Based on the slides, it seems that the RTX 4070 will retail at an SRP of US$599 (~RM2649), which is the same as the RTX 3070 Ti when it debuted back in 2021. So, when the GPU officially becomes available in Malaysia, it’s clear that AIB partners will be selling their own custom-cooled versions of the card at higher-than-SRP prices. As for availability, the card is also expected to be making its debut this month and more specifically, sometime at the end of next week. As for the seemingly confirmed specifications of the RTX 4070, the acquired slide shows that the card is obviously going be built around Ada Lovelace and more specifically, the AD104 GPU. There’s also the expected fare of 3rd generation RT Cores and 4th generation Tensor Cores, as well as the expected support for DLSS3 that provide access to things like Frame Generation and Optical Flow Accelerator. There’s also support for AV1 encoding, which has been a staple since the launch of the RTX 40 Series. What is surprising to see though, and if the slide is to be believed, is that the RTX 4070 Ti is set to ship out with 12GB GDDR6X graphics memory, and not the initially rumoured 8GB of memory. For that matter, it is equally surprising that NVIDIA is generously giving this mid-range GPU access to the more powerful GDDR6X memory, and not limiting it to the GDDR6 standard, as it did with its RTX 30 Series card. [Image: NVIDIA]In addition to the 12GB of memory, it appears that the RTX 4070 is also getting a huge bump in its L2 Cache. At 36MB, the cache is nine times greater than the 4MB on the RTX 3070 Ti, but as a trade-off, the memory bandwidth seems to be lower at 504GB/s, versus the 608GB/s of the RTX 3070 Ti. Lastly, there is the power consumption of the RTX 4070. Compared to the RTX 3070 Ti, the card supposedly draw a lower average TDP of 186W, versus the 240W of its predecessor. For that matter, its power consumption is even lower than the non-Ti RTX 3070. (Source: Videocardz)
