Dyndrite Enters 3D Printing with Their Own Geometry Kernel

The Dyndrite Additive Manufacturing Toolkit (AMT) is built on the company’s own multi-threaded "Accelerated Computational Geometry Engine" (ACE). It is the first GPU-powered geometry engine. (Picture courtesy of Dyndrite.)

The time of using 3D printing technology, with its completely digitalized workflow, is now. As we sit sequestered by COVID-19, unable to physically attend to manufacturing processes, we can only dream of a 3D printer that would accept a part design in its digital format and produce the part without doubt or question. A dream for them to be automatic and function unattended, with no machinist. Every manufacturing process that involves a mold, a casting, a sheet metal press, NC machine, milling machine or lathe, depends on a human operator being there. Unless that manufacturing process is considered to be “essential” by governments, business that must go on despite the risk of spreading COVID-19, and humans will be required to be present otherwise that part will not get made.

Even though 3D printers accept a digital format, usually as an STL file, that digital format is hardly automatic, easily created, efficient in its description, or economical in terms of storage. One company, relatively new on the scene, is built on the premise that STL has it all wrong; that STL “imprisons” a design and has a better idea.

Seattle startup, Dyndrite (pronounced DEN-drite, like the undesirable crystalline structures that form in lithium-Ion batteries) has created what it thinks is a better way to define and send geometry to a 3D printer. Their solution centers on their own geometry engine.

We know geometry kernels are old news. The one you have works fine. Similar to the engine in your car, it just runs, on its own, under the hood. You don’t get your hands dirty working on it. How can anyone improve on something that has a hundred years of refinement?

Here comes Dendrite’s 28-year old founder/CEO Harshil Goel, who promises a new engine, one that is more efficient, more economical, like the Elon Musk with his electric engine promising every advantage over internal combustion engine.

Dyndrite’s CEO Harshil Goel is 28. He uses the picture in his first suit for his LinkedIn profile.

A diploma from UC-Berkeley, one of the nation’s top engineering schools, would be a proud possession of any engineer. Goel has 3 of them. He carries a BSME and a bachelors at the same time and a year later added an MSME. He was on his way to 4th (PhD) but was lured out into industry, he says.

What he saw in industry did not please him. The mathematics and engineering were not coming together the way they should.

“I like to solve math problems for mechanical engineers,” says Goel, and here was one staring at him in the face. Model geometry that described one way, with one geometry engine which then had to be parsed into another format. A blocky, chunky, inexact, most inelegant format. It took a lot of processing to do it, considerable human intervention, and an enormous amount of storage space.

Why couldn’t geometry be described the same way, with the same format, the same graphics engine in both the design computer and in the 3D printer?

From Terabytes to Kilobytes

“We can take terabytes of model information and condense them into kilobytes by using the same geometry kernel in design and manufacturing,” says Goel. “We can pass a 4 KB line of code to a machine instead of a massive STL file.”. The geometry then unfurls from the line of code.

The Company

Dendrite is a Seattle-based startup that has received $10 million funding. Early investment included Carl Bass, ex-CEO of Autodesk, who immediately appreciated Goel’s skills in math and the machine shop. The last round included Google’s AI oriented Gradient Ventures. It is looking to hire developers to fulfill its mission, which is to not only build a GPU-based geometry kernel but apply it towards 3D printing. 

The company is holding a free online event featuring members of its Developer Council, including HP, EOS, ExOne, Impossible Objects, NVIDIA, Renishaw, SLM, and others. Speakers include Industry experts, Todd Grimm and former Autodesk CEO, Carl Bass.

Conference dates are April 21-22. Find more details here.