The RTX 4070 with GDDR6 memory (in place of GDDR6X) is launched today. Problems with the availability of GDDR6X chips have prompted Nvidia to propose a reliable solution from an availability point of view. While several rumors pointed to other changes for this RTX 4070, in the end, it will feature exactly the same technical specifications.
Discreet launch of RTX 4070 GDDR6
The GPU used will be the AD104-250/251-A1 with 5888 CUDA cores, 184 TMUs and 64 ROPs configured with 46 SM units. The graphics card will operate at a base speed of 1920 MHz, with a boost to 2475 MHz. The VRAM configuration will also remain the same, with a capacity of 12 GB running on a 192-bit interface, and this is where the main difference lies. According to early reports, the GGDR6s used will feature better characteristics than the standard models. Frequency is said to increase to 20 Gbps, compared with the usual 18 to 19 Gbps. It should be noted that GDDR6X offered 21 Gbps. Thus equipped, the memory of the “new” RTX 4070 would offer a bandwidth of 480 Gbps, slightly lower than the 504 Gbps of GDDR6X modules.
Moving on to performance, according to an initial leak on a battery of games (Alan Wake 2, Cyberpunk 2077, Metro Exodus and A Plague Tale: Requiem), the RTX 4070 GDDR6 achieves the same results as its GDDR6X version.
The arrival of this variant on the market is now a fact. In this story, Nvidia has not launched a Founders Edition version, and only green partners are supplying their usual variants. Nvidia claims that demand for the RTX 4070 is strong, and that the initial tension over GDDR6X was beginning to push prices upwards. Another important point is that the switch to GDDR6 will not, unfortunately, result in lower prices, since Nvidia’s official recommended price is exactly the same.
We’re obviously waiting to see what Nvidia’s magic recipe is for achieving the same performance with this card, without changing the GPU, frequencies or anything else.
