Home » its GDDR6X will be the same as in the RTX 30

its GDDR6X will be the same as in the RTX 30

We have rumors again from a reliable source like kopite7kimi, which kicks off strong today by talking about the new specifications of 2 future NVIDIA graphics cards, the RTX 4090 and the RTX 4070. This leak mentions that the RTX 4090 will have 24 GB of GDDR6X memory with a speed of 21 Gbpsleading to consumption of 600W. On the other hand, the RTX4070 it will be 12 GB of GDDR6 memory with a speed of 18 Gbps and a consumption of 300W. That would leave us with some graphs high-end that matches current ones in VRAMleading to a Slower VRAM provided that. Why will the RTX 40 keep the same GDDR6X?

By analyzing the rest of the data, the NVIDIA RTX 4090 It would consist of an AD102-300 chip and supposedly offer a big performance boost in exchange for higher power consumption and using PCIe 5.0 to power the 600W of consumption. where we stay disappointed it is in the aspect of VRAM, which is kept in 24 GB and his memory GDDR6Xwho has a speed slower than expected of a new generation of graphics. And it is that if we compare an RTX 3090 Ti and the future RTX 4090 we have 24 GB of VRAM and 21 Gbps in memory for both.

The NVIDIA RTX 40 and its GDDR6X: they deflate before their launch

According to NVIDIA RTX-4070it will be an AD104-400 chip with “a fairly moderate consumption” of 300W, considering that a large increase in performance is expected in the future RTX 40 series. However, here we find several advantages over the current generation and it is that it has 12 GB versus 8 GB of the RTX 3070both GDDR6, where the new generation wins again, 18 Gbit/s versus 14 Gbit/s.

While we know that bandwidth is extremely necessary for NVIDIA, we don’t know the reasons why the company decided not to significantly increase this bandwidth. However, if we analyze previous leaks, Kopite explains that NVIDIA CEO Huang Jensen designed these RTX 40 graphics around a “complex interconnected design”, separating the cache into parts. This system would be based on the interconnection between the “new” calculation units (YOU) and texture processing (TPC) from NVIDIA to improve the performance of the first and thus help reduce the bottleneck caused by GDDR6X bandwidth and GPU bus.

Besides, less speed in VRAM equals lower power consumption and temperaturesTherefore, if the latter is true, they would add benefits to future NVIDIA RTX 4000 graphics that are already going to have issues in these sections.

GOING THROUGH: Kopite7kimi

About the author


Add Comment

Click here to post a comment

Your email address will not be published. Required fields are marked *