIDIA RTX 2080 SUPER graphics card will feature an unlocked 2080 Ti chip
The RTX 2080 SUPER graphics card, on the other hand, will feature a non-OC 2080 Ti chip that has been unlocked. Most probably it’s going to have a core count of somewhere between a vanilla 2080 Ti and vanilla 2080. I have been told that it will have 11 GB of vRAM and have some features turned off. Edit: Igor is reporting that this card will have 8GB of vRAM, not 11 GB. The reported specs for this card are 3072 CUDA Cores.

Here’s the thing, however, NVIDIA previously bundled VRMs and vRAM with locked chips (non-OC) and the VRMs provided weren’t of the absolute best quality. This is going to change with the SUPER lineup as NVIDIA is now selling the entire lineup as unlocked chips, allowing AIBs to put in much higher quality VRMs and providing higher sustained clock speeds and performance. Like before, the RTX 2080 SUPER graphics card is expected to slot roughly into the price bracket of the standard 2080 or slightly above.

New chip: the NVIDIA RTX 2070 Ti SUPER 8GB exists
The RTX 2070 Ti SUPER graphics card is a brand new chip. This will be positioned between the RTX 2080 SUPER and the RTX 2070 SUPER (*duh). The only thing I know about this is that it will feature 8GB of vRAM. I am also told that the RTX 2070 Ti SUPER will have a delayed launch and will not launch with the existing SUPER lineup.

The NVIDIA RTX 2070 SUPER 8GB graphics card will feature an unlocked 2080 chip
The NVIDIA RTX 2070 SUPER will feature an unlocked, non-OC variant of the 2080 chip and will have 8GB of vRAM. Once again, we expect this to have a core count of roughly halfway between the vanilla 2080 and 2070. The RTX 2070 SUPER will slot in roughly to the price point (or slightly above) of the existing 2070. EDIT: The reported core count for this card is 2560 CUDA Cores

The NVIDIA RTX 2060 SUPER 8GB will feature unlocked 2070 chip
The same goes for the RTX 2060 SUPER, which will feature an unlocked, non-OC 2070 chip and 8GB of vRAM. The core count is expected to be roughly halfway between the vanilla 2060 and 2070. This is an upgrade not only on the performance side but for vRAM as well because the original RTX 2060 only had 6GB of vRAM. EDIT: The reported core count for this card is 2176 CUDA Cores.
 
Looking at prices...the 1660 is at the same price or *cheaper* than the 590 while being better in every single metric...some rumors says Nvidia will lower the prices after the Super launch and even if they aren't true the prices for AMD cards have no sense at all. I was waiting for Navi to upgrade but now I'm just going for a 1660 SC ultra until I can get a GPU to play at 1440 at less than 300 bucks that was what I though Navi would be.
 
Re: RTX 2080 Ti SUPER / brand new chip...

Since bandwidth remains bottleneck for non-game use, and while HBM2 remains pricey, anyone else hoping Nvidia eventually restore that 512-bit bus, to really make use of 16 Gbit/s GDDR6 (900 GB/s)?

Now that might be worth calling "super".

</wishful_thinking>
 
Re: RTX 2080 Ti SUPER / brand new chip...

Since bandwidth remains bottleneck for non-game use, and while HBM2 remains pricey, anyone else hoping Nvidia eventually restore that 512-bit bus, to really make use of 16 Gbit/s GDDR6 (900 GB/s)?

Now that might be worth calling "super".

</wishful_thinking>

Given they have no competitor, I doubt they will do that soon.
 
The wccftech article is too ambitious, whole new chip for 2080TiS? 2080Ti chip going into 2080S and somehow having only 8GB RAM?

The VC article makes sense in 2080S having full chip, 2070S moving to a cut 2080 chip and 2060S getting the full bandwidth from the 2070 chip.
 
The VC article makes sense in 2080S having full chip, 2070S moving to a cut 2080 chip and 2060S getting the full bandwidth from the 2070 chip.

They also say it increases the SM count to 34 instead of 2070's 36.
With equal bandwidth and ~6% difference in compute power, if the 2060S comes at the 2060's price then it's going to drive the 2070 out of the market quite fast, and make the 5700 non-XT DOA.

Not that people didn't know already that AMD put themselves in a corner with their Navi release prices.
 
AMD can always adjust there prices, if and when those Super cards arrives to the market.
The Navi cards are coming to the shelves July 7th. nVidia is apparently going to announce the new line-up next Saturday, with a hard launch in mid-July (a single week after Navi).
If these prices and specs are true, both Navi cards will need to get their prices adjusted before even hitting the shelves if they want to avoid getting DOA'd by reviewers and customers alike.
Backtracking on the price before release just shows very poor planning on AMD's side.
 
The Navi cards are coming to the shelves July 7th. nVidia is apparently going to announce the new line-up next Saturday, with a hard launch in mid-July (a single week after Navi).
If these prices and specs are true, both Navi cards will need to get their prices adjusted before even hitting the shelves if they want to avoid getting DOA'd by reviewers and customers alike.
Backtracking on the price before release just shows very poor planning on AMD's side.
C’mon, it was a given that the introduction of the Navi cards would be followed by a period of pricing and product rejiggling until they all found positions that were aligned with their respective strategies and economic framework. Neither Nvidia or AMD are idiots, they know that this will happen, they just don’t know the specifics of how the other party will choose to play their cards. We’re about to find out over the next six months or so, most of us perfectly safe munching popcorn from our spectator seats.
 
EVGA is offering credit toward a GeForce RTX card for owners of its 700- and 900- series
Here's how it works: if you bought an EVGA GeForce GTX 770, 780, 780 Ti, 970, 980, or 980 Ti card, you can get a credit toward your purchase of a GeForce RTX 2060, 2070, or 2080 direct from EVGA. You have to still have the card, and you need to register it at EVGA's website. If you bought a recertified card, you are explicitly excluded from the promotion, sadly. Likewise, if you live outside of the USA (including Puerto Rico) or Canada, you aren't eligible for the trade-up.

Graphics card Credit amount
EVGA GeForce GTX 980 Ti $150 USD
EVGA GeForce GTX 980 $100 USD
EVGA GeForce GTX 970 $75 USD
EVGA GeForce GTX 780 Ti $100 USD
EVGA GeForce GTX 780 $75 USD
EVGA GeForce GTX 770 $50 USD
https://techreport.com/news/34655/e...rtx-card-for-owners-of-its-700-and-900-series
 
Multitude of Nvidia GeForce RTX Super graphics cards hit EEC
Twitter-based tech news hawk Komanchi started to share screenshots and links of EEC trade body shipping manifests, like this one from Asus, from Gigabyte, and from Palit.

The Gigabyte haul might be the most impressive. This well-regarded PC, laptops and components firm has signalled its intention to ship 21x RTX 2080 Super variants, 26x RTX 2070 Super variants, and 22x RTX 2060 Super cards. Those familiar with Gigabyte's wares won't be surprised to hear there will be Aorus cards in the mix, and variations with and without OC, Pro, Gaming, WindForce, White, and other permutations.
91ce26fa-bcf7-4f74-b58c-868d199b42a0.png
https://hexus.net/tech/news/graphics/132140-multitude-nvidia-geforce-rtx-super-graphics-cards-hit-eec/
 
Back
Top