Chris Lohman

Setting up the boot.ini with both switches is a VERY good idea! Actually the 3GB switch does more than just free up 3GB; it also exposes the threshold at which the OS will release the memory. From our (exhaustive) testing on this matter we found that with the default 2gb limit, IRONCAD would hit the limit and then go nuts/crash. With the 3GB limit, IRONCAD would actually use about 2.4 gigs of memory and then suddenly the memory usage would jump back down to something like 500 megs and then climb back up to 2.4, then recycle back to 500 over and over. The developer explained to me that after IRONCAD uses memory and is finished with the data in memory (which occurs very quickly), IRONCAD marks the data has being ready for reuse. If you're in the middle of an operation though, the operating system doesn't get around to freeing up that marked memory however until it gets up into that higher range. (Once the operation finishes then it flushes the memory) So what I found was that while the default 2GB could only handle something like 400 parts (it also depends very heavily on the complexity of the parts); the 3GB would reuse the memory more efficiently and could thus handle several thousand parts. Of course, once you're using those high amounts of memory, you really should have that much RAM as well. Once you hit the pagefile, IRONCAD is going to slow down by 3-500% (because your pagefile is on your hard disk and your disk is that much slower than ram). So while the 3GB limit may prevent crashing, it may also slow ironcad down to a crawl (because you're using pagefile). So with something like 4000 parts, the recommendation would still be the /3GB limit (because the OS won't let us go higher) but it would also be "ideal" to have 4 gigs of ram (3 for ironcad, 1 for the OS) for optimum speed. And of course (I have to say this), since you're using an operating system that we do not test on nor do we certify will run IronCAD (that being XP Home), everything that we ever say about IronCAD and the OS can not be guaranteed to apply to XP Home. (Honestly though, XP Home is just missing the domain networking stuff in XP Pro so XP Home *should be fine* for running IronCAD {and I've had many people tell me that it -is- fine}. We just can't say that it is officially certified for use with IronCAD) id=size1>id='Arial'>

Hey Everyone, We're discussing creating some additional tech support offerings (at no additional cost) and were wondering what you would think of the following ideas. We're discussing creating: Some tutorials that cover a key design area; such as how to apply draft to a plastic part or techniques for creating fluid shapes, etc. These are not generic tutorials like our getting started guide; these would be more specific tutorials on how to create specific geometry (like adding ribs to a shelled plastic case). Even if you don't create that specific geometry, do you think you could still learn and benefit from tutorials on specific advanced geometry creation and would you like for us to create them? Do you have any specific topics in mind that you would like to see a tutorial on? (Aside from surface, which we realize NEEDS tutorials) A quarterly tech support newsletter. This newsletter would send out tips, tricks, tutorials, and other ironcad tech related information. We could even put some API examples in the techletter. Can you think of anything else that our support department could produce for you that you believe would be very beneficial to you as an ironcad user? What about a user gallery on this site; which would allow you to post renderings of your work. We would have to moderate incoming images of course; but once approved your work/name could be published into a gallery of user-generated images. If we had this, would you even be able to post your renderings, or is your work primarily confidential? Would you take the time to render a scene every few weeks (or months) in order to post it?

Ooh Jesse, I never thought about using a loft with associated faces for a stretchy animation. That's a very cool idea! Here's an example of what Jhahn was speaking of. It's two cylinders that are both shapes under the same part; then between the cylinders is a 2 profile loft that has both profiles associated with the inner faces of the cylinders. I then created a little custom smart motion path for the top cylinder and exported it and voila, a stretchy/boingy part. I initially would have suggested using the overlapping parts; and using the overlapping method will significantly reduce the export time and headaches with invalid geometry. But using this loft method does product a very interesting result. When using the loft method though; every frame of the animation requires a regeneration of the parts: so the export took a very long time and my first 3 attempts actually crashed because the geometry was too complex. Attached is an animated gif and a scene file for the stretchy animation . stretchy.gif stretchy.ics

This info has not yet been published, but we are very close to doing so. We should have a very high level list of stuff for you within the next couple of week (VERY high level, meaning you won't get to see some of the really cool little things that were added yet; but that will be published eventually). We're also looking towards releasing an alpha version for everyone with a support contract to play with within the next 4-6 weeks. As always; please don't consider any dates as exact or features as certain. Anything that we say about when or what features for 7.0 are approximations at this stage. We try to nail a certain release date but dates get pushed when bugs are found at the last minute. You should expect to see the high level list and release date within the next 2 or 3 weeks.

If you upload everything that it outputs, then it should all work. The html files use dynamic referencing in order to link to itself and down; so no matter where you put them they should still work. Allowing some other application to "Import" them and take control of them probably wont work; I would just use regular FTP to post all contents into a folder on a website.

The limitation is imposed by the operating system. Any single process is only allowed to "allocate" up to 2 gigs of total memory. Unfortunately you can't see the allocated blocks in Task manager correctly, only "used"; so task manager may say that it has only used 1.4 gigs of mem but the app has already allocated all that it is allowed and thus it crashes. You might be able to workaround this by bumping your allocatable limit up to 3 gigs (which is the highest option that is available). To set the 3GB switch on your OS: Edit your boot.ini so that it has the /3GB switch in it like my example has below: [boot loader] timeout=30 default=multi(0)disk(0)rdisk(0)partition(1)\WINDOWS [operating systems] multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="Microsoft Windows XP Professional" /fastdetect /3GB Please be aware that if you mis-edit the boot.ini that your system will not reboot properly. If you make a mistake then you'll have to boot on your recovery CD and create a new boot.ini; which is an exhaustive and painful process. Insure that you have all of the latest OS patches installed before using this boot.ini switch. Note that this /3GB switch is mentioned at the bottom of the Readme.txt which resides on the IRONCAD cd under \README\IRONCAD. You can also read more about the switch here: http://www.microsoft.com/whdc/hwdev/platform/server/PAE/PAEmem.mspx

Don we had a version of that boolean/linkage stuff in older revs but it was causing catastrophic failures in the models so we had to pull it out to prevent mass data loss. But even though it was pulled, it's not forgotten. It is such an increadibly complex feature for us to build (and keep easy) that we will have to put much more work into it to insure that it could never cause problems. So we probably won't see that added back into 7.0; but there remains hope for us to have that functionality in the future.

Ooh Mike...good call; nicely done.

Oh and David to answer your initial question; if you cut a negative shape using a boolean; aftercutting you delete your initial cutting tool, and then at a later date you need to get that tool back: Your only option is to perform another boolean in order to rebuild your tool. Unfortunately once you get into boolean world you lose some of the most flexible features of IRONCAD; the intellishapes: So editing that rebuilt tool must be done at the profile handle/profile level, or by adding positive or negative intellishapes back onto the cutting tool.

David what I think Mike is saying is that you model the negative using negative shapes to begin with (which is a good recommendation), not a positive shape that you somehow switch to a negative. See the attached for an example of said suggestion which contains your original part, and then a version of it that I rebuilt using intellishapes. Expand the parts in the scene browser to see the differences. Below is an explanation of the avoiding booleans statement. Booleans are far more mathematically complex than intellishapes are. So the more booleans you use; the more difficult the geometry will be to process; and thus the longer it will take to process. Booleans also create geometry that is not as flexible as an intellishape: So when you use a boolean you: 1) Lose the flexibility of IRONCAD's intellishapes: Which is a loss of one of the major features that makes IRONCAD so easy to use. (You still have the profile handles and profile editing though, so all is not lost after a boolean) 2) You could kill your processor IF your booleans are significant. David in your simple example; a boolean isn't a big deal for the processor to use. You lose the flexibility of the intellishapes; but you can just as easily go back and edit the original cutting tool and then re-boolean the part in your simple example. So using boolean or not really becomes an issue once your geometry grows to a significant (enough) level of complexity. Think of a plastic cavity with ribs and bosses...you could model the cavity, ribs, and bosses, in 3D and then just cut them from a block using boolean: Or you cut the cavity in the block using intellishapes (or a shell) and then build the ribs and bosses into it using regular positive intellishapes. In my cavity example, that single boolean could define almost all of the geometry in your part; which would make it difficult to go back to and edit and would also make it very very CPU heavy. The intellishapes calculate much faster and they're easier to edit than a boolean so you could win twice by modeling in the negative to begin with. Of course it's often very difficult to visualize and model a negative shape: So ultimately you should be using IRONCAD to work as fast and as productive as is possible for you. So if you can model in the positive and boolean that shape much faster than working with negative/intellishapes; and you don't think that you will need to make many (if any) design changes and it isn't killing your CPU: Then please use IRONCAD in a manner that makes your job the easiest and most productive. example2.ics

Sometimes when my shape is buried, I'll have to select it in the scene browser and then drag it from the scene. You are correct in finding that you can only drag it from the scene. Sometimes if my shape is very buried I'll have to suppress other shapes to get to it. In your situation David, if the catalogs are difficult for you, there is another way around this. You can duplicate the entire part (copy/paste) and then in the second part you can delete all of the shapes except for the one(s) that you want to keep in their own part structure.

I've actually reported that test to nVidia; it turns out that the software doesn't use card settings in a standard mode or something (it was developer speak that went a little over my head). nVidia recommended that we all use the industry accepted standard test (which I read about in almost every hardware review out there), which is SPEC Perfview, since those people actually work on insuring that everything they use is standard and it works for all cards. I just haven't been able to find the free time to use that test and come up with a template for all of us to follow. I'll try to get to it soon though. It's not that the Quadro's are slow; it's that the test app isn't using their hardware.

Ah yeah good 1 Joel. Just to ammend what Joel said; it could also be a surface smoothness at the part level. Your global smoothness could be low but your part could be high. Part Properties/Rendering...

Can anyone help me answer this question? I don't know any architects... When creating an architectural drawing; when displaying the VIEW SCALE units; could the units be displayed as anything other than "Major unit - Minor unit"? For example 3/32" = 1' 0" or 1 1/4" = 1' 0" If the scale was exactly 12 inches, could it be displayed as 1' instead of 1' 0" ? Can the view scale ever be displayed in some other format, like "inches on paper" = "inches" or "inches on paper" = "yards", etc. on an Architectural drawing? Metric? test.zip

Some possibilities: 1) Your card can't handle the volume of data in your scene. The only solution here is to upgrade your card or suppress data. (Data = Parts/Cameras/Lights) 2) You've recently installed a graphics driver that is going nuts. 3) Try hitting [ctrl]+8 or [ctrl]+9 (try both) and rotate afterwords. This forces opengl or software rendering. 4) Is it a problem with a scene that contains only a single block? 5) You could just be out of system resources. Try rebooting?

IronCAD lights are not capable of volumetric light rendering, which is the name for what you are asking.

Ah hehe :-) This is very disturbing. I'm running a Quadro FX 3000 on Windows 2003 Server (with hardware acceleration enabled) and it's only reporting 47 fps. It "must" be a problem with the fact that the drivers are not officially supported on Win2K3 or -something-. This just isn't right. ...going back to windows xp...

wow, 4300 is so high that I'm experiencing difficulty even comprehending it.

5700? not 570? Is it that high when the window is active (IE you have clicked on it) or when the window is inactive? OpenGL fps is falsly reported as MUCH higher when the window is inactive. Where are you guys getting the EdgeVBM test from? I only found a single hit for it on google.com. The test had a rotating male face in shaded and wireframe format; is that what you're running?

lol scary

The vmem that is recommended by Microsoft is to have the same amount of vmem as you do ram. So in your case that would be one gig. One theory about increasing vmem speed is to set your minimum to 15 megs below your maximum. By doing this the operating system pre-allocates your disk space and thus when you need to use virtual memory; you only need to write into the pre-allocated vmem block instead of also building the block as you require it. This used to make a big difference back when Windows NT 4 was the latest OS and 5400rpm disks were amazing. There is no such thing as good performance with virtual memory though. If you need virtual memory then you would be MUCH better off by purchasing more ram. If you are using ironcad you can expect to see a speed reduction of at least 50% once you start consuming virtual memory.

Thanks

IronCAD doesn't render part edges during rotation by default; so any graphics card will produce the same disappearing edges during rotation. Shall I file an enhancement request on your behalf for part edges to be drawn during rotation? Insure you have instructed IronCAD to always use opengl through tools/options/rendering (uncheck automatic, check opengl [not camera only]). Upgrading a graphics card will show the most improvement in scene camera movements (orbit, pan, zoom). It's when moving the camera that we hit the graphics card. A stationary render in realistic mode uses the software pipe for rendering (Not the graphics card's processors). Exporting a rendering uses the software pipe for rendering. I have seen that inside out flipping but I've never intentionally reproduced it. I'm glad that you mentioned that as a result of zooming because now I may be able to produce a bug report from it. One other note with regards to the graphics card settings. While I can't recommend any settings that will enhance performance; I have found that if you setup these newer cards that allow you to "hard code" the antialiasing and anisotropic filtering; that if you set them to 4/8 (highest levels) by default for all apps; that this will cause the ironcad window to ghost itself in a wierd way when you exit the application. So just in case you notice that suddenly when you exit ironcad; the window does some wierd shrinking/ghosting/jumping before it closes; you'll know which setting did that.

Perhaps it's a problem with saving to your external usb hard drive. Perhaps you don't have enough space in your %temp% disk for saving large data. We have never tested nor officially certified an external USB hard drive for saving to/from IronCAD. It would probably work but; it is an unknown factor. I have grown into the habit of saving sets of files with incremental revisions for my own work. It's not a big deal with fast scsi drives until my files get into the 20-30 meg range and by then I have linked externally to bring my file sizes back down.

We install the latest stable kernel whenever we release a major rev. We heavily test the latest kernels before our next release to insure that they are stable. We usually find that the latest kernel is highly unstable. By the time the kernel people fix all of the bugs that we find in their kernel; we have usually released by then with whatever is the most recent "rock solid" kernel.