Multi-Thread

[jfjacques]

This is more of a complaint than anything else. 

Honestly why don't you guys multi-thread. 

I have to use Revit for work, but I miss the slickness of Archicad's viewport. Just the speed. Revit has so many windows to open and nested menus, but it's display performance is awful. 

I am running on an i7-8600k, overclocked to 5ghz on all cores: pretty much as fast as you can get at the moment, but pretty much all wasted thanks to this product.  I only see one core being used-  the gpu barely moving as well. 

I can't imagine what someone on a slower machine might be experiencing. 

It's the same with several autodesk products- I'm looking at you 3ds. Is it all running on some common half-baked architecture? Is autodesk just hoping to wait it out as processors get faster?  

When will you use more than one core - it's clearly possible, your competitors do it. What is the inherent issue?

Details pls - i'd love to have one of your software engineers answer this question.  

[nicholastpreston]

We need drafting speed yesterday!!!!

[jeroen.van.halderen]

I completely agree, multithreading would greatly improve my Revit experience as well. I am developping plugins for Revit and I have to test my code in some big revit projects, it always annoys me how long it takes to load the project or perform some tasks. I think the upcoming release of .NET 6 is a great chance for Autodesk to address both the switch to the new .NET and the multithreading issue.

Also, it would be great if more processes in Revit could make use of GPU acceleration. The way it is now Revit peaks at about 30% GPU usage, maybe I'm doing something wrong but I think Autodesk can do better here.

Thanks in advance for taking my suggestions into consideration.

[daniyar.assylbekov]

I know Revit team is actively working on performance improvements, especially in terms of taking advantage of multiple CPU cores. Nonetheless, I'd love to see more tasks optimized asap, especially for regenerating complex families hosting multiple nested families and so on. It takes minutes to change single dimension parameter. I did make a lot of model optimizations I could think of, but time efficiency is still not there. I'm working on such project and there are no other software to my knowledge that can do it like Revit. Moreover, I really enjoy Revit, I just wish it could be optimized much earlier.  

Thank you!

[kimberly_fuhrman-jones]

Hi, @daniyar.assylbekov ,

You are correct! Our teams are actively working to improve performance with Revit. Would you be willing to share a model with our product team working on performance and provide additional feedback? I can provide additional details to submit a support case, and we will direct this to the appropriate team. Please feel free to message me directly with any questions.

Thank you for your Idea!

Kimberly

[TaylorLilly]

Please vote for visibility.

Autodesk- in the past 5 years your market cap has grown by 333% to an impressive 70 billion dollars. 

We know that when you apply the resources and the right team, you can build amazing products like fusion360, which have been well received. 

So why is Revit so neglected? Why do I need a $3,000 machine to run your program smoothly, but I can play GTA5 on 15 year old console software? A game with much more complexity and a much higher graphic load than Revit. I can run unreal engine on a basic laptop faster than Revit 3d views can draw. I can orbit in Lumion fully rendered and animated with wind, clouds, etc, but Revit struggles with a simple shaded view with materials. Just loading elevations with realistic materials and depth fogging can crash Revit easily. 

There are some core inefficiencies in the software. The program that some of us spend every day in brings no joy. Its not smooth, or fun, or sleek, or pretty. It's clunky, laggy, hesitant, and slow. Doing light research has pointed me to a single reason why- Revit is built on old software architecture and does not support multi-threaded tasks natively. It relies heavily on processors that have a single core high clock speed. If this is true, its a huge issue. 

So I am proposing something - rebuild it. Ground up. You have the resources. Your competitors are going to leave you in the dust very soon if you do not. We don't want to see that. We have a lot invested in this software. 

I'm actually begging here. You guys can do great things when you decide to. 

Also, if you do not rebuild it, none of the incremental processes you improve will be enough. The graphics industry is leapfrogging and Revit is *crawling* toward improvements. I'm not sure I can tolerate it much longer. It's costing money to move so slowly. 

[jeroen.van.halderen]

I agree with you on some points, especially the multithreading, it would that Revit could take much better advantage of multi-core processors, which seems important since that's the way the industry's heading. However, totally rebuilding Revit would really be a lot of work, but personally I would really like to take on such a job, it's always fun to improve software right? While I'm at it I would also add an unmanaged API, just for fun:)

[magnus.enstromBIM]

I agree with you. If you compare how Revit looks now and how it looked in the year 2005, it is not much of a difference. And from using Revit in the last 8 years I haven't seen much improvement regarding speed and UI. It feels like the software is constantly extended on a platform built a long time ago. A fresh start with modern tech would be beautiful. 

What I would like to see is somebody doing what Figma did to transfer the design industry. But for Revit, basically. Skip the whole version limitation, it just slows down everything. Constant processing, freezing, regenerating views etc just feels old.

[paul.t.macknight]

@jeroen.van.halderen Well, it would be a lot of work, but at this point if they were to give us "everything we wanted and more," which is what I've heard is the promise with 2022 (and am highly suspect of that statement), they don't have many other options. Honestly, to try to rework a program that wasn't even originally built by them to get something close to the above request in addition to other MUCH needed fixes and improvements would likely take such an overhaul that it would in fact be cheaper and easier to build from the ground up. And I'm saying this fully recognizing that they already essentially tried overhauling it by transitioning the code base so they could more easily make changes.

Revit, at this point, is archaic, even in terms of other BIM software that are also behind when compared to other modelling software that are so much more resource efficient that they practically allow real time editing in rendered views with the same hardware it takes just to run Revit smoothly. It's almost an insult to the AEC industry to trick decision-making execs who don't fully understand Revit's flaws with pretty words, shiny features, and in some cases even empty promises because of their own lack of understanding of Revit's limitations (I'm specifically thinking of our company's fiasco with Revit Server, which three times caused the deletion of all doors and many dozens of hours of rework in one of the largest projects we've had in the past decade, and Autodesk's insistence that it did in fact work and they will make it work for us when in fact it couldn't).

I agree 1000% with Taylor that Revit needs to be retired in favor of a new software (they don't even need to stop calling it Revit, just give it a more modern infrastructure through a rebuild rather than a retrofit). Let the team currently working on improving Revit move over to the new project and cease all major updates on the software, leaving just a handful of people primarily responsible for pushing interim updates to fix major bugs that break intended features and prevent users from actually using the software until the new one is ready to release.

[GGMSU]

This is exactly what they are doing from my understanding.  There have been a number of articles written and if you find their last town hall YouTube video with the engineers they seem to imply as much.   Revit will be replaced with a cloud based application and users will have a thick (most likely) or thin client on their local machine to interface with.  When this will happen is anyones guess. I hold no hopes of seeing anything for at least a couple of years.  And then there is the initial phase of new complex software in the wild that will certainly reveal sizable issues that need to be fixed. So all in all, I’m guessing 2025 we might actually have something revolutionary as opposed to whatever we want call the Revit update cycle. 

[paul.t.macknight]

@GGMSU And that's another problem I have. We don't need something like that because not everyone can use an application like that unless they're willing to invest in secure options. All these projects that are currently using Revit on projects involving government, classified information, UCNI, or otherwise undisclosable information in the commercial market won't want to allow designers to use a cloud based storage unless Autodesk can prove it's entirely secure, and I can tell you from experience that they can't or won't because our company does not use Autodesk 360 because they don't offer a secure service, among other issues. The alternative is to allow this cloud to be put onto a secure server under the control of another entity (be it the client or the AEC firm), which Autodesk would never allow because they're as bad about controlling their own products as Apple is.

While the idea is nice in theory, it just doesn't work on a more practical level because it actually reduces the number of potential users, meaning they'll likely up the pricing to make up the difference and potentially drive even more people away. And it shouldn't be needed because it should be doable on the end-user's computer, just like other modelling industries. My computer is spec'd to do well or excel with anything, and is only adequate when it comes to Revit. Even Microstation/OpenBuildings runs better and as a better built-in renderer (though we typically opt for an external GPU based renderer anyway), with it's biggest sticking point being usually unnoticeable: the inefficiencies in accessing the model's database of elements. Even it's not as slick as other modelling non-AEC software, but it's still a huge improvement to Revit and it's ENTIRELY client based. But when Autodesk get a "bright" idea they won't let go of it, especially since they likely have too much money put away in this concept to scrap it in favor of a more wide-reaching solution.

[GGMSU]

@paul.t.macknight , I don't disagree.  However there is still not enough real information for any assessment.  Who knows what's going to shake out and I have believe they are aware of the valid security points you make given their position in the industry.  I also don't necessarily know that having a cloud based backend requires storing files in the cloud, which is partially why I think it will be a think client.  I do some government work too and frankly my biggest concern with this premise will be how they handle the cloud dependencies.  I don't always work where I have a reasonable internet connection or one at all.  I too prefer local software.  But we will see.  I may very well be retired by the time they get their act together.  

[JamieMann]

@kimberly_fuhrman-jones Did daniyar get back to you with a model? I'm happy to give you a couple of large models to work with <450mb.

I have a fully optimised AMD 5900x CPU running some cores at 5ghz that have affinity to Revit and I still feel like its chugging. I'm not talking about parallel processing here just that I noticed I got no decrease in "latency" coming from an older processor to bleeding edge hardware. Also running fantastic RAM and an M2 drive all cooled appropriately - I'm just not getting the value out of this hardware when using Revit.

[kimberly_fuhrman-jones]

Hi, @JamieMann ,

We're always looking for examples we can use to test performance. Please email me at kimberly.fuhrman@autodesk.com and I will get you connected with the right team.

Thank you!

Kimberly

[Petermartens]

Autodesk is more concerned about their business model, then they are about producing proper software. It is ridiculous that in 2022 (almost 2023) Revit only uses one core from the processor. 

We buy fast computers with expensive graphic cards... all for nothing. 

Stop adding new features / gimmicks, first take care of the basics!! 

What are we paying autodesk else for? 

ps: This is asked by thousands of users for over more then 10 years....

Why is it not happening?? 

[kfandrich]

Making Revit multi-core is my top request! We buy amazing workstations with loads of computing power, and Revit can only use a small fraction of it. Not many people are aware of this situation, so it probably doesn't get many votes, but a truly multi-core Revit would likely improve every aspect of the user experience and enable the program to be far more capable! While you're at it, make a setting so that IT folks could specify how many cores or what percentage of cores Revit would use. This would enable us to tune processor usage and give a better experience for our users. Right now, I turn off hyperthreading in the bios so that the cores aren't further divided up. This seems to improve Revit performance. I also buy the highest mhz I can get and don't worry so much about buying a lot of cores. It really disappoints me that I have users who work on huge models and can only use a fraction of their processor. Imagine how much better the performance could be if they could use 3/4 of their processor instead of 1/8 or 1/6!

Thanks!

[Lorenzo.Taccini.STA]

Please allow multi threading while PDF exporting

[b.kronemanK9DTX]

Curious to hear what the status of this topic is (since Autodesk is working on this since 2021, according to their reply on this topic)? This would improve Revit so much!

[l.wichert]

Significantly accelerate the overall system, including multi-processor core support.

[l.wichert]

Hey people, my idea "Accelerate overal system, make it quicker!" was moved/merged with this thread. Ok but i wanten to keep it simple, without "techic-insider-knowledge"... so to say - form the user-point-of-view. as i am the user (architect) which doesent care HOW it will be quicker! 

So

MAKE-IT-QUICKER!

now

[aizaiwutuobang]

After several years, I'm back here again. In the past two years, there have been significant changes in Revit, especially with the migration to .NET 8. My first thought was whether Revit was undergoing a rejuvenation through a re-coding of its underlying framework. However, since I haven't used it yet, I'm not sure if there have been truly remarkable changes. But I see hope. It seems that Revit is capable of implementing multi-threading applications, which is an achievable task. All it needs is Autodesk's commitment to addressing it! But there's one limitation that has been a constraint for Revit all this time. That is the object limit of 10,000, where imported CAD points and objects cannot exceed 10,000 each. I hope this can also receive attention! Especially when importing points for terrain solid model creation, as the widespread use of Revit will result in an increasing number of terrain points exceeding 10,000.