Does Rhino 5 benefit from double or triple video cards?

Does Rhino 5 benefit from dual or triple video cards hooked up together in one machine. a computer expert tells me no, but i wanted to get a second opinion. Thanks

I think there is something about that on the Wiki pages.
Did you look?

Found it:
http://wiki.mcneel.com/rhino/twodisplayadapters

Thanks for the link. The answer YES but:(1) the video cards must be exactly the same,(2) they must support SLI,(3) they must be physically connected together using the SLI bridge adapter,(4) and they must be running on a motherboard that supports SLI/Crossfire.

Given all of that, the drivers will allow you to enable SLI support, and Rhino will work just fine across multiple video cards…-mcneel wiki

So it can be done.

Apparently so but I have no experience with it.
I can’t imagine it would have much effect on Rhino performance considering the cost of the setup.
It is useful information if you have a more demanding OpenGL graphics primary application that does benefit and want to know if Rhino is OK on the same computer.

A hardware expert told me that on games that supported SLI it would improve the performance by 20-30%.

I am interested in concept and development of detailed models that are BIG Files, so the subject is important to me.

A computer is like a 3 dimensional box that one works in, the more powerful the computer the bigger that box is, and thus the more detailed your models can be, it seems, to me. Of course when one gets down to brass tacks, this statement is overly abstracted to be of much use to anyone.

As far as I know all Rhino files are composed of: points, lines, surfaces, materials, and meshes, and some text as well: I think that about sums it up, maybe I missed a few things. Therefore it seems like ‘File Size’ is a good measure for determining hardware, because all the files are made of the same things, and everyone is apply the same sorts analysis and physics tools on their designs. So using ‘File Size’ to determine computer size seems like a valid relationship to me. Please correct me if I am wrong.

It may be but that assumes the files are all made of the same approximate percentages of the objects types and that is not universally true.

What I’m concerned about is you making the assumption that graphics performance is the bottleneck in your process. I suspect it’s not so a uber fast graphics setup will not likely yield the kinds of performance increase you think it will.

It’s much like commuting to work. Buying a super fast car isn’t going to reduce the time you spend driving to and from work because the top-end speed of the car is not what’s slowing you down.

If you spend a bunch of money on this dual/triple card set up and it’s no faster than what you have, you’re going to be pissed off at us.

If it were me and I was convinced I needed fast graphics, I would follow Jeff’s advice on the system requirements page and get an AMD FirePro v7900 card.
Since I don’t need super fast graphics, I’m running an Nvidia GeForce GT 750M and very happy with it for my needs.

Hi Justin,

I’m not totally following your thoughts on “file size to determine computer size” but wanted to offer an opinion on SLI. I tried it once and it had zero effect that I could see. If you are trying to speed up the Rhino display with large models. You should send in the model to tech@mcneel.com by way of rhino3d.com/upload we might be able to help determine where the slow down is occuring. It is often the case that render meshes can be customized, meshes joined or bad objects repaired to speed the display. Just offering a bit of caution before investing in more hardware.

I understand what your talking about I already took a so called basic to advanced training course to learn how to use rhino, to rebuild curves, and repair meshes, and close holes in the model to make sure it was a ‘solid’, and build anything ones imagination could conceive of. In my opinion it was only a basic to intermediate course, because it was too artistic and didn’t the delve into the construction and manipulation of complex polyhedron, used almost no math at all, and made no use of scripting at all. But it was still a good course, and I learned a lot.

basically one always ends up with a bunch surfaces, then may convert them to meshes.

I know my current laptop, a Thinkpad T400s with 4GB RAM, dual core 2.53GH processors; can handle file sizes up to about 15MB, then its starts to get unworkable. Primarily in the use of the camera. it seems like in theory if one used a strictly text based input method, that made no use of Pan, Zoom, and Rotate of the camera, and did not use the mouse at all. Then one could work with files around 150MB on this machine. I have done this I can load up big files of this size, but I cannot work on the file using the camera, its too big for this machine

IDEA 1: So their may be another great idea for Rhino 6: is to build out the text based input so one can work with big files, in every way, without using the graphical bottleneck that most computer seem to have. One of the great things a bout textually inputting models, is the sense of personal gratification one gets from knowing he can express his design in complete verbal detail if necessary, one feels like a true expert. Using the mouse and camera I always feel like a kid playing with play-dough and sticks.

anyways So the Primary bottleneck appears to be GRAPHICS.

Its very easy to test this: just load up some stuff that composes a design and use the array command to blow it up to a big file size till your computer becomes ‘stalled’ and unusable.

IDEA 2: Their is free unfinished plugin called ‘Yeti’ on food4Rhino that aims to allow for complete textual input based design modeling, that could be very useful for those who push the limits of their computer, and want to feel like super-experts.

IDEA 3: Another idea that would probably be necessary for this subject would be a ‘simplified camera’ that does not offer totally fluid Pan, Zoom and Rotate, but instead uses a ‘snap to’ options, Thus not requiring so much graphical power.

IDEA 4: Computer Games for years have used engines that automatically drawn only sections of a 3d environment, rather than trying to draw everything at once. Installing optimization tools something like that into rhino environment would be good, allowing people to work with much larger files on less powerful computers.

With Computer chip technology hitting a wall scientifically, good software design needs to pick up the slack.

Since it seems graphics are the number one bottleneck in CAD it seems like more SLI graphics cards is better for performance on large files. But that doesn’t mean I would buy 3 SLI NVIDIA® GeForce® GTX TITAN’s 6GB’s, that would be a $7000.00 machine. Its just a matter of deciding what computational constraints one is going to put themselves under. It seem to me that one cannot use the graphical interface, without a lot of RAM, virtual memory and a powerful graphics card.

In Sum. File Size is a very useful number and a lack of Graphical power from a good video card, and a lack of Memory to hold the large file seems to be the big bottle necks in CAD computing. Please correct me if I am wrong.

You may not of had the SLI setup correctly. Also did compare and contrast on big files?

Justin:
You seem to be making the assumption that a faster graphics setup will translate into faster Rhino display performance. That is a bad assumption.

Rhino is designed to work well on inexpensive, mostly off-the-shelf hardware. That means it had acceptable performance for users with inexpensive computers.
It also means that we have not take advantage of any of the performance features available in higher versions of the OpenGL specification (Rhino V5 uses OpenGL 2.0 and Shader 1.2), and no support for brand specific tools.

I see so your saying their is already a built in ceiling when it comes to building large file sizes.

apparently OpenGL 4.4 is the newest version

according to your own wiki, rhino will work across multiple video cards as long as they are the same and connected correctly.

But now you saying that this ability to work across multiple video cards does not equate into better graphical performance.

Does McNeel plan to build out the performance of the software in Rhino 6 so that it benefits from multiple graphics cards and utilizes available performance feature on newer technology?

No. I’m saying that Rhino V5 uses OpenGL specification 2.0 and takes no advantage of any of the improvements in newer specifications. My guess is V6 will probably move up the OpenGL specification a bit but certainly not to the latest. That would kill a lot of existing Rhino customers that can’t afford to replace their hardware. Even moving away from 2.0 will be a problem for lots of schools.

Yes, it will “work” but don’t read into that that it will translate into the performance gains that the dual card SLI was designed to do. Your assumption that Brian had connected his incorrectly was not a very likely scenario.

You asked:
Does McNeel plan to build out the performance of the software in Rhino 6 so that it benefits from multiple graphics cards and utilizes available performance feature on newer technology?

No. Our business plan is to continue to make an excellent surface modeling tool that will run well on average computer hardware. That way Rhino is affordable to a large number of professional and academic users. If Rhino required fancy hardware, our current sales and development model would put us out of business.

Alias is an excellent NURBS based surface modeling tool that does take advantage of high-end equipment. Have you priced Alias, the annual maintenance fees, and the required hardware to run it?

1 Like

Hi John, I strongly believe that Rhino would benefit from a dual approach. An OpenGL 2 pipeline (“just” use V5’s) for the medium hardware and also a “pro” high end version that can feed the fastest cards.

Quite a lot of designers, architects and ship builders are willing to buy K5000 cards just to speed up Rhino when working on massive and complex scenes. And that is why you upgraded Rhino 5 to support 64 bit OS with lots of RAM.

Thus I built Holomark2, to figure out if and where they would benefit from such an investment.

Rhino can run in software, and you did build a DirectX pipeline once, so having multiple OpenGL versions “is” an option. I also strongly believe that you could sell the pro OpenGL pipeline as a plugin to cover the cost.

That said: Nobody needs a graphic card that feeds a 60Hz monitor with much more than 60fps. So it all depends on the complexity of the scene one is working with.

1 Like

I could be wrong, but don’t think having DirectX, low-spec OpenGL, and high-spec OpenGL display code streams to develop, maintain, and keep in sync would translate into more copies of Rhino sold. If it wouldn’t sell more Rhinos, how do we pay for it? Diverting more developer resource to it would just cost more, potentially translating into higher prices, and slow the development of other tools that everyone benefits from. This development could increase the price of Rhino even if you didn’t benefit from it (like tools you don’t use), or do you think the small number of users willing to buy high end cards would also pay $10k (number pulled out of the air), for a Rhino that would use the fancy cards?

They are making 120hz and 240hz monitors now.

I was making no assumptions. Just fact finding. I see your business model is deeply flawed. Your pushing customers out the door instead of drawing them in that’s always a bad thing. Your software can be learned on a $200.00 laptop computer. But when one plans to make money working in CAD, they buy a better machine to accommodate there career, this software is way to complex to be used idley, people who take the time to learn this stuff plan to make money using it. Therefore software needs to take advantage of hardware as much as possible.

You need to keep your customers. if your software works for student using old school computers, building nothing at all, but doesn’t work well for the professional who are building something, like me, then its useless and all student who are wise will skip you and go to autodesk or some other brand offering better hardware compatibility.

Currently your cheaper price and greater intitial accesibility is what drew me here. Autodesk is way toooo big and expensive, they are offering so many software packages at double the price you are asking. They are not a good fit for me. They seem geared towards large firms with fairly large budgets of some sort, with well paid professional Autodesk experts building skyscrapers and city infrastructure and expensive automobiles.

If you can keep your price point about the same you will retain your customers better. It is not possible to promise more sales of Rhino but, every person as far as you are concerned is a computer, therefore to reach the audience you need to reach out to those computers and give them what they want.

Computer users may be mostly made up of mediocre machines. But I am sure there are very few serious CAD users anywhere working on mediocre machines, unless they are confident they will never be building anything bigger than a 20MB file. CAD users are power users. So your missing your audience by focusing on non-professionals.

And of course the secret to all business modeling is the Golden Rule(Mt 7:12). Don’t give people something you wouldn’t be willing to accept. Therefore one can see that you are very close to being unacceptable to professionals. Your audience is the professional.

Oh by the way. thanks for your honesty.

So I won’t bother with a dual Video card setup. Maybe in the future we’ll see better hardware support.

Your an expert So here is a hardware configuration below that I want to use to build Files around 6-7GB. Do you believe it will work?

	Quiet Fan Pack
	SilverStone® 500 Watt 
	Asus® Z87 Plus 
	Intel® Core™ i7 4770K 3.5GHz 
	Asetek® Liquid Cooling
	No Processor Overclock 
	1866MHz 4x8GB (32GB) 
	NVIDIA® GeForce® GTX 780 Ti 3GB 
	On-Board Audio 
        On-Board Ethernet 
	**Crucial® M550 SSD 512GB**
	Western Digital® Black™ 3TB 
	Asus® 16x DVD Writer 
	Microsoft® Windows 8.1 

EDIT: I forgot to add the 512 SSD that was also a part of the original spec.
EDIT 2: the manufacturers of this machine have assured me over and over again that this machine will be able make easy work of my demanding 6-7GB file sizes. However according to this forum I fear the problem may come from the software side of things, not the hardware.

That looks like a nice setup. I haven’t used the GF GTX 780 Ti personally.

If I remember correctly, over half of all Rhino seats are educational licenses, and that’s counting 30-seat lab licenses as 1. This hasn’t changed much for many years. I rely on Bob for the business plan and strategies. I’ve been here for 27 years; since before the Rhino project started. It’s been a great ride and I have no plans to throw in with anyone else.

Be well

1 Like

I’m using a Dual Xeon engine and my bigger files are around 1GB. During Vray rendering the RAM usage can be quite high, 10 GB or more for Rhino only. Often I’m working on one big file, but I need to open other big files at the same time too. In this moments my 32 GB are not enough some times. Your file size sounds extreme large. Best you don’t fill all RAM slots so you can build in more later.

Graphic … I use a Quadro 6000 and I’m not happy with the Rhino performance. If my model display slows down, than the card stay at low GPU usage, the bottleneck is the Rhino display pipeline. So I must say for Pro user Rhino can be a problem. Only I can hope more people will have the same problems, report it and the display pipeline will enhanced for the pro needs. My current CPU/GPU power is very high, but Rhino is power resistant and using a few percent only. :wink:

So, the answer to your question “will it work?” could be “no”, but not because the hardware. If you want work with complex NURBS models you could get big problems. If you use a lot of mesh objects than the GPU is better used and complex scenes are not so much a problem.

PS: I’m not sure about your choice, but your OS should be 64bit.

So meshes work better performance wise. interesting. I have no problem using meshes instead of regular breps, it doesn’t matter to me.

everything on my spec is 64bit. well anyone unhappy with the rhino display pipeline probably just went elsewhere without saying anything.

So rhino needs to utilize the hardware better…

what do you build with rhino?