BIM projects are becoming larger and more complex, which means that – regardless of the computer's capacity and the excellence of the software – you will experience waiting time while the BIM program runs its processes. There are many comparisons of different hardware options on the Internet where you can see how much faster a given processor or a given graphics card is compared to another, but there are very few comparisons between different BIM programs. We are trying to do something about it with this unformal speed test. It’s important to remember that our test should not be taken for more than it is: two examples of performing the same tasks in two different BIM programs, while observing the speed difference.
The experiment was conducted in Archicad 24 and Revit 2021, which are the latest versions of Graphisoft's and Autodesk's primary BIM programs, respectively. Both programs were installed on the same laptop with the following specifications:
No plugins have been used and no BIMcloud or BIM360 has been connected. The experiment is performed as the BIM programs normally work during ordinary installation and use. The computer has been restarted before each attempt in each BIM program. To be able to document the process, we have made screenshots throughout the experiment, just as we have had Windows statistics for the hardware load open throughout the experiment. These two utility applications will, of course, seize some of the computer's resources, but the premise is the same for both BIM programs.
We have teamed up with Revit expert Simon Olafsson, CEO at IKT-lederne , who has many years of experience with Revit, as he, among other things, has been working as a BIM manager at various Danish design studios. In this way, we have ensured that both Revit and Archicad are managed in the best possible way.
To test the speed while the two BIM programs handle many objects, we have performed the same task in Archicad and Revit:
The small house is without windows, doors, stairs, furniture etc. to make the modeling as uniform in Archicad and Revit as possible so that the comparison is as close as possible. Using only the most typical object types, that are also used on virtually all BIM projects and in the simplest version.
With our stopwatch ready, the experiment starts by copying the housing so that there are two. Then the two houses are copied so that there are 4, and these are copied so that there are 8 and so on to 16, 32, 64, 128, 256, 512, 1,024, 2,048, 4,096, 8,192 and finally 16,384 small houses. The whole experiment takes place in 2D top view alone. For every doubling of houses, the copying takes longer and longer. While the first houses are copied so fast that we did not even have time to start and stop timekeeping, after a few copies it takes a few seconds, and then we experienced real waiting time. Here you can see the measurements from both Archicad and Revit.
It seems that the waiting time is increasing exponentially in both programs, but with a much faster growth in Revit than in Archicad. When copying to 256 houses takes 12 seconds in Revit, while copying to 512 houses only takes 4 seconds in Archicad. However, the difference is getting bigger and bigger. Copying to a total of 2,048 houses, it takes 10 seconds in Archicad and over 5 minutes in Revit. With a total of 4,096 houses, it takes 20 seconds in Archicad and almost 20 minutes in Revit. Subsequently, we stopped the measurement in Revit, as the waiting time became disproportionately long. In Archicad, copying to a total of 16,384 houses took 2 minutes and 17 seconds. Then Archicad “crashed”, and we stopped the experiment.
Interestingly, however, the difference between the two programs also increases exponentially. When copying to 2,048 houses, Revit is 30x slower than Archicad, while when copying to 4,096 houses, Revit is 60x slower!
During the experiment, we kept an eye on the computer's use of the processor and RAM. While the RAM consumption in both programs was around 10 GB and fairly uniform throughout the experiment, there was a big difference in how the two programs utilize the processor power of the computer. While Revit occasionally switched between a short load of each processor core, Archicad loaded all 6 processor cores (12 threads) fairly evenly. This means that the task of copying objects utilizes modern computers with many processor cores better in Archicad than in Revit, which might work better with a processor with only one very fast processor core. However, it is not possible to buy computers today that have only one processor core. Most computers on the market have between 4 and 8 processor cores – some up to 64!
While we acknowledge that copying many thousands of objects repeatedly is not a common task, this experiment shows how much it takes to push the two BIM programs to wait times, which for both Archicad and Revit users will feel like a very long time. If you work on large, complex projects, such as hospitals or large landscape assignments, you know both programs have long waiting times when copying, moving, editing objects in the model or e.g. generates sections, facades and quantity extracts.
It is striking, however, that the difference in how the two programs handle this task of copying many simple objects is so profound. If you have to wait more than 20 minutes for a program to perform a given task, in many cases you will wonder if the program has "gone down" and must be restarted, which can delay the task further.
If you look at the file size of the project with 4,096 small houses, the Revit file takes up approx. 60 MB, while the Archicad file takes up 37 MB. The Revit file is thus more than 60% larger than the Archicad file, which makes a difference every time you open and save the projects. Or when you, for instance, store many versions of BIM files on the company's servers, back up and upload to online platforms in the “cloud”.
To see how the two programs handle import of IFC files, we modeled a simple building with a deck, 10 walls and 6 windows at each floor with a total of 20 floors. The windows were quite simple with only a fixed part, while the walls and decks were modeled as layered building parts consisting of concrete, insulation and concrete. We copied this building to 16 identical towers and exported the project to an IFC file (IFC 2.x.3. Coordination View 2.0). Modeling as well as export were performed only in Revit. Then we restarted the computer and imported the IFC file into both Revit and Archicad. This means it was the same IFC file from Revit that was imported to both programs. We did not change the default import or export settings in either Revit or Archicad.
The import did not take that long with either Revit or Archicad. It took a little more than 3 minutes in Revit and a little less than one minute in Archicad. While we know that IFC files are often heavy and take a long time to import, it is weird that it takes Revit over 3 times as long as Archicad. Especially considering that the IFC file comes from Revit, so one must assume that the translation will be faster than when the software must translate between two different BIM programs' way of modeling each building part.
Conclusion of the comparison
Year by year, the BIM programs are optimized, and the hardware becomes faster, but at the same time our BIM projects also become larger and increasingly complex. Therefore, when you are working with BIM you often experience waiting time while the computer operates. Interestingly, we cannot measure how long it takes to copy 64 small houses to 128 in total in either Archicad or Revit. We actually modeled the 20-story tower in both Archicad and Revit to test the speed of generating a section through the building, but this happened completely without waiting time in both applications. We have become accustomed to the fact that even quite complex tasks are performed as fast as we can click with the mouse. On the other hand, it has a big focus when it comes to waiting seconds or minutes for the computer to work.
Perhaps rightly so, as a waiting time of 15 seconds every 5 minutes, add up to 20 minutes of wasted work per day, which equates to 2 weeks of work per year. It is a hidden expense when choosing software if it turns out that one program provides significantly more latency than another. To prove the point: Imagine if you could switch to another BIM program and then get two weeks extra summer vacation paid ;-)
In our simple test, it seems that there is a big difference between how Archicad and Revit handle the same task and how the two programs utilize the computer's hardware. It does not help to buy a new fast computer with lots of RAM, a strong graphics card and lots of processor cores if the task is mainly performed on only one of the many cores. This makes it difficult to eliminate the waiting time.
Copying a lot of objects takes up to 60 times longer in Revit than in Archicad and importing IFC takes 3 times as long, however, is more than we had expected before we started the experiment. It's a bit of an eye opener that the two programs – which are so similar in so many ways – still have big differences in terms of speed. This test does not show that one program is better than the other. There is a difference in the functions and workflow in the two programs, just as there is a price difference. Our two tests are by no means complete, and you can easily imagine that you test differently – e.g., on other hardware Revit may be faster than Archicad for some tasks, just as many tasks will take the same amount of time in both programs.
And then one last thing: it is important that in this type of theoretical experiments, you acknowledge that it is a theoretical, and not a practical, endeavor. We made the test with a wink in the eye (ay ay, say no more!), as the world is much more complex than it appears when copying 16,000 small houses in two BIM programs. We want to show the differences between different BIM programs as accurately and objectively as possible, but at the same time, being a Graphisoft partner, we have an obvious advantage in presenting Archicad in a positive light. – We don’t presume that this comes as a big surprise for anyone…