Hi Everyone.

I am Chris Ziomkowski, the CEO of XTend Online. A couple of issues I wanted to address. First, do not be fooled by the small size of our team. Google was founded by just 2 people, and today has nearly 100,000 employees.  Same with Apple. In fact, all companies begin with a small number of founders. We are no different.

With regard to chip design, nearly every chip ever created is done with a small team of "wizards". This includes complicated processors designed by companies like Intel. The remainder of the staff helps to write test procedures and speed the development by writing some of the more repetitive logic, but creation of an architecture does not require a large team. It just requires a few dedicated and experienced people who understand the problem they are solving. Large teams can make it go slightly faster.

My cofounder and I have been working on this architecture since 2017 when we first realized there was a lot of misinformation floating around. At the time, nobody was interested in hearing our opinion, so we decided the best way was simply to show people. 18 months later, we now have a real architecture to show, and are ready to bring this product to manufacture.

The small size of our team shows in things like our website, which unfortunately hasn't been updated since September, and still has vestiges of things that are no longer relevant. We are looking to add a social media manager and a web programmer to our team for the coming media push.

The technology is very real. We have tried to demonstrate to everyone that GPUs use a shared memory bus which creates an artificial bottleneck. Deep DRAM stacks limit the throughput to a few hundred GB's/sec. The Hyperminer architecture uses a fine grained mesh by contrast, which has a total memory bandwidth on the order of 2 Petabytes/sec. Removing this bottleneck pushes the performance limit out to the total bisection bandwidth of the mesh,  which is more than an order of magnitude greater than what is achievable by a shared memory bus.

You can imagine it like a car manufacturer. A GPU puts all the parts in one pile, and then says we can't go any faster because there are limits on how fast a person can pick a part from the pile and hand it out to a skilled craftsman who will build the whole car at his workshop.   Our architecture is more akin to a factory, where we presort all the parts into small piles, and then use an assembly line where each worker only does the specific task his station is specifically optimized for.  The limit becomes not in grabbing parts, but in moving partially built assemblies between various stations. This is horribly simplified, but it illustrates the point metaphorically.

I invite anyone with questions to engage me. I will also be at MPWR in Vancouver on March 12, so if you will also be there I will be happy to meet anyone and discuss how this will change the industry.  You don't have to believe me, but you do have to accept physics. This is coming and will have the same effect that GPUs had on CPUs. The industry is simply too large, there is too much at stake, and too many skilled people are getting involved.  As a community we need to transition away from GPUs and onto a 3rd generation of programmable hardware optimized for digital mining challenges.