Samsung begins mass production of 16-GB GDDR6″

Samsung Electronics announced that it has begun mass producing the industry’s first to 16-GB of GDDR6 memory. Memory GDDR6 promises to be in demand for the release of gaming devices, graphics adapters, automotive electronics, networking equipment and systems with artificial intelligence. The company believes in a “timely” release of a new type of memory, which is widely demanded by the industry and strengthen the position of Samsung among manufacturers of all types of DRAM.


Samsung начинает массовое производство 16-Гбит GDDR6"

For the production of GDDR6 chips Samsung uses process technology with the norms of class 10 nm. Judging by indirect signs, about which we say below, this can be a process with rules 18 nm, which belongs to the first generation of process technology Samsung’s 10-nanometer class. At the same time, the company began producing memory chips with use of the second generation process technology 10-nm class. According to Samsung, the production of GDDR6 chips with the norms of class-10 nm increased by 30 % compared to the volumes of production of 8-Gbit GDDR5 chips manufactured with the norms of 20 nm. If GDDR6 chips were manufactured using the process technology of the second 10-nm generation, the yield would have been much more.

It is also interesting to note that the memory GDDR6, which went to series, has more capacity than the samples shown at CES 2018 and won an award of the exhibition. So, 2018 at CES Samsung showed a 16-GB GDDR6 bandwidth 16 GB/s to contact, and the series went to the PSP memory 18 GB/s for contact. The overall throughput of 16 Gbit chip GDDR6, thus, up to 72 GB/s. It is more than two times faster if you compare the new product with 8 GB of GDDR5 with a speed of 8 GB/s to contact.

Samsung начинает массовое производство 16-Гбит GDDR6"

Finally, GDDR6 memory operates at a voltage of 1.35 V, which is 35% less than the voltage of GDDR5 memory (1.55 V). Memory with low power will consume less and, importantly for the gaming adapters will potentially be better to accelerate given the increase in power during acceleration.

Источник