One among NVIDIA’s secret weapons when it got here to the efficiency of its present graphics playing cards was the adoption of a proprietary Micron commonplace for VRAM. Nicely, evidently the GDDR6X will repeat within the RTX 40. This time a lot quicker and with higher capability. For the reason that second era of this reminiscence has simply been formally introduced. What’s going to carry us once more?
Video reminiscence or VRAM is essential for the efficiency of a graphics card, which is why NVIDIA opted for GDDR6X for its high-end RTX 30, on account of the truth that they’ll present a lot larger bandwidth than GDDR6 typical. Nicely, evidently we will take into account Micron’s VRAM for unique use to accompany NVIDIA’s most superior GPUs as successful.
And it’s that the corporate specialised within the design and manufacture of reminiscences has simply introduced the second era of stated reminiscence. Let’s see, due to this fact, how the GDDR6X will likely be within the NVIDIA RTX 40 and what new efficiency and capability it brings us in comparison with the present one. Will we see AMD lastly undertake this sort of reminiscence in its future RX 7000 or will it proceed to be unique to the inexperienced emblem model? Who is aware of, for the second we guess to see it solely within the subsequent era of the inexperienced model, however we could possibly be in for a shock.
That is the GDDR6X for RTX 40: 2 GB per chip and 24 Gbps velocity
If there’s something that has been extremely criticized concerning the GDDR6X through the NVIDIA RTX 30 era, it’s the truth that its capability per reminiscence chip has been 1 GB for a very long time. So it was vital to put twice as many chips, in shared information bus mode, in an effort to have very excessive reminiscence capacities. Which doubles the price of reminiscence on the graphics card. Nonetheless, that GDDR6X exists with 2GB capability per chip just isn’t a novelty, as is the case with the RTX 3090 Ti.
It’s the improve in bandwidth the place the GDDR6X takes all of the drive, since when it turns into 24Gbps We’re speaking about right now’s hypothetical RTX 3090 with a 384-bit bus that might go from having a bandwidth of 1008 GB/s to having a bandwidth of 1152GB/s. In any case, we preserve questioning if this will likely be sufficient to feed the as much as 144 shader cores that this beast of a graphics card can have.
Nor can we overlook that this enchancment has been made doable by the rise in energy consumed, because of the brand new 12+4-pin connector. Because of which the graphics playing cards will not have a consumption restrict of 375 W at 675 W. On the identical time, the query that assails us is to know if for the mid-range of the RTX 40 we may even see using the GDDR6X this time , though we anticipate it at switch speeds decrease than that 24 Gbps.