VRAM, also known as graphics memory, is one of the most important parts that shape today’s graphics solutions. This component is used to store data and items that the GPU needs to be able to work efficiently, and without it it could not function in the way that we all know. Its evolution is key to facing new challenges, and Samsung has shown us the future in the short term with the presentation of GDDR6W memory.
This new type of memory is the South Korean company’s response to the challenges posed by virtual reality and the metaverse. To create a detailed world in a virtual environment and reproduce it in real time, you need a huge amount of graphics memory, and also a high computing capacity. With this in mind it is easy to understand the role played by GDDR6W memory, an important innovation that allows to double the capacity and increase the performance, and that is based on the encapsulation technology “Fan-Out Wafer-Level Packaging”, also known as FOWLP.
The attached image shows us the distribution that GDDR6W memory uses, and perfectly illustrates the differences it presents compared to traditional GDDR6 memory. As we can see, we have a stack that allows us to increase the density of chips from two to four, and we also have a reduction in height and occupied space. This new type of memory allows 36% thinner configurations, and despite his multilayer design offers the same thermal properties as GDDR6 memory.
With GDDR6W memory it is possible to create configurations with a bandwidth of up to 1.4 TB/s, and its maximum speed is 22 GHz (22Gbps). Taking into account that the GDDR6X memory, developed by Micron, used by the GeForce RTX 4090 and GeForce RTX 4080 already works at 21 GHz we can make it clear that what is really interesting about this new type of memory from Samsung lies in the encapsulation technology it uses, and in the fact that it can double the total capacity.
In theory, GDDR6W memory reduces the amount of I/O (input and output) to about one-eighth compared to HBM2E memory, making it unnecessary to use “microbumps” (micro gaps), and also Eliminates the need for a middle layer. Production of this new type of graphics memory will begin in the coming months, and it could certainly have a significant impact on next-generation graphics cards.
