Issue #1,100 | Inside the Business of CAD | 14 June 2021
Written by Gustav Näslund
Mathematics software for engineers is not something I am especially familiar with. When I took calculus, we had no calculators, and in my field of study at the time, transportation engineering, we tended to work with simple trigonometry, rather than advanced equations.
So, reader Gustav Näslund suggested running an article about the software available in the engineering math editor space.
- - -
Software for engineering math are not strictly CAD, but I think nonetheless that they are essential and powerful tools for design engineers to perform and document engineering calculations.
I think this class of software is not as well known, nor used, as much as it should be in our industry, which is why I would like raise awareness of them.
If you are not familiar with this kind of software, think of it as a free form digital notepad that does calculations interactively. It supports units of measurement and live links between all variables and equations. The figure below is typical:
In general, this kind of software lets you add equations, text, and images in a freeform manner, and you can rearrange them as you like. All equations are live-linked, so that they update their values when something changes.
Because you work in a paper-like format, the format is self-documenting. When you are done, just save it as a PDF file, and you have your calculations documented.
The equations support all units of measurements. This means, for example, that we can mix imperial and metric units in equations without fear. The mandatory reference here is to the NASA Mars Climate Orbiter debacle; see en.wikipedia.org/wiki/Mars_Climate_Orbiter.
I find that the programs have a very low threshold for getting started, and run fast, so they are suitable for even the simplest calculations — never mind the most advanced.
Math Software Packages
The figure below is of SMath Studio; MathCAD and Maple Flow look similar.
MathCAD has long been a player in engineering math editor field, 35 years in the business. It was taken over by PTC in 2006, and then received a total makeover as MathCAD Prime in 2011.
The latest version of MathCAD is available now by subscription only, in the ballpark of $1,000/year for a commercial licence. Educational licenses are available. www.mathcad.com and en.wikipedia.org/wiki/Mathcad
Maple Flow is a very recently released (2021) direct competitor to MathCAD from Canadian company MapleSoft. It appears to me to be aimed directly as a competitor to MathCAD, as it has a very similar interface and works in the same way.
Maple Flow seems to be priced at $2,390 for a single perpetual commercial license. Educational licenses are available. maplesoft.com/products/mapleflow
SMath Studio is a free alternative created by Andrey Ivashov, first released in 2006. It has capabilities that are very similar to the original MathCAD, before its remake as MathCAD Prime.
Being free, it is not as advanced as MathCAD, but for 90% of engineers it will be just what they want (and need). It has a thriving community that creates plug-ins and so on, and is funded through donations. www.smath.com
[Gustav Näslund owns GN Tech of Sweden, and specializes in product development and mechanical design. www.gntech.se/en]
And in Other News
The next release of NX is shipping later this week on June 18, while Solid Edge 2022 is scheduled to become available later this year in October.
In other Siemens news, the company acquired Nextflow Software of France for its meshless simulation technology. Smooth-Particle Hydrodynamics does particle-based computational fluid dynamics for gas and liquid flows. www.nextflow-software.com
- - -
Radica Software’s Electra Cloud schematic electrical design software now exports BOMs (bills of material) directly to OpenBOM for report creation, with further links to accounting systems. A future upgrade will export data from OpenBOM back to Electra and its Vecta.io SVG [scalable vector graphics] editor. radicasoftware.com
- - -
3D Systems has divested itself of its on-demand additive manufacturing service, named On Demand Manufacturing, for $82 million to Trilantic, which promptly renamed it Quickparts. Details at schnitgercorp.com/?p=18523
- - -
Jon Peddie Research reports that the GPU market last year grew by 39% to 199 million units. The market share between suppliers remains unchanged:
-
AMD - 15%
-
Intel - 17%
-
nVidia - 68%
Letters to the Editor
Re: U-Splines for Predicting Failure
Though well outside of my profession, I find FEA [finite element analysis] interesting. Does Coreform run on GPUs [graphics processing units]? I searched their forum for the term, but there was only one unrelated result.
How does the speed of their software compare to competitors that do run on GPUs? Many FEA software packages are optimized to run on nVidia GPUs: nvidia.com/en-us/gpu-accelerated-applications.
It would seem that if their core calculations are more efficient on CPUs, writing the code to get their software to run on GPUs would be a further advantage.
- Peter Lawton
Affiliated Engineers, Inc.
Greg Vernon replies: Currently, we only execute on CPUs, but we’ve architected our algorithms and code to be maximally parallelize-able. GPU-compute is definitely on our roadmap.
A large bulk of our compute time is spent within the linear and nonlinear solvers provided by the open-source PETSc project, which is currently developing GPU support. See mcs.anl.gov/petsc/features/gpus.html.
We’re currently focusing on classes of problems that have [in the past] required incredible amounts of compute resources, but rather than throw more resources (more CPUs and GPUs) at the problem, we’re consistently finding that throwing better math at the problem is giving us orders of magnitude faster solves.
I like to think of it in terms of Exascale* ambitions. If the point of Exascale compute is to solve problems, then there are two ways to achieve this:
-
Build faster/larger parallel computers
-
Use better mathematics that make the problem smaller for existing computers
We’re taking the second approach!
-Greg Vernon, director of product management
Coreform
*) Exascale = computing systems capable of calculating 10¹⁸ floating point operations per second, or more.
Mr Lawton responds: The two approaches are not mutually exclusive, of course, so Corefrom may consider using both — the outcome will be even greater advantage over competitors.
It sounds, too, like the math they are improving requires a different coding team than would be required to get Coreform to run on CUDA, so the two improvements can happen in parallel, once you have the right personnel.
Notable Quotable
“He’s at home recovering for a few days. He saw an offensive tweet.”
- Management Speak
Thank You, Readers
Thank you to readers who donate to the operation of upFront.eZine:
-
Martin van der Roest: “One of those decisions that is a no-brainer. I really appreciate your perspectives and contribution to our industry. Thank you!”
-
Dairobi Paul
To support upFront.eZine through PayPal.me, I suggest the following amounts:
-
$25 for individuals > paypal.me/upfrontezine/25
-
$150 for small companies > paypal.me/upfrontezine/150
-
$750 for large companies > paypal.me/upfrontezine/750
Should Paypal.me not operate in your country, then please use www.paypal.com and use the account of [email protected].
Or mail a cheque (US$ or CDN$ only, please) to upFront.eZine Publishing, Ltd., 34486 Donlyn Avenue, Abbotsford BC, V2S 4W7, Canada.
*4720
Comments