kicad-developers team mailing list archive

[Nick Østergaard <oe.nick@xxxxxxxxx>]

.2016-03-05 0:22 GMT+01:00 Mário Luzeiro <mrluzeiro@xxxxx>:
> Hi NickOe,
>
> tanks for testing it!
>
> The issues you describe I didn't experienced it yet, would you like to try a latest version?
> I am testing on a linux machine and on a windows machine.

I was testing 5909 Fix an issue on raytracing it was not reset a bbox.
of lp:~mrluzeiro/kicad/kicad_new3d-viewer

This is on archlinux x86_64.

I will retest with 5912.

>
> What CPU do you have?

Intel(R) Core(TM) i5 CPU       M 540  @ 2.53GHz

That is two cores, and I have enabled hyperthreading.

>
> On Raytracing, complex boards will take about 10seconds per CPU core just for rendering, it means if you have more cores it should take less. (depending also on the resolution)
>
> Regarding the load times, I was expecting the same or faster (it should be faster in the future even more).
>

Currently I am mostly concerned about the time that I open the
3dviewer untill I can see my board (in opengl).

>
>> I have also noticed that there are not holes rendered in the solder
>> paste in the opengl view.
>
> (I believe you mean solder mask ?)

On a second though, I mean solder paste and solder mask actually. :)

> Since I received lots of requests (you and Maurice :P)
> I implemented that feature (on solder mask) in my latest commit.
>
> I was not very clear on if / how should I implemented the holes of the vias.

The old viewer shows holes in the solder make and paste at least.

> Since there is no information to remove solder mask, that in the fab. will be "tented" vias.
> I leave it as discussion.Maybe I can add an option to remove it or not... but at moment it is by default implemented.
>
> There is an issue with that.. this is only work (at moment?) with THT vias.
> So it will not remove if it is a "micro via" between B_Cu/F_Cu and a inner layer.

I am not too concerned about the microvias, personally. My concern was
only on though hole pads such and the normal vias.

I see your implementaiton of the holes in the mask now. But it looks
like the plating tube of the hole is smaller than the hole, is that
how it was intended?

In trying to test the render speed, I tried to change an option in the
of the mask, it takes about 19 seconds, measured with a stopwatch
manually. It looks like only one thread is execution when looking at
the CPU usage in top. It is on 100% and the other three are below 15%
as when idling.

A slight bug I noted that, I saw that the pivot point is reset after a
setting change to the viewer.

>
> Hope advanced users that develop multilayer boards dont request it much! :P

On a sidenote; what was your plan on actually cutting board artworks
such as silk and pads and such that was outside of the board edge? If
implemented, I hope this will happen sometime in the future, then it
should be an option that can be enabled or disabled for the viewer.

> Let me know when you try the latest updates...
>
> Mario Luzeiro
> ________________________________________
> From: Nick Østergaard [oe.nick@xxxxxxxxx]
> Sent: 04 March 2016 18:23
> To: Mário Luzeiro
> Cc: easyw; kicad-developers@xxxxxxxxxxxxxxxxxxx
> Subject: Re: [Kicad-developers] Some new sweet features on the new 3D-Viewer
>
> Hi Mario
>
> I just tried your branch the other day, but there the raytracing did
> not seem work properly at all. It was not rendering anything sometimes
> and when it did it was one of the very low res views. I don't remeber
> the revno, have you see this before? (I have not tried the latest of
> today yet, in case you already fixed it).
>
> I have a desing in similar complexity as that board and maybe a bit
> less, but it takes 17 seconds to load the board in the 3d view,
> whereas in the 3d viewer in 4.0.2 it takes 3 seconds for it to render
> in the 3d viewer.
>
> I have also noticed that there are not holes rendered in the solder
> paste in the opengl view.