Model.
GeForce RTX 3080.
RTX 3070, 3080, and 3090: Specs, Pricing, and Launch Dates.
GeForce RTX 3070.
.
GeForce RTX 3090.
Cuda Cores.
10496.
8704.
5888.
Base Clock.
1.4 GHz.
1.44 Ghz.
1.5 GHz.
Boost Clock.
1.7 GHz.
1.71 GHz.
1.73 GHz.
VRAM.
24GB GDDR6X, 19.5 Gbps.
10GB GDDR6X, 19Gbps.
8GB GDDR6, 16Gbps.
Memory Bandwidth.
935.8 GB/s.
760 GB/s.
512 GB/s.
Bus Width.
384-bit.
320-bit.
256-bit.
RT Cores.
82.
68.
46.
Tensor Cores.
328.
272.
184.
Ampere.
Ampere.
Architecture.
Ampere.
ROPs.
96.
88.
64.
* Graphics Card Power.
350W.
320W.
220W.
Suggested PSU.
750W.
750W.
650W.
Yes.
Custom Samsung 8nm.
Yes.
DirectX 12 Support.
Yes.
Yes.
Manufacturing.
Custom Samsung 8nm.
Yes.
Yes.
Yes.
Launch Date.
Yes.
Custom Samsung 8nm.
Yes.
Nvidia DLSS.
PCIe Gen 4.
09/24/2020.
09/17/2020.
10/2020.
That said, bear in mind that there will still be AIB partner models with custom coolers, with more standard designs..
A few more quick points. Nvidia once again revealed The Marbles during the night demo, only this time there was no rasterization; the Marble demonstration was totally path trace, all real-time, and the ray tracing was processed simultaneously with the shading. Nvidia displayed brand-new lighting effects, specifically aerial and spherical lighting. There was no pre baking at all in the demonstration, which means absolutely nothing was pre-rendered, lighting wasnt pre calculated, etc.
The RTX 3080 is being billed as the new flagship, while the RTX 3090 is being positioned more as a Titan-class card. The 3080 will be $700 and launch on September 17th, with 10GB of GDDR6X, and TFLOPS claims at 30, 58, and 238 (for shader, RT, and Tensor efficiency, respectively). Were waiting on more specifications beyond that. The RTX 3090 will have 24GB GDDR6X and is supposedly efficient in 8K video gaming with RTX on and DLSS made it possible for.
Almost right away, the essential takeaway from the discussion was how Nvidia is positioning the RTX 3000-series. Nvidia is actually attempting to sway Pascal owners here who elected to avoid last-gen. Turing barley gets a reference in the 3000-series statement, as it was a slow burn that never truly took off the way Nvidia intended..
Turing is standard 1.0 x, while Ampere is showing about a 60% to 100% uplift with Blender cycles, Autodesk Arnold, and V-Ray, and its showing gains in the 50%- and-up variety for gaming applications, including Minecraft RTX that we just recently tested. Now, that said, thats with ray tracing allowed– if RT doesnt matter to you, the gains may be smaller sized for rasterization just. Well have to see.
Cooling & & Power Design.
Positionally, NVIDIA is attempting hard to bury its 1080 Ti: “To my Pascal buddies, its safe to update now,” stated Jensen Huang during the keynote. Once again, Turing barely got a mention in this sector, and was in reality gotten rid of from some of the charts when NVIDIA was attempting to make the case for 1080 Ti upgrades to the 3000 series.
$ 700.
The Latency Discussion.
Ray Tracing Demo & & Broadcast.
Editorial, Host: Steve BurkeEditorial, Reporting: Eric HamiltonVideo: Keegan Gallick.
Completing the basics are some surface-level specifications for Ampere: 2nd gen RTX, 28B transistors for the densest Ampere GPU, a customized 8nm manufacturing process from Samsung, and Ampere performing 2 shader calculations per clock (versus 1 per clock on Turing). These are all points well require to go back over when we have explicit details on the Ampere architecture..
In addition, whatever is declared to be dynamic. The characteristics here point towards things like diffused global lighting and volumetric rendering in real time, to name a few technologies..
For dimensions and case compatibility, were seeing dimensions of 12.3″ (length) and 5.4″ (width) for the FE RTX 3090, which is the greatest card. Thats just partially bigger than GTX 1080 Ti models from AIB partners. The RTX 3090 will be of a 3-slot design, while the 3080 and 3070 must be 2-slot designs..
* Nvidia lists power specs as “Graphics Card Power,” which isnt necessarily the like TGP or TDP. This number might represent GPU-only or GPU + memory, but we will verify this in our evaluation. All numbers above are from NVIDIAs official materials and have actually not yet been separately confirmed.
In addition to the tough specs, were also seeing some official numbers that can put to rest the wild rumors that have been walking around concerning the both the size and power requirements of these cards. Nvidia is noting the Total Graphics Power for the RTX 3090, 3080, and 3070 at 350W, 220w, and 320w, respectively. Nvidia is recommending a system power of 750W for both the RTX 3090 and 3080, and 650W for the RTX 3070.
Nvidia is also noting 2x 8-pin adapters for both the RTX 3090 and 3080; meanwhile, the RTX 3070 is listed with 1x 8-pin. For Nvidia NVLink (SLI Ready), youll see the feature is supported on the RTX 3090; however, for both the RTX 3080 and 3070, Nvidia only lists a dash. One part of the latency discussion is certainly about pushing ever greater frames per second, however the other part is Nvidias brand-new application, Nvidia Reflex. With regards to Nvidia Reflex, Nvidia declares this improves response time and revealed this with Valorant, Destiny 2, and Fortnite.
This might have simply entirely changed the value and argument of closed-loop liquid coolers over CPU tower coolers, as you all of an unexpected have a CPU cooler that will have to handle a greater reliable ambient temperature level, thanks to the GPU stressful directly into it. This is an aspect of the cooler we mean to test extensively to see what effect this will have on the air vs liquid argument..
Nvidia is likewise noting 2x 8-pin connectors for both the RTX 3090 and 3080; on the other hand, the RTX 3070 is noted with 1x 8-pin. If youre wondering what occurred to the 12-pin adapter, that still exists– it can be seen in the renders. What is likely happening is that 2x 8-pin cable televisions will be adapted into a single 12-pin. Well know more when we can physically evaluate the cards..
Concerning the fans on the cooler, theres 2 fans, with one fan on the backside pulling air in through the front and another fan exhausting (pressing) the warmed air out the back. This style implies that air can be found in the underside of the GPU and will be discarded out the back of the card and into the CPU cooler tower/finstack (if you have one) in a standard ATX setup..
RTX is going to get a big push– and likely only become significantly more relevant in this generation– but for those who arent interested in ray tracing in games, or are playing video games that dont support the feature, rasterization efficiency is what we need to understand in order to recommend an upgrade..
Returning to the new specs, the RTX 3070 claims to be faster than the 2080 Ti at $500 with an October launch window. The card will come geared up with 8GB of GDDR6 (not 6X) and declares 20.4 shader TFLOPS, 40 RT TFLOPS, and 163 TFLOPS for Tensor performance, but we dont have further specs yet. And keep in mind, FLOP numbers are of restricted usefulness, especially within the context of marketing..
Then theres a set of heatpipes– it looks like 4 flat pipes in the render, but that may simply be art– which then round-out towards the end. After this, there are two separate fin stacks that look to be taller in size, bridged by the heatpipes in between. After that, its simply the shroud.
Nvidia is suggesting a system power of 750W for both the RTX 3090 and 3080, and 650W for the RTX 3070.
Youll see that both RT cores and Tensor cores have made generational shifts, with RT cores for Ampere being 2nd generation and Tensor cores being 3rd generation. We dont have any information on what this generation is altering about the core design, however Nvidia should be offering us with details in the next couple of days..
The most interesting aspect of this set of cards, from a screening viewpoint, is the brand-new cooler design. NVIDIA released a rendered explosion of it, showing a backplate that must ideally utilize thermal pads to leverage the area, then the PCB, which has actually been reduced in size for this generation..
Theres more conversation on our YouTube channel, where weve now produced one video on the launch specs and another on RTX 3000 customized cards.
It seems one of Nvidias next big pushes will be among latency. This push appears to be two-pronged. One part of the latency discussion is definitely about pushing ever higher frames per 2nd, but the other part is Nvidias new application, Nvidia Reflex. Both of these points tie into Nvidias push to bring 360Hz G-Sync displays to market..
$ 500.
MSRP.
Therefore far, we likewise understand that NVIDIA is declaring a 1.9 x efficiency/ watt enhancement for Ampere versus previous generations. For ray tracing performance, NVIDIA is dunking on Turing, as you can obtain from the following chart..
Beginning with the essentials: The RTX 3080 will release at the exact same price as the RTX 2080, at $700 MSRP from NVIDIAs website, and around September 17th for partner models. The RTX 3090 will go for $1500 on September 24th. The RTX 3070, meanwhile, will release for $500 in October..
Concerning the charts, well rapidly point out a few things. Of all, the gap was made to look bigger than it is due to the reality that the RTX 3000 series had ray tracing and DLSS made it possible for, whereas with the 10-series, DLSS isnt readily available– not a surprise to anyone. Requiring ray tracing to be on sort of muddies the water here and makes the 10-series look worse than it really is. What we do not know strongly is the rasterization relative efficiency between the 2 generations.
Completing the cooling conversation is a couple of points about the PCB design and VRM. The referral PCB is compact to fit the cooler, and will be running an 18-phase total design– a few of that is for memory, clearly. The VRM flanks the memory modules on either side, so we have a couple of suspicions about where hotspots will emerge. The memory modules are oriented so that 4 are left of the GPU, 4 are right of the GPU, 3 are above the GPU, and 1 is listed below it. We can also see where the VRM flanks the bottom and leading memory modules..
$ 1,500.
With regards to Nvidia Reflex, Nvidia claims this enhances response time and revealed this with Valorant, Destiny 2, and Fortnite. While Nvidia is being unclear about its terminology and metrics here, we think the basic essence is improving total system end-to-end latency. Nvidia Reflex is another point well have to review when we have more details, but its something Nvidia atleast believes will be a highlight for this generation..
The latency battleground isnt a brand-new one, as AMD has had similar innovations in the past however weaker marketing behind it. Latency becomes more vital as rasterization efficiency scales ever upward, and raw FPS becomes less appropriate..
Nvidia also introduced Nvidia Omniverse Machinima, which it says is a “RTX-powered mixer app for video game storytelling.” While telling stories with computer game isnt new, maybe its something brand-new to Nvidia. We d like to see Jensen Huangs response to Red vs. Blue. At any rate, Nvidias Omniverse Machiniam will include video game libraries and can do AI position simulations through a webcam to successfully do movement capture without motion capture devices, which is interesting. The demonstration showed impacts like fluids and particles, and theres supposed to be a beta in October.
In general, it reminded us of Ansel, where its a niche application that some people will use, rather than being a broad market application.
It appears RTX Voice has been rolled into Broadcast, and it will get a bigger push with the RTX 3000-series. Broadcast will provide innovations such as a green screen result, where it will presumably trace the user and their chair and change the background. These are going to be more streamer target utilities, however its something Nvidia plans to introduce more of.
With its new cooling design, Nvidia is staking claims to 55% more air flow, 3x quieter, and 30% more effective compared to previous designs. While we ought to be able to test the airflow claim, the 3x quieter claim will be challenging. Its difficult to quantify subjective multiples of volume. Theres also things like acoustic power (dB) and viewed sound to the human ear, and theyre not mutually congruent. As generally, well dive more into acoustic metrics and claims when we get the cards..
To that end, NVIDIA removed the 20 series from its charts and revealed the 1080 Ti vs. 3080 with ray tracing and DLSS– not precisely a fair comparison, and on its rather arbitrary chart of “typical perf across popular graphics-intensive games,” it states that the 1080 Ti is doing about 1440p over 60FPS while the 3080 RT is doing over 60FPS 4K. This is with DLSS on the 3080 and not the 1080 Ti, so keep that in mind. We wouldnt put too much stock in first-party charts just yet.
Additionally, theres a surprise caution moving down the specs list. For Nvidia NVLink (SLI Ready), youll discover the function is supported on the RTX 3090; nevertheless, for both the RTX 3080 and 3070, Nvidia only notes a dash. We presume this dash to imply no. This is fascinating, because Nvidia appears to be pressing multi-device readiness and communication through a bridge to the most expensive cards..
For at least the 3070 model, we noticed that the power adapter is positioned in the center of the card. This is a design move that Nvidia likewise pulled with the RTX 2060 FE, and it was a godawful design in regards to taking apart the card. We d hope that Nvidias mechanical engineers took this into represent this generation, however well see..
