Technology

NVIDIA war and GDDR6X

It is surprising how a paperwhite can change everything, opinions and impressions in between. But it is that what has just been known transcends four companies on different fronts. Let’s start at the beginning: Innosilicon, the Chinese-funded company (covert or not, it matters little) is making strides to compete in desktop GPUs and HPC with NVIDIA, AMD, and now Intel. His intentions reach such an extent that we now know that the GDDR6X It is developed by them together with Micron!

In case you don’t understand the importance of this fact, it turns out that the development and production of this type of memory has always been attributed to him, which is only being used by NVIDIA… Or not. And it is that if we remember the first graphic to include it was precisely from Innosilicon, but since the power was not representative it was not given more importance.

Micron and Innosilicon, USA and China, NVIDIA and AMD, who wins and who loses?

Innosilicon-GPU-GDDR6X

The reality is that there is an agreement signed by Micron and Innosilicon where the development is a joint part and this implies that NVIDIA is a mere spectator dominated in this area by a company that is fighting at the same time to be equal to it, which is at least ironic.

The cooperation of both companies for the GDDR6X left Innosilicon with an important part in the agreement with Micron, it is really the most crucial part: PAM4 modulationessential to reach the current speeds and voltage.

Innosilicon-GDDR6X-Whitepaper

But there is more, at the Zhuhai 2022 Integrated Circuits and Software Industry Annual Conference Innosilicon presented and talked about the new 21 Gbps GDDR6X, that is, the memory that integrates the new RTX 3090 Ti and supposedly the next RTX 4090. In said conference it was assured that with a bandwidth less than 256 bits were able to overcome 5Tb/sa performance similar to that achieved by some DDR4 modules.

Although pricing was not discussed, Innosilicon said that currently in terms of performance and price, not ratios as such, but separately, GDDR6X is the high-bandwidth storage solution. most profitable todaysurpassing HBM in some points, but at a much lower cost.

Innosilicon-GDDR6X-1

Therefore, both AMD and NVIDIA are at the mercy of memory being developed by equal parts American and Chinese companies, where the latter could be a threat in less than a decade. And for this reason we launched two questions/reflections into the air, is AMD the only one that has seen the move and decided at the time to opt not to use a greater internal bandwidth with Infinity Cache so as not to depend on a future direct competitor? Is this a covert war between the US and China to dominate the gaming and HPC performance via GPU? Argue your answers.

VIA: Mydrivers

Source link

Related Articles

Leave a Reply

Your email address will not be published.

Back to top button