Micron in their latest research article has revealed the full potential of their upcoming GDDR6 memory which will be equipped with the latest graphics cards from GPU manufacturers. The GDDR6 memory has been in works for quite some time and many manufacturers have already started mass production so we can expect to see the memory shipping soon in the latest cards.
Micron GDDR6 Memory Research Article Reveals Full Potential - Up To 20 Gb/s Possible With Little But Effective Voltage Bump
Within the research paper, Micron discusses several aspects of the GDDR6 memory which includes silicon changes, channel enhancements, and performance measurements. While GDDR6 follows an evolutionary path over GDDR5 and GDDR5X memory, there are still some significant changes in the underlying architecture to boost memory bandwidth while saving power. This makes the VRAM a viable option for next-generation consumer graphics cards such as NVIDIA's upcoming line of the GeForce products.
One detail that caught my eye was the Micron has already determined in performance measuring that their GDDR6 memory could extend beyond the 16.5 Gb/s range. The result proved that with a slight but helpful bump in I/O voltage, the memory chips can push speeds as high as 20 Gb/s which is a significant jump from the JEDEC defined 14 Gb/s target.
While the preceding results demonstrate full DRAM functionality up to as high as 16.5Gb/s, it is possible for the overall performance of an architecture to be capped by timing limitations in the memory array itself. To determine if this GDDR6 interface could extend beyond the 16.5Gb/s range, the device was placed into a mode of operation which exercises only the I/O while bypassing the memory array.
The oscilloscope measurement presented in Fig.15 confirms that when bypassing the memory array, and with a small, but helpful, boost in I/O supply voltage, it is possible to push Micron Technology, Inc.’s GDDR6 I/O as high as 20Gb/s.
We know that from our own experience that GDDR5X proved to provide a big boost in memory performance with clock bump. At reference voltage, it was able to achieve the same clock rates that are offered on the Titan XP with a GTX 1080. If we are looking at similar performance scaling with GDDR6, then 20 Gb/s overclocks on upcoming cards can provide a similar performance boost.
So just what kind of performance we might expect from a 20 Gb/s clock bump, well to put things into perspective, a 256-bit card with such speeds would able to deliver 640 Gb/s bandwidth which is close to Titan V's 652.8 Gb/s (HBM2). A 384-bit card would almost hit the 1 Tb/s barrier with an approximate bandwidth of 960 GB/s surpassing NVIDIA's Tesla V100 solution.
GPU Memory Technology Updates
|Graphics Card Name||Memory Technology||Memory Speed||Memory Bus||Memory Bandwidth||Release|
|AMD Radeon R9 Fury X||HBM1||1.0 Gbps||4096-bit||512 GB/s||2015|
|NVIDIA GTX 1080||GDDR5X||10.0 Gbps||256-bit||320 GB/s||2016|
|NVIDIA Tesla P100||HBM2||1.4 Gbps||4096-bit||720 GB/s||2016|
|NVIDIA Titan Xp||GDDR5X||11.4 Gbps||384-bit||547 GB/s||2017|
|AMD RX Vega 64||HBM2||1.9 Gbps||2048-bit||483 GB/s||2017|
|NVIDIA Titan V||HBM2||1.7 Gbps||3072-bit||652 GB/s||2017|
|NVIDIA Tesla V100||HBM2||1.7 Gbps||4096-bit||901 GB/s||2017|
|NVIDIA RTX 2080 Ti||GDDR6||14.0 Gbps||384-bit||672 GB/s||2018|
|AMD Instinct MI100||HBM2||2.4 Gbps||4096-bit||1229 GB/s||2020|
|NVIDIA A100 80 GB||HBM2e||3.2 Gbps||5120-bit||2039 GB/s||2020|
|NVIDIA RTX 3090||GDDR6X||19.5 Gbps||384-bit||936.2 GB/s||2020|
|AMD Instinct MI200||HBM2e||3.2 Gbps||8192-bit||3200 GB/s||2021|
|NVIDIA RTX 3090 Ti||GDDR6X||21.0 Gbps||384-bit||1008 GB/s||2022|
Micron mentions that they are very confident in claiming that GDDR6 data rates are gonna extend beyond the Jedec defined speeds so this is a just a hint of the overclocking potential we would get to see on the next-generation gaming cards.