Qspice wins over LTspice on a basic boost converter

It was nice to experiment a comparison between Qspice and LTspice of the same very basic boost converter.

On this example, Qspice runs quite faster, with Total elapsed time: 0.100831 seconds., while LTspice needs Total elapsed time: 0.510 seconds (mean times, default settings for both programs).
Furthermore, LTspice speed is not repeatable, sometimes needing quite a longer time (Total elapsed time: 1.161 seconds.) for the same simulation (not understood when and why, just rerunning the same simulation). Qspice is instead almost deterministic, with < 50 ms variability.
Simulation esults are almost identical.
Let notice that both simulators cannot give reasonable results without specifying the Maximum Timestep, showing the same type and amount of error.
(Win10 ent, i5-6300HQ CPU @ 2.30GHz, 2301 Mhz, 4 cores, 16 GB RAM)


Thank you for sharing this. I find it interesting to see what folks think of it. I’ve used so many different simulators and I wasn’t really an LTSpice user, but when I found QSPICE, it was a breath of fresh air.

I started spice simulation from Pspice when I was in university. After I started my career, our CTO introduced LTspice to me and I have been using it since 2010. LTspice was the most important tool in my career and study in electronics. Whenever I train a junior engineer, I encourage him/her to learn LTspice.

As you know, Mike Engelhardt is author of LTspice, and in past few years, he started this new project and it is Qspice now. Frankly, I immediately switched to Qspice since the first time I used it. One of my major area is switching mode power supply, and simulation of SMPS is one of the biggest challenge as discontinuity in equation. Mike knows exactly what is required to make thing work. I am amazed by Qspice simulation speed at first. And later on, I found that Qspice offers .bode (frequency response analysis) and C++. Thanks to Mike, he accepted my suggestion to offer .bode with selectable input and output node which allow .bode to calculate transfer function between any two nodes in circuit, which allowed open loop or close loop frequency response measurement easily. Frequency response analysis can be done with .ac analysis but the limitation is that cannot handle switching circuit (e.g. logic, SMPS etc). .bode offers a time domain method for frequency analysis. Another amazing feature is C++, for example, I just worked with @RDunn for a general C++ PID controller. This is amazing for a “free” software (I think I should also thanks Qorvo too) which offer analog and digital domain simulation with C++/Verilog, and with an easy learning curve and simulate with this speed in a general laptop.

Beside of Pspice / LTspice / Matlab Simulink, I didn’t use other simulator so it is unfair to say Qspice is the best or not. But if you are seeking to learn a spice simulator, no doubt I suggest you Qspice. Some people may struggle if they want to change from LTspice to Qspice just for simulation speed (but frankly, benefit is NOT only speed) and have to adopt to a new GUI they not familar. But as you are not LTspice user, just go ahead for Qspice, you won’t regard it.


I think for fair comparision you need tune both SW for maximum performance.

“… for maximum performance”

There are two variants available

  1. for max. speed.
  2. for given precision.

And you need to use the same settings, because LTspice and QSPICE default settings are different (afaik)