Hi Michael ..
The subject line says it all. I'm just beginning to anticipate. My fingers are tickling.
eric

:)

> My fingers are tickling...

Tickle who? ;-)

Hi eric, I've been wondering when someone was going to ask about that! :)

Yes, I've been working on v3 for the last while here and a new beta is nearly ready.

The new things are - completely overhauled PDF import to be more robust (previously some kinds of PDFs would get garbled), new Extrude to point and Extrude tapered options, new Isocurve command for building an isocurve along a surface, and a new Isocurve option for Trim to cut a surface by isocurve.

The last thing that I'm trying to get done is possibly using multiple CPU cores in the display engine, I've been experimenting with that the past couple of days and it will probably be a few more days yet before I'll see if that will actually yield anything good or not, once that is done then I'll be releasing a new v3 beta.

- Michael

> using multiple CPU cores in the display engine...

Whoa!!! =-)


CPU Cores: How would that work. Say for instance, I have 2 cores on my home PC and 8 on the one at work.
Would MoI take the current view your in and divide it's workload on multiple cores, or would MoI take each view and give it it's own core?

IsoCurves: Awesome!!! ...I think. :-/

Would this be the same as Projecting a line to form a new surface curve around the object.
Or would there actually be some kind of switchable on/off feature to the iso line?

Could you do things like rebuild a surface?

I know for one thing... Rhino's view of it's objects look really cluttered and bad with them, maybe there's some good uses for them.
So they're like the control-point grid, but on the surface?

Hi Mike,

> Would MoI take the current view your in and divide it's workload on multiple cores, or
> would MoI take each view and give it it's own core?

It's about dividing up the workload for just one view to speed up the display. But I'm not sure yet if it's going to actually work ok or not, it has to synchronize things up when it actually talks to the video card and too much synchronization can cripple multi-core performance. I won't really know for sure if it's going to work well until it's actually up and running. I'll probably have an initial piece working tonight and then I'll be able to see if it's actually going to be better or worse. It is actually possible for multi-core code to make things worse if the different cores spend too much time bottlenecked at synchronization points.

But it's something that I've wanted to try for quite a while now so it is cool to be in experimentation mode right now.


> Would this be the same as Projecting a line to form a new surface curve around the object.

Yeah it's similar to that - an isocurve basically one of the U or V curves that come from the inbuilt surface structure. They're pretty convenient to use with certain kinds of surfaces, like on a sphere a U isocurve will be a latitude line and a V isocurve will be a longitude line so they are convenient ways to get at those curves on a sphere. Similarly if you have a fillet surface, one of the U or V directions of that fillet will be arc curves if you want to cut the fillet at it's sort of natural boundary the isocurve trim will be useful for that.

> Could you do things like rebuild a surface?

It's usually for trimming, but yeah say you do a loft between 3 curves then using isocurves you could extract out a bunch of new curves on the surface in between those, on a loft surface one of the U/V directions will be like the profile sections.

- Michael

Hi Mike,

> I know for one thing... Rhino's view of it's objects look really cluttered and
> bad with them, maybe there's some good uses for them.

The Rhino isocurve display thing is a side effect of Rhino being designed primarily for wireframe display mode, since I originally made it on a Pentium 90 machine way back when.


> So they're like the control-point grid, but on the surface?

Yeah, each isocurve is a 3D-space version of a horizontal or vertical 2D line in the UV space of the surface. Since a NURBS surface has a regular row/column grid layout to it, there is a kind of 2D rectangle built in to each surface as part of the surface definition.

- Michael

Before August ?

Hi Pilou,

> Before August ?

Yup, it will probably be ready sometime next week, at least that's what it looks like right now. I often times mess up time estimates though! ;)

- Michael

Cool for the camping :)

Michael - If you need a 2nd quick validation on the multi-core before you release, I've got an i7 3930k 6 core cpu.

Ed

Michael!

I just thought of something about the multi-core display...

Does this have anything to do with MoI's "Insufficient Memory" errors I get when I export object to an .obj?
Funny how I get those errors and the export fails making it's high-count poly's, yet Kerkythea and apps like it boast bazillion-poly capabilities.

Maybe it's something else causing the occasional IM errors (Red on yellow type). I generally have to hide objects and try the export on the rest with more conservative values.


Extremely high-res polys are sometimes overkill, but when I'm working on a close-up or glass/refraction objects for that matter, poly-number is key.

I also suspect that it could be the fact that I've only got a gig of ram in my home PC. :-/

Hi Mike, yeah the multi-core use for mesh export can tend to make a greater spike in the amount of memory used at any one point in time, since for example processing 6 things in parallel requires 6 times the amount of memory used all at once.

There are a couple of things that I have on my list to tune-up on that export to try and conserve some memory.

If you're running into it often you can limit the number of cores to be used as described here: http://moi3d.com/forum/index.php?webtag=MOI&msg=5124.16

That won't really be as much of an issue for the display since generating an export mesh on a big surface like a long tube or something requires a whole lot more memory to process than the display mechanism needs to use.

- Michael

""""""""""Funny how I get those errors and the export fails making it's high-count poly's, yet Kerkythea and apps like it boast bazillion-poly capabilities.""""""""""

It would be a more relevant comparison if Kerkeythia could import a 500 mb "CAD" model, then ask IT to dice it up into a bazillion little polygons, then compare the results....

I see your point Burr, MoI has a remarkable tessellation engine.

But, I figured it was more of a poly/memory thing that was causing the error.
Most likely it was the fault of my flea-powered home PC. ;-)

Yeah while it is in the process of dicing MoI is actually dicing up NURBS surfaces, they only become polygons in the last stages. Also on top of that trim curves are diced up into small pieces as well to make trim boundaries. That whole process takes up a bunch of additional overhead to sort of "set the stage" for stuff before the polygons are even created. It's usually not the polygon data itself that really takes the most space, more all that in between stage calculation stuff.

When you're in the polygon app there is none of that other stuff existing at that point, only the final result, that's why the capacity is different.

- Michael

Also yes "flea powered" does not make a great natural combination with "high density" as well... ;)

If you have a lower powered machine it just goes with the territory that you need to avoid higher complexity stuff.

- Michael

Sounds great Michael.... Bottoms up.

Peace,
Rich_Art. ;-)

Looking forward to testing the extrude to point and taper!

Michael T.

Looks like it will be generally in the range of a 2x to 3x drawing speedup (depends on different aspects of the model) with the multi-core use!

I was hoping for maybe a straight 4x speedup on a quad-core but it looks like that won't happen because of too much waiting for the graphics driver to just process all the incoming requests. So probably more cores beyond 4 or so cores won't give much boost. But despite that a 3x speed-up is nothing to sneeze at! :)

I need to test it a bit more and also see if I can wring a little bit more yet out before releasing it, so a few more days yet...

- Michael