NVIDIA has introduced the new Ampere architecture, as well as the first GPU based on it – the flagship GA100 for computing accelerators. Now, unofficial details have appeared on the internet about the characteristics of other Ampere GPUs, which will form the basis of NVIDIA GeForce RTX 30 series consumer video cards.

According to leaks and rumors, NVIDIA has finished preparing a digital design for the main Ampere graphics processor and is ready to send it. A large portion of the rumors have been centered around the RTX 3080 and RTX 3080 Ti, leaks about other graphics cards in the upcoming line have been accumulating also. How likely are we to see the Nvidia GeForce RTX 3080 and RTX 3000 line this year?

We are just hardly any days away now from the RTX 3000’s supposed release date, so Etechwall thought to write all that we know and have heard so far about these next-generation RTX GPUs.

Let’s discuss the leaked specifications, price, and other details related to GeForce RTX 3080 and 3080Ti.

Nvidia 3080

An unknown NVIDIA Ampere GPU, probably the RTX 3080, has been seen in 3DMark Time Spy, revealing its performance.

Although the benchmark could be any of the RTX 3080, 3080 Ti / Super, or RTX 3090. There is reason to believe that it is the RTX 3080, the gaming performance shown here puts the flagship RTX 2080 Ti far behind. Nvidia may launch a very successful RTX 3000 series if these results are true.

The benchmark posted by @_rogame shows that the NVIDIA Ampere GPU scored an impressive 18257 points in the Time Spy graphics test. The RTX 2080 Ti FE scores around 13,939 points, making the Ampere GPU roughly 31% faster. However, we see that it lags behind the overclocked RTX 2080 Ti KNIGPING. The question is, Is it the RTX 3080, RTX 3080Ti / Super or the RTX 3090?

Leaked Specification Details The leaks and rumors report that the card will include the GA102-200-KD-A1 SKU. This SKU will include 4352 CUDA cores, the same as the RTX 2080 Ti that will be organized in a total of 68 SMs. The card is apparently going to include 20 GB of memory that is likewise going to be GDDR6X, the 7nm manufacturing process, running at 19 Gbps over a 320-bit bus interface with a data transfer rate of 760 GB/s. The switch to GA102 from TU104 on the RTX 2080 may be a fascinating one.


Source : Chiphell Forum

Over the main few days of June, the first affirmed image design configuration came on the discussion forum Chiphell, which you can see below. The photo shows the front and back of the affirmed card, including the extremely significant RTX 3080 moniker. The design is with fans on the two sides of the affirmed RTX 3080.

Aside from its unusual two-sided fan design, the image does not give a lot of data. We can not see the sides of the graphics card, for instance, it is extremely unlikely of determining what sort of display outputs it has or what sort of power connector it requires.

Obviously, it’s absolutely impossible to know whether this is the final design of the RTX 3080 until Nvidia officially disclose it themselves.

Leaked Price Details

The cost for Nvidia GeForce RTX 2080 (2944 CUDA cores, 8GB RAM GDDR6, 14 Gbps Memory Speed, 448GB/s Memory Bandwidth, 215W Power), the price was around $799. Etechwall thinks that the RTX 3080 will cost somewhere in the region of $800, too.

Nvidia 3080 Ti

NVIDIA is moving with small steps with the Ampere series of cards, we expect a big jump in performance (improved clock speeds, core count and of course IPC).

Leaked Specification Details

Over the past few months, we have been getting news about the RTX 3080 Ti, but it seems we already have its main specs. The data suggests the increase in performance compared to its predecessor: the RTX 2080 Ti. So let’s see what this flagship wears.

It seems that it is an image taken quickly with the intention of not being caught. The user who leaked this information is CyberPunkCat.

The RTX 3080 Ti shows the GA102-400-A1 core in its specifications, including 5376 CUDAs. It would follow a 7nm process, the “boost” frequency would be 2,200 MHz, it would equip 12 GB of GDDR6 memory manufactured by Samsung and its 320W TDP.

If we take these specs for reference and compare them to the RTX 2080 Ti, you will see that GA102 – 84 SMs / 5376 CUDA cores / 12GB GDDR6 / 384-bit bus – 40% faster than RTX 2080 Ti. (Source : KittyCorgi on Twitter) But again the question is the same, Is it the RTX 3080, RTX 3080Ti / Super or the RTX 3090 ?

Leaked Price Details

The cost for Nvidia GeForce RTX 2080Ti (4352 CUDA cores, 11GB RAM GDDR6, 14 Gbps Memory Speed, 616GB/s Memory Bandwidth, 250W Power), the price was around $1199. As such, Etechwall would not be surprised if the RTX 3080 Ti’s price (or whatever it’s going to be called) is close to $1200 when it eventually comes out.

Latest Launch Update

So, the old rumors and leaks are done. Let’s take a look at the latest update about the launch.

NVIDIA seems to be preparing everything for a big announcement that probably has a lot to do with its new Ampere graphics, the GeForce RTX 3000.

Through Nvidia’s Twitter, the hardware company has published what they refer to as #UltimateCountdown, the last countdown, along with an image that has subsequently appeared, to wait until August 31 to have new news about the brand.

The rumor of the launch of the GeForce RTX 3080 in September. It is quite logical to think that NVIDIA will finally present its Ampere-based GeForce RTX 3000 graphics that are expected to arrive this year and that, in fact, it is rumored that they will come out on September 17 with its two top models, the RTX 3080 Ti and the RTX 3080. Later, the GeForce RTX 3070 would arrive, which would be launched in October and before the end of the year, in November, the RTX 3060 would go on sale as the entry-level range of this new generation of graphics for this 2020.

If you have any thoughts, put down your comments. Stay Tuned for more such articles !

More Technical Reading if you did not understand any term :

1. NVIDIA Ampere Architecture

2. IPC and GPUs

3. What Is TDP for CPUs and GPUs

4. GPU Programming and Streaming Multiprocessors

5. CPU Count/CPU Core Count

Biplab Das
mobomotion.tech@gmail.com
My name is Biplab Das. I’m a writer, Blogger, Youtuber and full time IT support engineer whose childhood obsession with science fiction never quite faded. A quarter-century later, the technology that I coveted as a kid is woven into the fabric of everyday life. I’ve spent the past years to learn these technologies, I recently published a book on computer science fundamentals. People say smartphones are boring these days, but I think everyone is beginning to take this wonderful technology marvel for granted.

Leave a Reply

Your email address will not be published. Required fields are marked *