Visual China.
Text |leitechIn October 2022, NVIDIA released their latest 40 series graphics cards, of which the RTX4090's performance parameters far exceeded everyone's imagination, becoming the white moonlight in the minds of countless game lovers. But obviously, such a graphics card with full performance ** can't be close to the people, and the official price of 12999 yuan has been able to dissuade many people, not to mention that the non-public version of the graphics card has to add a few thousand yuan to get it.
ROG JD.com is officially self-operated***
However, the RTX 4080 and 4070 in the lower gear are regarded by many consumers as IQ detection cards because of serious parameter castration and overconfident pricing, so that for a long time, Xiaolei has no idea about the 40 series graphics cards, and even plans to wait until the 50 series graphics cards are launched.
Nvidia official)
In the early morning of January 9th, Nvidia officially released the RTX 40 Super series graphics cards at CES 2024, which are RTX 4080 Super, RTX 4070 Ti Super, RTX 4070 Super, and the prices are 8099 yuan, 6499 yuan and 4899 yuan respectively. But if nothing else, there are less than 10 months to go, and the RTX 50 series graphics card is about to meet us, at this time, is buying a 40 series graphics card considered a 49-year income?
Before the official ** three 40 series graphics cards are not worth buying, let's first briefly understand their parameters, and what are the improvements compared to before.
First up is the RTX 4080 Super, which has been upgraded from AD103-300 to AD103-400 in terms of core aspects, the number of CUDas has been increased from 9,728 to 10,240, and the core clock speed has been increased by 90MHz and 50MHz in terms of base and acceleration, respectively.
Nvidia official)
Interestingly, the RTX 4080 Super's memory parameters have been greatly improved, although the bit width and capacity of the 256-bit 16GB GDDR6X remain the same, but its equivalent frequency has been reduced from 224GHz has been increased to 23GHz, and the bandwidth has also been increased to 736GBS, which is even higher than the RTX 4090.
Nvidia official)
Officially, its performance is twice that of RTX 3080 Ti, combined with the performance of the previous 4080 is about 40% stronger than that of 3080 Ti, it does have an improvement, but the magnitude is not too big.
The second is the RTX 4070Ti Super, which Xiaolei is most concerned about this time, after all, it is a product with both TI and SUPER suffixes in the history of NVIDIA graphics cards, and it is also the only one of the three products that has replaced the core chip, the core has been upgraded from AD104-400 to AD103-275, the number of CUDA cores has increased from 7680 to 8448, and the video memory has increased from 192-bit 12GB GDDR6X to 256-bit 16GB GDDR6X, the bandwidth has also been increased to 672GB S, which is a huge increase.
Nvidia official)
However, according to the official data, the overall performance of RTX 4070Ti Super is 1 of RTX 3070Ti6 times or so, and according to the evaluation, the performance of the RTX 4070Ti is 1 of the RTX 3070TiIt's about 4 times, so the RTX 4070 Ti Super isn't a big boost.
Finally, it is the most accessible RTX 4070 Super, which is also the most conscientious one of the three graphics cards, first of all, its core chip has changed from AD104-250 to AD104-350, but CUDA cores have increased from 5888 to 7168, almost catching up with the level of 4070 Ti. The core reference clock has been increased by 60MHz, and the boost clock has remained unchanged at 2475MHz.
Nvidia official)
The official said that its performance surpassed RTX 3090, and this positioning is a little strange, but the corresponding evaluation has been done on the Internet before, saying that the performance of RTX 4070 is only about 20% stronger than RTX 3070, and it is no wonder that netizens joked that this card was an IQ detection card at that time, and the gap between RTX 3090 is far away.
Nvidia official)
This also shows in disguise that the RTX 4070 Super has improved far beyond our imagination, and it can even be said that its cost performance is much higher than today's RTX 4070TI, after all, the average price of RTX 4070TI on the market today is about 6000 yuan. The most important thing is that its video memory has been upgraded from 12GB 192bit to 16GB 256bit, which makes it easier to run 4K games.
According to the current summary of information, the three super graphics cards will be officially launched in China in the middle and late of this month, and the manufacturers that have been confirmed to be the first are: ASUS, Rainbow, Gainward, Yingchi, Gigabyte, Yingzhong, MSI, ZOTAC, Renaissance, Xiangsheng, Mingxuan and Weconex, and the pricing should also be relatively reasonable.
All in all, RTX 4070 Super is the most cost-effective of the three graphics cards, which is the level of increasing the volume and reducing the price in the real sense, while the other two are to increase the amount without increasing the price, if there are plans to install the machine in the near future, these three graphics cards are really worth starting.
Although these three graphics cards all have a certain cost performance, considering that the 50 series graphics cards will be released at the end of this year at the earliest, will it be a bit 49 years to buy the 40 series graphics cards that have just been released?Ray's answer is: no.
Nvidia official)
The reason is also very simple, according to Moore'Judging from the current leaked information of S Law is Dead, the upgrade of the 50 series graphics card is not large, and it can even be said that there is some toothpaste. After all, the Blackwell architecture GPUs of the 50 Series focus on energy efficiency rather than performance, so there will be no performance improvement of nearly 100% from the 30 Series to the 40 Series.
And considering the product positioning of the RTX40 series, Xiaolei guesses that the improvement of RTX's three people-friendly graphics cards may be minimal, and the focus is still on the two high-end graphics cards of RTX 5080 and RTX 5090. In other words, buying a 40-series super graphics card today may be exactly the same as the 50-series in terms of performance, but it may be cheaper in terms of **.
However, Nvidia's current position supports it to do so. On the one hand, data center revenue has completely surpassed that of gaming graphics cards, and it is clear that NVIDIA's focus has shifted to the data center, and the gamer's experience is not the primary consideration;On the other hand, AMD's graphics cards are really not powerful, rumors say that their next-generation RDNA 4 architecture will not have a flagship model, which cannot put pressure on NVIDIA, not to mention Intel, the migration of business focus and the reduction of competitive pressure, Nvidia naturally does not have to work hard to engage in R&D and upgrades as before, who doesn't like to make money lying down?
Nvidia official)
This is certainly not good news for us gamers, NVIDIA and AMD were originally one of the must-have options for gamers, and now AMD no longer pays attention to the high-end market, and NVIDIA is dedicated to the AI market, which will inevitably make gamers, especially civilian gamers, a little dissatisfied.
However, the only thing to be thankful for is that the current Nvidia has not forgotten gamers, and a number of AI technologies announced at this press conference can be converted to games, bringing consumers a more extreme gaming experience. But will NVIDIA in the era of 60-series, 70-series, and 80-series graphics cards still care about the gaming market as much as it does today?No one knows.