Modeling benchmark for OSDM

[clausb]

Hi all,

encouraged by the support of so many folks here in the forum for the "viewbench" graphics benchmark, I asked myself whether it would make sense to also have a public modeling benchmark for OSDM, and whether people out there might already have written code which they would want to contribute. (Of course we have internal benchmark code here at CoCreate, but it contains a very large number of tests which most people won't really be interested in, and also partially depends on our internal environment.)

Any ideas/pointers/suggestions?

Claus

[May Kung]

By public benchmark, do you mean one that is readily available to the general public that does not require OSD to run?

I've only tried using the Viewspec benchmarks, Viewbench, or a standard set of models that I time for loadtime and update time.

[clausb]

May,

I was thinking of a benchmark which runs inside OSDM and executes modeling functionality.

Claus

[stefano_ME30]

Claus

could it help putting clipping test also?

I saw performance drop in the better hardware

Stefano

[clausb]

Originally posted by stefano_ME30
could it help putting clipping test also?


Thanks for the proposal.

Clipping tests belong into viewbench, IMHO, not into a modeling benchmark as discussed here. I'll make a mental note to look at adding such a test the next time I make a major revision of viewbench.

Claus

[pkehoe]

Many years ago I created a recorder file while creating the
"Frontplate" part in the old training material. I've had to
update it based on changes to the syntax. I also changed
it from doing a layout at the end to doing some clipping.
And as computers have gotten faster I found that the
variability on the same computer was almost as much as
between computers, so I run through the recorder 5
times to increase the time to magnify the differences
between computers.