Samsung's new GDDR6W memory technology promises doubled bandwidth and capacities over GDDR6
Faster memory using 3D stacking - 3D innovation comes to GDDR6
Published: 29th November 2022 | Source: Samsung |
Samsung delivers staggering memory performance with their new double capacity GDDR6W modules
Samsung has today revealed their new GDDR6W memory type, a new kind of GDDR6 memory that utilised Samsung's Fan-Out Wafer-Level Packaging (FOWLP) technology. Samsung calls this new memory type their "next-generation graphics DRAM technology", as it doubles the capacity and bandwidth that GDDR6 memory can offer, mostly because it is effectively two standard GDDR6 memory chips on a single package.
GDDR6X can offer users two times as much GDDR6 memory per module than standard GDDR6 modules and can offer users a 2x bandwidth increase through its doubled I/O design. With GDDR6W, I/O has been doubled from x32 to x64, and Samsung has stated that their current GDDR6W modules can deliver per pin bandwidths of 22 Gbps. That's a lot of bandwidth.
Developing 'GDDR6W' Graphics Memory, with Doubled Capacity and Performance Based on the Cutting-edge Fan-Out Wafer-Level Packaging (FOWLP) Technology
High performance, high capacity and high bandwidth memory solutions are helping bring the virtual realm to a closer match with reality. To meet this growing market demand, Samsung Electronics has developed GDDR6W (x64): the industry’s first next-generation graphics DRAM technology.
GDDR6W builds on Samsung’s GDDR6 (x32) products by introducing a Fan-Out Wafer-Level Packaging (FOWLP) technology, drastically increasing memory bandwidth and capacity.
Since its launch, GDDR6 has already seen significant improvements. Last July, Samsung developed a 24Gbps GDDR6 memory, the industry’s fastest graphics DRAM. GDDR6W doubles that bandwidth (performance) and capacity while remaining the identical size of GDDR6. Thanks to the unchanged footprint, new memory chips can easily be put into the same production processes customers have used for GDDR6, with the use of the FOWLP construction and stacking technology, cutting manufacturing time and costs.
As shown in the picture below, since it can be equipped with twice as many memory chips in an identical size package, the graphic DRAM capacity has increased from 16Gb to 32Gb, while bandwidth and the number of I/Os has doubled from 32 to 64. In other words, the area required for memory has been reduced 50% compared to previous models.
Using their Fan-Out Wafer-Level Packaging (FOWLP) technology, Samsung has managed to create DRAM modules that double the capacity and bandwidth of GDDR6 while reducing package thickness and improving heat dissipation.
The primary advantage of GDDR6W memory is that it offers more bandwidth and higher memory capacities per chip. This is very desirable for chips that require high bandwidth levels, and applications where PCB space comes at a premium.
Generally, the size of a package increases as more chips are stacked. But there are physical factors that limit the maximum height of a package. What’s more, though stacking chips increases capacity, there is a trade-off in heat dissipation and performance. In order to overcome these trade-offs, we’ve applied our FOWLP technology to GDDR6W.
FOWLP technology directly mounts memory die on a silicon wafer, instead of a PCB. In doing so, RDL (Re-distribution layer) technology is applied, enabling much finer wiring patterns. Additionally, as there’s no PCB involved, it reduces the thickness of the package and improves heat dissipation.
One of the largest advantages to GDDR6X is that it offers more memory capacity and bandwidth per unit of PCB area, but the fact that it uses a x64 I/O complicates matter. GDDR6W effectively acts as two GDDR6 chips on a single package, which means that it needs to connect to twice as many places on a GPU's memory controller. Don't expect GPU manufacturers to replace GDDR6 chips with GDDR6W chips to double their bandwidth. For starters, GPU makers would need GPUs with memory controllers that are two times larger.
To put things another way, a single GDDR6 chip needs 32-bits from a GDDR6 memory controller to connect to a graphics card. A GDDR6W chip needs a 64-bit connection, effectively allowing it to take the place of two GDDR6 memory chips. This complicates PCB design, but allows for the creation of GPUs that have their memory allocated PCB space halved.
The height of the FOWLP-based GDDR6W is 0.7mm – 36% slimmer than the previous package with a height of 1.1mm. And despite the chip being multi-layered, it still offers the same thermal properties and performance as the existing GDDR6. Unlike GDDR6, however, the bandwidth of the FOWLP-based GDDR6W can be doubled thanks to the expanded I/O per single package. Packaging refers to the process of cutting fabricated wafers into semiconductor shapes or connecting wires. In the industry, this is known as a 'back-end process.' While the semiconductor industry has continuously developed towards scaling circuits as much as possible during the front-end process, packaging technology is becoming more and more important as the industry approaches the physical limits of chip sizes limits. That’s why Samsung is using its 3D IC package technology in GDDR6W, creating a single package by stacking a variety of chips in a wafer state. This is one of many innovations planned to make advanced packaging for GDDR6W faster and more efficient.
With GDDR6W memory, Samsung can rival the bandwidth levels offered by HBM2e memory using GDDR6 technology. With a 512-bit memory bus, Samsung's GDDR6W memory can achieve performance levels that are close to a 4096-bit HBM2e setup. Samsung expects designs that use GDDR6W to be cheaper than equivalent HBM2e products, a factor that could make GDDR6W a popular addition for newly designed products.
The newly developed GDDR6W technology can support HBM-level bandwidth at a system level. HBM2E has a system-level bandwidth of 1.6TB/s based on 4K system-level I/O and a 3.2Gpbs transmission rate per pin. GDDR6W, on the other hand, can produce a bandwidth of 1.4TB/s based on 512 system-level I/O and a transmission rate of 22Gpbs per pin. Furthermore, since GDDR6W reduces the number of I/O to about 1/8 compared with using HBM2E, it removes the necessity of using microbumps. That makes it more cost-effective without the need for an interposer layer.
Samsung's GDDR6W technology is an interesting new addition to the memory market, and a clear showcase of how 3D packaging technologies can enable major steps forward in most areas of the semiconductor market. It remains to be seen if future gaming GPUs will utilise this technology, but as gaming workloads become more bandwidth heavy, there are some clear upsides to using Samsung's latest memory technology.
With their latest generations of graphics cards, both AMD and Nvidia have been increasing cache sizes and investing in new technologies like AMD's Infinity Cache to work around the bandwidth limitations of modern GPU workloads. These limitations can be tackled more directly with faster memory, and Samsung's GDDR6W memory is one potential way to increase overall memory bandwidth. Will we see next-generation GPUs with 512-bit memory buses and GDDR6W memory?