NVIDIA today officially released the upgraded version of the RTX 40 SUPER series of graphics cards, including RTX 4070 SUPER, RTX 4070 Ti SUPER, and RTX 4080 SUPER.

The core chip of RTX 4070 is still AD104, but its number has changed from RTX 4070 AD104-250 to AD104-350 .

The number of CUDA cores in this graphics card increased from 5888 to 7168, an increase of 21.7%, which is the largest number and the largest increase among the three models.

The core base frequency has been slightly increased by 60 MHz to 1980MHz, while the boost frequency remains at 2475MHz .

The graphics memory specifications remain unchanged, still 192-bit 12GB GDDR6X, with an equivalent frequency of 21GHz and a bandwidth of 504GB/s .

Due to the substantial increase in core size, the overall power consumption increased from 200W to 220W , an increase of only 10%, which is the only increase in power consumption among the three models.

Officials claim that the RTX 4070 SUPER uses only a fraction of the power consumption of the RTX 3090 to achieve performance beyond that of the RTX 3090. If DLSS 3 is considered, the performance advantage will reach 1.5 times.

The evaluation of the FE public version and the standard frequency version of this graphics card will be lifted on January 16, while the high-frequency version will be lifted on the 17th and will be fully launched. The starting price is 4,899 yuan, which is 100 yuan higher than the initial price of RTX 4070. Although the US dollar price has not actually changed, this small gap may be caused by factors such as exchange rates.

This graphics card is the first time in history that the Ti and SUPER names appear on the same product at the same time. The core number has been upgraded from RTX 4070 Ti AD104-400 to AD103-275, which is the only case among the three models that the core chip has been replaced.

The number of CUDA cores increased from 7680 to 8448 , an increase of exactly 10%.

The core frequency has also been slightly increased by 30MHz to 2340MHz , while the acceleration frequency remains unchanged at 2610MHz.

The most significant change is the video memory, which is upgraded from 192-bit 12GB GDDR6X to 256-bit 16GB GDDR6X. The frequency remains at 21GHz, so the bandwidth increases significantly to 672GB/s.

Power consumption remains unchanged at 285W .

RTX 4070 Ti SUPER is aimed at 2K or even 4K games. Officials claim that compared to RTX 3070 Ti, the performance is improved by up to 1.6 times. At the same time, after enabling DLSS 3, the performance can be improved by up to 2.5 times .

The evaluation of the MSRP standard frequency version of this graphics card will be lifted on January 23 (there is no public version), and the overclocked version will be lifted on the 24th and will be fully launched. The starting price is 6,499 yuan, which is exactly the same as the initial price of RTX 4070 Ti.

The core number of this graphics card has changed from RTX 4080 AD103-300 to AD103-400 , and the number of CUDA cores has increased from 9728 to 10240, an increase of 5% .

The increase in core frequency is considerable, with the base frequency and acceleration frequency increased by 90MHz and 50MHz respectively, reaching 2295MHz and 2550MHz .

The graphics memory specifications are more powerful. The bit width and capacity of 256-bit 16GB GDDR6X remain unchanged, but the equivalent frequency is increased from 22.4GHz to 23GHz, setting a record for the RTX 40 series, even exceeding the 21GHz of the RTX 4090, and the bandwidth has therefore reached 736GB/s .

In terms of computing power, this graphics card provides up to 52 Shader TFLOPS, 121 RT TFLOPS, and 836 AI TOPS.

Power consumption remains unchanged at 320W .

RTX 4080 SUPER is suitable for 4K ray tracing games. Its performance is 1.4 times higher than that of RTX 3080 Ti. When frame generation and DLSS 3 technology are enabled, the performance advantage can reach 2 times .

The evaluation of the FE public version and MSRP standard frequency version of this graphics card will be lifted on January 30, while the overclocked version will be lifted and fully listed on the 31st. The starting price is 8,099 yuan, which is 1,400 yuan cheaper than the initial price of RTX 4080.

It will be launched in the middle and late this month. The first brands include: ASUS, Colorful, Gengsheng, GALAXY, Gigabyte, Inzon, MSI, Zotac, Renaissance, ASL, Maxxuan, and Yeston.

After the new card is released, the RTX 4080 and RTX 4070 Ti will be delisted immediately, while the RTX 4070 will continue to be sold.

Affected by new products and market news, as of the close of U.S. stocks on January 8, Nvidia (NASDAQ: NVDA) stock price surged 6.43%. It once again set a record high during the session . Overnight, the market value increased by approximately 560 billion yuan. In the past year, Nvidia's stock price has risen by more than 230%.