NVIDIA officially announces GeForce RTX 30 series, the top model enables world's first 8K60 fps gaming

Chip maker NVIDIA has announced three types of next generation graphics card ' GeForce RTX 30 ' series. The RTX 30 series is the second-generation RTX series that adopts NVIDIA's

Ampere architecture and Samsung 8nm process. The high-end model RTX 3090 is the world's first to support 8K resolution and 60fps gaming.

GeForce RTX 30 Series Graphics Card Overview-NVIDIA

NVIDIA Announces the GeForce RTX 30 Series: Ampere For Gaming, Starting With RTX 3080 & RTX 3090

The following table summarizes the specifications of the announced GeForce RTX 30 series 3 types and the high-end model RTX 2080 Ti of the previous generation RTX 20.

NVIDIA GeForce RTX 3090 RTX 3080 RTX 3070 RTX 2080 Ti
architecture Ampere Ampere Ampere Turing
Manufacturing process Samsung 8nm Samsung 8nm Samsung 8nm TSMC 12nm(FFN)
Number of CUDA cores 10496 8704 5888 4352
VRAM 24 GB 10 GB 8 GB 11 GB
GPU core clock (boost) 1.7 GHz 1.71 Ghz 1.73 Ghz 1.54 Ghz
Memory clock 19.5 Gbps/GDDR6X 19 Gbps/GDDR6X 16 Gbps/GDDR6 14 Gbps/GDDR6
Memory bus width 384 bit 320 bit 256 bit 352 bit
Thermal design power (TDP) 350 W 320 W 220 W 250 W
Number of transistors 28 billion 28 billion unknown 18.6 billion
Single-precision floating-point number 35.7 TFLOPS 29.8 TFLOPS 20.4 TFLOPS 13.4 TFLOPS
Half-precision floating point number (FP16)
(Tensor core)
RT core performance 69 TFLOPS 58 TFLOPS 40 TFLOPS unknown
Release date September 24, 2020 September 17, 2020 October 2020 September 20, 2018
(Only yen conversion)
(Approximately 159,000 yen)
(About 74,000 yen)
(Approximately 53,000 yen)
(Approximately 106,000 yen)

The RTX 30 series uses the Ampere architecture, which was just announced in May 2020. The adoption of the Ampere architecture has significantly improved power efficiency, and 'performance per power consumption has jumped 1.9 times,' NVIDIA says. However, AnandTech, a technology news site, said, 'The 1.9-fold number is a little exaggerated, so it may be about 1.5-fold at most.' On the other hand, the power consumption itself is also rising, and the thermal design power of the top model RTX 3090 is 350 W, the RTX 3080 is 320 W, finally exceeding the 300 W level.

Although the GPU has not been announced, it is likely that GA 102 is adopted from the photograph of the GA 102 die

leaked from Hong Kong PC hardware manufacturer Zotac .

Furthermore, unlike the RTX 20 series, which used TSMC's 12nm process, the RTX 30 series uses Samsung 8nm process. In fact, AnandTech from the photo of the die that has been leaked, with 500mm 2 little over a large area of GA102 is estimated, has pointed out that is even smaller than the TU102 of 754mm 2. As a result, the transistor density has increased significantly, and the number of transistors in GA102 has reached 28 billion.

The number of

CUDA cores, which is the shader core of the GPU, is 10496 cores for the top model RTX 3090, 8704 cores for RTX 3080, and 5888 cores for RTX 3070. In addition, the computing power of the RT core , which can be called an innovation of RTX, is up to 69T FLOPS , and the computing power of the Tensor core in charge of AI is up to 285TFLOPS. In addition, the top model RTX 3090 is the world's first graphics card compatible with 8K and 60fps.

The release date and price of the RTX 3090 is $1,499 on September 20, 2020, RTX 3080 is $699 on September 17, 2020, and RTX 3070 is $499 during October 2020. However, the price has risen a little more in Japan, the RTX 3090 is 229,800 yen, the RTX 3080 is 109,800 yen, the RTX 3070 is 79,980 yen, and it is purchased as a personal import even including shipping and tariffs. It may be cheaper.

You can see the state of the announcement event from the following movie.

NVIDIA GeForce RTX 30 Series | Official Launch Event-YouTube

in Hardware,   Video, Posted by log1i_yk