Recently, it has been reported that Nvidia as well as AMD’s mainstream graphics cards currently on sale have received varying degrees of price reductions, both down to the recommended retail price, and even some models have been priced below the recommended retail price.
According to the report, the average minimum price of NVIDIA and AMD graphics cards has previously been reduced to below the suggested retail price. This time, the media statistics are including RX 6800, 6800 XT and RTX 3080 10GB, RTX 3070 and RTX 3060 Ti, including the “major models”.
As of August 21, these “major models” are priced at 97% of the manufacturer’s suggested retail price. In China, NVIDIA’s high-end models have seen significant price cuts, but RTX 3060 and RTX 3060Ti and other mainstream graphics cards have seen lower price cuts, and some high-end branded graphics cards are still slightly above the suggested retail price.
The main reason for this price cut, in addition to the inventory pressure of graphics card manufacturers, is also partly due to the upcoming RTX 40 series graphics cards to make way.
Previously there was a lot of information about the RTX 40 series graphics card specifications, including RTX 4090 was exposed that its performance will be about double compared to RTX 3090. The video card field of information kopite7kimi previously said that the RTX 4090 in DMark Time SpyExtreme scores more than 19,000 points.
Looking back at the RTX 3090 and RTX 3090 Ti, their previous scores were around 10,440 and 11,400 respectively, and the RTX 4090 was 82% and 66% higher than them. In other models of RTX 3090 graphics cards, RTX 4090 lead even up to nearly 100%.
Of course, such extreme performance must be built on top of higher frequencies and power consumption, and the RTX 4090 is no exception.
RTX 4090 is expected to have 16,384 CUDA cores, 52% more than RTX 3090Ti, but the performance increase is even greater.
It is reported that the RTX 4090 will consume 450W of power, and the power performance at the limit of performance is not yet known. In the face of 450W power consumption of graphics cards at every turn, both the graphics card cooling or chassis cooling will usher in a huge challenge. In particular, the power supply is not an ordinary power supply can completely eat.
Previously, the foreign media also broke the RTX 40 series of some specifications information, RTX 4090 GPU model AD102-300-A1, 16384FP32 core, GPU frequency up to 2750MHz, memory for 384bit 21Gbps 24G GDDR6X, power consumption 450W.
RTX 4080 GPU model AD103-300-A1, 10240 FP32 core, 256bit 21Gbps 16G GDDR6X memory, power consumption 420W.
RTX 4070GPU model AD104-275-Kx-A1, 7168 FP32 core, 160bit 18Gbps 10G GDDR6 memory, power consumption of 300W.
And RTX 4090 AD102 core is expected to use TSMC 4nm process, frequency 2 ~ 3GHz, single floating point accuracy in the 80 ~ 100T or so. Of course, no surprise, there is RTX 4090 Ti or TITAN in the future.
And RTX 4070Ti will probably become a new generation of sweet card, the news that this “RTX 4070 Ti” will use the full version of the AD104 core, open all 7680 CUDA cores, 160 ROP units, 48MB L2 cache, compared to RTX 3070Ti core number 25% increase in the number of cores and 12 times more L2 cache.
But its core frequency is not known, but in TSMC 4nm process support will not be too low. Video memory with 192-bit GDDR6X, capacity 12GB, equivalent frequency 21GHz, bandwidth is 504GB / s.
Public PCB design number PG141-SKU331, the entire card power consumption limit open to 400W, the performance is expected to easily reach the RTX 3090Ti level. In contrast, the news that the first lineup of RTX 4070 using AD104 GPU, CUDA core of 7168, 10GB of video memory, power consumption of 300W.
You can see that Nvidia is simply squeezing the toothpaste out of the RTX 40 series, and it seems that the second half of this graphics card launch will bring great surprises to users.
