In this article, we will explain to you how GDDR6 memory works, as well as what its applications are and in which markets this type of memory has been implemented and what is the reason. We will also casually comment on its future and the problem that exists to create new generations of GDDR memory.
GDDR6 memory specifications
The Sixth Generation Graphics Dual Data Rate, or better known as GDDR6, differs completely from previous generations in that while each chip has a maximum bandwidth of 32 bits per clock cycle, we now have two 16-bit channels running at the same time instead of a 32-bit. This is what happened in previous generations.
This means that GDDR6 can serve two memory requests at the same time, but the counterpart is that in order to deliver the information with sufficient bandwidth, the bandwidth had to be increased compared to GDDR5, this led to Because the throughput binary has doubled and therefore GDDR6 transmits or receives twice as much data through its pins as GDDR5.
The advantage of having two channels is that it avoids the contention phenomenon, which occurs when the memory access is occupied by a different process, adding latency to those in the queue. So, GDDR6 not only has higher bandwidth but also lower latency due to it.
GDDR6 memory specifications
GDDR6 memory can operate at speeds of 12 Gbps, 14 Gbps, and 16 Gbps. At least those that are currently on the market in different products, but especially graphics cards and new generation consoles. Although the JEDEC JESD250B standard on which GDDR6 is based specifies speeds of up to 18 Gbps and even speeds of up to 22 Gbps have been achieved in the lab, so far we have not seen any products with GDDR6 at this speed.
Each of the GDDR6 memory chips uses 180-pin BGA package, its voltage varies between 1.25V-1.35V. When the memory is working at 12Gbps, it does so under the voltage of 1.25V, it is Also called low power consumption mode, which can be activated at any time by the processor through the memory controller to reduce power consumption at times when bandwidth is low or in low power systems such as laptops.
Regarding its capacity, it’s 1 GB or 2 GB per chip, with 1 GB being more common than 2 GB in different products. Although the 2 GB configuration is gradually becoming standardized and it is possible that we will see configurations with a higher density such as 3 GB and even 4 GB in the lifetime in which GDDR6 is used.
In terms of bandwidth, GDDR6 can achieve the following data transfer speeds:
|Interface bits||12 Gbit / s||14 Gbit / s||16 Gbit / s|
|128||192 GB / s||224 GB / s||256 GB / s|
|192||288 GB / s||336 GB / s||384 GB / s|
|256||384 GB / s||448 GB / s||512 GB / s|
|320||480 GB / s||560 GB / s||640 GB / s|
|384||576 GB / s||672 GB / s||768 GB /|
You have to take into account that the bandwidth in Gbps can be moved in any space between 12 Gbps and its maximum speed, but we have chosen the speeds of 12 Gbps, 14 Gbps and 16 Gbps.
El Modo x8 o modo Clamshell
This mode was already available in GDDR5, it is based on having two memory chips share the same data bus of the controller, so as not to have to increase the number of pins and therefore the interface. This allows to increase the storage capacity of GDDR6 without increasing the bandwidth and therefore not the pins of the memory controller on the edge of the chip.
For this, what is done is that instead of taking a GDDR6 memory chip, connect its 16 pins to the data bus, one bit per pin, of the memory controller corresponding to a channel. What is done is connect 8 pins to the data bus a GDDR6 chip and the other eight pins to a second chip.
The x8 mode as well as the standard mode, x16, allow the following storage capacities:
|Interface bits||No. of Chips (x16)||Capacity (x16)||No. of Chips (x8)||Capacity (x8)|
|128||4||4 u 8 GB||8||8 or 16 GB|
|192||6||6 or 12 GB||12||12 or 24 GB|
|256||8||8 or 16 GB||16||16 or 32 GB|
|320||ten||10 or 20 GB||20||20 or 40 GB|
|384||12||12 or 24 GB||24||24 or 48 GB|
This is not RAM for low power systems
Since it was designed for systems with tremendous graphics power, whether they are next-gen consoles or the most advanced graphics cards, it is not a type of memory designed for use in low consumption systems. The reason is that its power consumption is higher than that of other types of memory.
For example, a GDDR6 14 Gbps memory chip and with a 32-bit bus when transmitting data which has a power consumption of 7.5 pJ / bit, which results in a consumption of 3.36 * 1012pJ /, which is 3.36 J / s and therefore 3.36 W. This for a single memory chip, so we are not dealing with a type of RAM designed for use in low power systems .
GDDR6 on next-gen consoles
GDDR6 is also used in next-generation consoles, not only as video memory, but also as system-wide RAM in a unified memory or UMA configuration. This means that not only the GPU and its coprocessors have access to GDDR6 but also the CPU and its coprocessors.
Since all three next-gen consoles are based on AMD Ryzen architecture SoCs and use GDDR6 memory, they bring with them a number of common RAM access features. First, in terms of addressing, the CPU and GPU space is separate and depending on the memory address that the memory controller points to, the bandwidth will vary.
This leads to GDDR6 being handled differently when using RAM for a next-gen console than when using VRAM for a graphics card.
The case of Microsoft’s Xbox Series S and X
In the Xbox Series X and S for example, we have two speeds in the bandwidth, it is not that there are two different memory sinks, which can be seen by opening the two consoles, but that ‘there are two different ways to access the memory.
If we talk about the Xbox Series X GPU, it directly accesses the RAM, we have 10 GB of total memory that the GPU can access using exclusively all the available bandwidth and therefore occupying both channels. As for the Xbox One S, the amount allocated only for the GPU is 8 GB, out of a total of 10 GB. As for the rest of the memory, it is accessible by the GPU but using only one of the channels , since the other channel is used by the CPU.
In either case, the CPU accesses the RAM using the classic AMD Ryzen UMC, which communicates at the memory controller clock speed and with a 32 byte bus. In GDDR6, this speed is obtained by dividing the GDDR6 Gbps by eight, as in the console it is 14 Gbps then the memclk is 1.75 GHz. Therefore, in new generation consoles, the CPU does not use the entire memory bus of the GDDR6 channel, but it remains at 56 GB / s.
What about the PlayStation 5 versus GDDR6?
As for the PlayStation 5, SONY did not give differentiated clock speeds like Microsoft. But, in terms of performance, it is to be expected that the SONY console will have separate access regions in order to avoid conflicts between CPU and GPU, which reduces the effective bandwidth and therefore the total performance, by particular of the GPU.
So, in the absence of information, our guess is that the behavior is very similar to that of the Xbox series. Ultimately all three consoles with Zen 2 with RDNA 2 and GDDR6 and we don’t think AMD has gone for different solutions for common issues in different consoles.