Do it real test for SW2007 or SW2006 on your workstation !!!

We don't need the 3D Mark points, this is the better way to understand that
SW is still in 1999!
Do it like I do:

1. Sketch rectangle 2000x1000 mm (left corner in origin cordiantes 0,0,0)
2. Extrude 1 mm or sheet metal plate
3. On plate sketch 2 equal circle: 1. diam. 4 mm, cordinate (5; 5)
2. diam 4 mm, cordinate
(8; 10.2 )
4. Click on Tools-Linear step and repeat on: x-axis 332 times on distance 6
y-axis 96
times on distance 10.2
5. ENTER & WAI(S)T YOUR TIME!

Best result till now:

MacPro Dual Xeon-5 GB RAM ON WIN XP 32-bit...............1h53min(OS
recognize only 2 GB RAM-a, constant use of each processor )
MacPro Dual Xeon-6 GB RAM ON WIN XP 64-bit...............2h5min(OS recognize
6 GB RAM-a, only one processor active ??, the second is inactive all the
time)
AMD Athlon 2800+ 64-bit on WIN XP 64-bit.........................> 3 hour,
waist time
Pentium IV 3200 64-bit on WIN XP 64-bit..............................> 3
hour, waist time

This is the example of each day on my work. I need better result, people.
Can anyone do better time?
I'm waiting also the result on SW2008?

Keep work the good job

Davor

 

OK, how many dipshits can you pack into a single newsgroup?

This guy has a lot of time to waist.

Comparing macpros agains old pcs? If all you model is plates with holes
in it, it sounds like a greta benchmark.

Daisy.

 

Not sure what you mean with the total time but,.. on my old P4 3.0mhz,
2 gigs, FX500.. using SW2006sp5.1
(my ram usage peaked at 1 gig so, there is no need for more ram)

~1hr8min to solve the pattern (blue, sketch still open)
~25min to exit or save the sketch (grey, sketch closed)
or,..~1hr33min total

Will try it on SW2008 sp0 later..

...

 

Correction,.. (Dammit!) it for some reason it did not pattern the
second circle sketch?? So, I'll guess (factor of 2X) that this is
then on par with your values?

...

 

Some testing was done a couple years ago (by Matt Lombard, I think, and
probably TOP too) on the problem of large patterns. Sketch-based patterns
(step and repeat) rebuild the slowest. If the pattern allows for it, make
original holes in the plate, then pattern the feature. Geometry patterns
will probably be slower. There have also been reports of improving
rebuild times by making patterns of patterns.

So, my recommendation would be to make a feature pattern of the original
holes, make a pattern of them, then make a pattern of the pattern. It
still won't be fast, but should will be better. You might also examine
whether it is necessary for all of the holes to be modeled, or if a few
holes and a note on the drawing will be sufficient.

 

There is also a new feature in SW 2008 called cosmetic pattern...to
show holes like this in the drawing without actually making them. Need
to use shaded views and real view turned on.

// Krister

 

Actually,

The cosmetic Pattern that was presented in Alpha and Beta is not what
was implimented in release of 2008. The idea was to do a cosmetic
pattern and have it show thru to drawings as an annotation but that
was dropped. the part of it that survived the release is a cosmetic
pattern that is essentially a RealView material that can be applied
parts and faces.

 

Hmmm....the other parts seems to have survived in the help file. I
remember playin with it in Beta and just checked yesterday before I
answered here that it still was there.

// Krister

 

I can't help you with better hardware, but maybe I can help with a
different way to make it.

Yes, doing it all in a sketch pattern is a great way to take an extra
long lunch and catch up with friends. Additionally, in my experience,
patterns of patterns is a great way to take a long break (check email,
read 'the onion', etc)

I had a patterning job a few months back that required thousands of
instanced patterns (this is where having two computers comes in handy
- while one crunches, work on another job on the second) I wish I had
clocked it - all I know is I gave up on both basic approaches early on
- sketch pattern (immediately) and then feature pattern after a try or
two.

The only way that worked for this gig in anything approaching a
reasonable time (and the pattern was along curved surfaces which kicks
Deejays problem up a notch) was to model one complete instance
(solid, holes, fillets, etc), then pattern the body that was created,
then merge the final bodies. It still took a decent chunk of time
time, but at least it was do-able. Before this I spent a few hours
with other approaches and almost 'no-quoted' the job because SWx could
not handle it.

In full-diclosure mode, I patterned a series of bodies that described
10% or so of the whole, merged them, then patterned those new bodies
and merged them. That seemed to give me the best result-to-annoyance
ratio of all.

On the planar pattern, if you do this again, I would suggest you do a
corner of your pattern as a body, and then pattern that body (in one
or two steps) to fill up your 2000mmx1000mm area.

***And, of course, ask yourself if you actualy NEED to model a true
pattern or if you can represent it on a drawing and have your vendor
do it***

Hope this helps. I know that patterning bodies instead of features
(dumb things instead of smart things) has helped me save time multiple
times on my path to the same final model.

Ed