Re: gEDA-user: How do I really design footprint with pcb?
On Tue, 2010-11-02 at 18:42 -0500, Mark Rages wrote: Sorry, I have no idea about your intention. Maybe DJ has, I have seen he has responded already. Maybe you can make a picture? If making footprints is your problem: For that we have many helpful tools. many helpful tools means that everybody gets to write a footprint generator in their own favorite scripting language. There are no helpful GUI tools, only the unhelpful tool of pcb itself. Footprint generation seems like a GUI task to me, but I'm sure others will disagree. This is not really true. We have different non-GUI like footgen and my sfg, web-based ones like DJ's, the new one with Java-GUI recently mentioned in dev mailing list. And of course we can draw footprints in PCB program, I have never done that, Kai-Martin has, and he has explained ho to do it. ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: How do I really design footprint with pcb?
On Wed, 2010-11-03 at 15:26 +0100, Stefan Salewski wrote: mentioned in dev mailing Sorry, I think it was geda-announce. And maybe this tread is helpful http://archives.seul.org/geda/user/Mar-2009/msg00459.html http://www.penguin.cz/~utx/pstoedit-pcb/ ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Wed, 2010-11-03 at 17:29 +, Peter Clifton wrote: Hello people, I've got a load of changes I've been working on recently in PCB+GL, this time on my local_customisation_no_pours branch. Since long time I am wondering about the terms before pours or no pours. I have still no idea -- is there a special meaning or is that only a arbitrary name like ubuntu's release names. ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Wed, 03 Nov 2010 19:03:16 +0100, Stefan Salewski m...@ssalewski.de wrote: On Wed, 2010-11-03 at 17:29 +, Peter Clifton wrote: Hello people, I've got a load of changes I've been working on recently in PCB+GL, this time on my local_customisation_no_pours branch. Since long time I am wondering about the terms before pours or no pours. I have still no idea -- is there a special meaning or is that only a arbitrary name like ubuntu's release names. Peter C's branch includes experimental support for real copper pours. If I remember correctly. Peter -- Peter Brett pe...@peter-b.co.uk Remote Sensing Research Group Surrey Space Centre ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Wed, 03 Nov 2010 17:29:04 + Peter Clifton pc...@cam.ac.uk wrote: Here is the gotcha.. the VBO code didn't really work on the NVidia machine.. rendering got really slow. If you discover this, you can force it back to using Arrays rather than VBOs with this patch: Peter, Maybe, just by chance, you have the same problem I had (855 chipset). There is a bug going around in the Intel drivers, which was finally solved in the latest kernel, 2.6.35, but even there, a last minute error crept in, and req'd a patch. If this is the case, I can fetch the solution. John ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Wed, 2010-11-03 at 17:51 +, Peter TB Brett wrote: Peter C's branch includes experimental support for real copper pours. If I remember correctly. Peter Ah! http://en.wikipedia.org/wiki/Copper_pour For my restricted English pour was always strong rain. ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Wed, 2010-11-03 at 15:19 -0300, John Coppens wrote: On Wed, 03 Nov 2010 17:29:04 + Peter Clifton pc...@cam.ac.uk wrote: Here is the gotcha.. the VBO code didn't really work on the NVidia machine.. rendering got really slow. If you discover this, you can force it back to using Arrays rather than VBOs with this patch: Peter, Maybe, just by chance, you have the same problem I had (855 chipset). There is a bug going around in the Intel drivers, which was finally solved in the latest kernel, 2.6.35, but even there, a last minute error crept in, and req'd a patch. If this is the case, I can fetch the solution. The performance problem was on an NVidia card (I forget which). I've been over every way I can imagine of using VBOs, and I still get about one frame per second. Using Vertex arrays on that same card manages 20fps. I'm usually developing on a laptop with Intel G45 graphics chipset. (Same 3D drivers as 965+ upwards, but with card specific defines etc..) Anything 945 and prior uses a completely different 3D driver, (covering 915-945) and I also see an i810 driver in the tree too. At this stage, I don't expect my code to work on i8xx, and I'd even be doubtful of it working on 945, as it (at the moment) requires uses pixel shaders. I'd be interested to see the output of glxinfo for your machine though. John ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user -- Peter Clifton Electrical Engineering Division, Engineering Department, University of Cambridge, 9, JJ Thomson Avenue, Cambridge CB3 0FA Tel: +44 (0)7729 980173 - (No signal in the lab!) Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me) ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
Peter Clifton wrote: If VBO rendering slows you down, Ehm, how would I know, that this is the the actual bottle neck? (What is VBO, anyway?) Throw your usual complexity of board at it. Tried it at work. This is a 3.5 years old, moderate hardware. AMD single processor, 4400 BogoMIPS. Graphics card is nvidia NV37GL Quadro PCI-E Series. Graphics driver is the closed source nvidia, 173xx legacy version. Pcb was run in a maximized window on my left 1280x1024 monitor. pd-ac: (a small board) http://bibo.iqo.uni-hannover.de/dokuwiki/doku.php?id=eigenbau:photodiodenverstaerker_20_mhz pcb-head (compiled in August 2010: 35 FPS (thin_draw_poly:44 FPS) pcb-GL_before_pours: 20 FPS pcb_local_customisation_no_pours: 20 FPS pidpeltier: (a more koomplex board) http://bibo.iqo.uni-hannover.de/dokuwiki/doku.php?id=eigenbau:temperaturregler_diodenlaser current debian/squeeze pcb-gtk, version 20091103: 25 FPS pcb-head: 12.5 FPS pcb-GL_before_pours: 16.5 FPS pcb_local_customisation_no_pours: 12 FPS lasertreiber: (My most complex real world board. I usually switch off polygon planes) http://bibo.iqo.uni-hannover.de/dokuwiki/doku.php?id=eigenbau:lasertreiber current debian/squeeze, version 20091103: 17 FPS (thin draw poly 25: FPS) pcb-head: 11 FPS (thin_draw_poly:18 FPS) pcb-GL_before_pours: 9 FPS pcb_local_customisation_no_pours: 10 FPS Looks like v20091103 is consistently the fastest binary on this particular set-up. Any idea, why this is so? ---)kaimartin(--- -- Kai-Martin Knaak tel: +49-511-762-2895 Universität Hannover, Inst. für Quantenoptik fax: +49-511-762-2211 Welfengarten 1, 30167 Hannover http://www.iqo.uni-hannover.de GPG key:http://pgp.mit.edu:11371/pks/lookup?search=Knaak+kmkop=get ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
Peter Clifton wrote: I've got a load of changes I've been working on recently in PCB+GL, this time on my local_customisation_no_pours branch. For those not familiar with git, these are the commands I ran to install Peters version in /usr/local/bin/pcb-test : / git clone git://repo.or.cz/geda-pcb/pcjc2.git cd pcjc2 git checkout -b local_customisation_no_pours ./autogen.sh ./configure --disable-doc --enable-dbus --enable-gl make sudo make install cd /usr/local/bin/ mv pcb pcb-test \ ---)kaimartin(--- -- Kai-Martin Knaak tel: +49-511-762-2895 Universität Hannover, Inst. für Quantenoptik fax: +49-511-762-2211 Welfengarten 1, 30167 Hannover http://www.iqo.uni-hannover.de GPG key:http://pgp.mit.edu:11371/pks/lookup?search=Knaak+kmkop=get ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: KiCad's polygon library
On Thu, 2010-10-28 at 18:44 +0100, Peter Clifton wrote: Speaking of KiCad (and its source), I noticed they've found a GPL polygon boolean operations library which I'd not come across before. There is a fresh post on cairo mailing list: Clipper performs boolean clipping operations (intersection, union, difference xor) on 2D polygons. http://lists.cairographics.org/archives/cairo/2010-November/021050.html http://lists.cairographics.org/archives/cairo/2010-November/021051.html ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Wed, 2010-11-03 at 21:54 +0100, Kai-Martin Knaak wrote: Peter Clifton wrote: If VBO rendering slows you down, Ehm, how would I know, that this is the the actual bottle neck? (What is VBO, anyway?) Vertex buffer object: http://en.wikipedia.org/wiki/Vertex_Buffer_Object Throw your usual complexity of board at it. Tried it at work. This is a 3.5 years old, moderate hardware. AMD single processor, 4400 BogoMIPS. Graphics card is nvidia NV37GL Quadro PCI-E Series. Graphics driver is the closed source nvidia, 173xx legacy version. Pcb was run in a maximized window on my left 1280x1024 monitor. pd-ac: (a small board) http://bibo.iqo.uni-hannover.de/dokuwiki/doku.php?id=eigenbau:photodiodenverstaerker_20_mhz pcb-head (compiled in August 2010: 35 FPS (thin_draw_poly:44 FPS) pcb-GL_before_pours: 20 FPS pcb_local_customisation_no_pours: 20 FPS pidpeltier: (a more koomplex board) http://bibo.iqo.uni-hannover.de/dokuwiki/doku.php?id=eigenbau:temperaturregler_diodenlaser current debian/squeeze pcb-gtk, version 20091103: 25 FPS pcb-head: 12.5 FPS pcb-GL_before_pours: 16.5 FPS pcb_local_customisation_no_pours: 12 FPS lasertreiber: (My most complex real world board. I usually switch off polygon planes) http://bibo.iqo.uni-hannover.de/dokuwiki/doku.php?id=eigenbau:lasertreiber current debian/squeeze, version 20091103: 17 FPS (thin draw poly 25: FPS) pcb-head: 11 FPS (thin_draw_poly:18 FPS) pcb-GL_before_pours: 9 FPS pcb_local_customisation_no_pours: 10 FPS Looks like v20091103 is consistently the fastest binary on this particular set-up. Any idea, why this is so? Hmm, that is interesting. Just note that the stock PCB's thindraw poly is faster, but the PCB+GL one will actually increase the workload due to the fancy translucency it does in that mode. It is interesting to note that git HEAD PCB is slower than 20091103. I wonder what I broke ;) (There might be some performance trade-offs which have been made to improve other activities). I might have to dig into that, as it is quite a worrying regression on some of the boards. -- Peter Clifton Electrical Engineering Division, Engineering Department, University of Cambridge, 9, JJ Thomson Avenue, Cambridge CB3 0FA Tel: +44 (0)7729 980173 - (No signal in the lab!) Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me) ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: Ben mode feature request
John Griessen wrote: That is a standard feature of the usual chip design software Not only chip design software. Many mechanical design applications provide this feature, too. autocad, varicad, ustation are just those, I got in touch with. ---)kaimartin(~~~--- -- Kai-Martin Knaak Öffentlicher PGP-Schlüssel: http://pgp.mit.edu:11371/pks/lookup?op=getsearch=0x6C0B9F53 ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Wed, 2010-11-03 at 22:04 +0100, Kai-Martin Knaak wrote: Peter Clifton wrote: I've got a load of changes I've been working on recently in PCB+GL, this time on my local_customisation_no_pours branch. For those not familiar with git, these are the commands I ran to install Peters version in /usr/local/bin/pcb-test : / git clone git://repo.or.cz/geda-pcb/pcjc2.git cd pcjc2 git checkout -b local_customisation_no_pours Did this work and give you the 3D / GL stuff? (Does git automatically pick the right remote branch?) I would have done: git checkout -b local_customisation_no_pours origin/local_customisation_no_pours But since I learnt to do that some while back, things might have got easier since. -- Peter Clifton Electrical Engineering Division, Engineering Department, University of Cambridge, 9, JJ Thomson Avenue, Cambridge CB3 0FA Tel: +44 (0)7729 980173 - (No signal in the lab!) Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me) ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Thu, 2010-11-04 at 00:48 +, Peter Clifton wrote: It is interesting to note that git HEAD PCB is slower than 20091103. I wonder what I broke ;) (There might be some performance trade-offs which have been made to improve other activities). I might have to dig into that, as it is quite a worrying regression on some of the boards. I've just read over the git log and can't see anything which jumps out at me. Perhaps some potential changes in performance on polygon operations, but those aren't happening during benchmarking anyway, the NoHoles polygons (are / should be) cached. I'd be fascinated to see results of a git bisect between the pcb-20091103 tag and git HEAD to see where the performance dropped. pidpeltier looks like the board with the most dramatic difference between the versions. If you fancied sending me any of the boards off-list, I'll have a play and see if I can make PCB+GL faster with them ;) -- Peter Clifton Electrical Engineering Division, Engineering Department, University of Cambridge, 9, JJ Thomson Avenue, Cambridge CB3 0FA Tel: +44 (0)7729 980173 - (No signal in the lab!) Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me) ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Wed, 2010-11-03 at 21:54 +0100, Kai-Martin Knaak wrote: Peter Clifton wrote: If VBO rendering slows you down, Ehm, how would I know, that this is the the actual bottle neck? (What is VBO, anyway?) And to answer your other question... you don't know unless you apply the patch in my earlier email, something like adding buffer-local = true; If that makes things faster, the VBO code was causing problems. The VBO code seems fine on my laptop, but I just can't make it work fast on my own NVidia. (With whatever latest driver Ubuntu installs for it). I was getting 1fps vs. 20fps on the same board when I encountered the problem. -- Peter Clifton Electrical Engineering Division, Engineering Department, University of Cambridge, 9, JJ Thomson Avenue, Cambridge CB3 0FA Tel: +44 (0)7729 980173 - (No signal in the lab!) Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me) ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
Kai-Martin Knaak wrote: Now for the test at home: This is recently bought moderate hardware. AMD dual core, 1050 BogoMIPS. Graphic card: ATI Radeon Sapphire HD5450. Driver is fglrx. pcb was again maximized to 1280x1024 on the left screen. I only tried the lasertreiber.pcb layout: version 20091103: 24 FPS (thin-draw-poly: 16 FPS) pcb-head: 13 FPS (thin_draw_poly:19 FPS) pcb-before-pours: 19 (thin draw poly: 15.5) pcb-local-customization-no-pours: 18 (thin draw poly: 16) This is quite a bit faster than my hardware at work. About twice the speed except for pcb-head which gained only 2 FPS. Now, I am eager to test my shiny used Radeon HD4670... ---)kaimartin(--- -- Kai-Martin Knaak Öffentlicher PGP-Schlüssel: http://pgp.mit.edu:11371/pks/lookup?op=getsearch=0x6C0B9F53 ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
Hi, On Wed, 2010-11-03 at 17:29 +, Peter Clifton wrote: Throw the most complex usually-hideously-slow board at it.. how does it perform? I've run the benchmark on the most complex board that I've currently designed in PCB. Run on a AMD Phenom 9750 (4822 bogomips) with nVidia GeForce 9400GT using the latest binary driver fullscreen at 1280x1024: Design available at: git://srobo.org/boards/power-hw.git Photo renderings: https://www.studentrobotics.org/~rbarlow/pcbs/power/ PCB head: 16.3FPS PCB before_pours: 25.2FPS PCB local_customisation_no_pours (VBO): 85.6FPS PCB local_customisation_no_pours (array): 93.1FPS All benchmarks were performed three times and averaged. I have compiz enabled on this computer and tried running the VBO local_customisation_no_pours branch with it disabled which gave an average frame rate of 89.4FPS. It seems that compiz doesn't affect things too much. Richard ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Thu, 2010-11-04 at 02:34 +0100, kai-martin knaak wrote: Kai-Martin Knaak wrote: Now for the test at home: This is recently bought moderate hardware. AMD dual core, 1050 BogoMIPS. Graphic card: ATI Radeon Sapphire HD5450. Driver is fglrx. pcb was again maximized to 1280x1024 on the left screen. On the nvidia (or whatever you fancy), see if you get a change in performance with the attached patch.. or the one I sent with the original email to disable the VBO usage. Your full-screen FPS numbers aren't that bad, but I'm sure we can do better eventually ;) Now, I am eager to test my shiny used Radeon HD4670... I'm taking bets that you'll not see huge improvements. I'd imagine we'll be CPU bound before much more framerate can be extracted. I'll have to send you a patch / testcase to benchmark _without_ any drawing calls, to determine the CPU limit on frame-rate. Also.. try shrinking the PCB window to the smallest possible size.. does that increase framerate much? If you know how for your drivers (I don't), try disabling sync to vblank. -- Peter Clifton Electrical Engineering Division, Engineering Department, University of Cambridge, 9, JJ Thomson Avenue, Cambridge CB3 0FA Tel: +44 (0)7729 980173 - (No signal in the lab!) Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me) diff --git a/src/hid/common/hidgl.c b/src/hid/common/hidgl.c index 3ab77c6..aad818c 100644 --- a/src/hid/common/hidgl.c +++ b/src/hid/common/hidgl.c @@ -145,14 +145,14 @@ hidgl_flush_triangles (triangle_buffer *buffer) buffer-triangle_array = NULL; } - glEnableClientState (GL_VERTEX_ARRAY); glVertexPointer (3, GL_FLOAT, 5 * sizeof (GLfloat), buffer-local ? buffer-triangle_array : BUF_OFFSET (0)); - glEnableClientState (GL_TEXTURE_COORD_ARRAY); glTexCoordPointer (2, GL_FLOAT, 5 * sizeof (GLfloat), buffer-local ? buffer-triangle_array + 3 : BUF_OFFSET (3)); + glEnableClientState (GL_VERTEX_ARRAY); + glEnableClientState (GL_TEXTURE_COORD_ARRAY); glDrawArrays (GL_TRIANGLE_STRIP, 0, buffer-vertex_count); glDisableClientState (GL_VERTEX_ARRAY); glDisableClientState (GL_TEXTURE_COORD_ARRAY); ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Thu, 2010-11-04 at 01:50 +, Richard Barlow wrote: PCB head: 16.3FPS PCB before_pours: 25.2FPS PCB local_customisation_no_pours (VBO): 85.6FPS PCB local_customisation_no_pours (array): 93.1FPS All benchmarks were performed three times and averaged. That as the kind of change I was _hoping_ to see ;) (Although I've no idea why arrays is faster than VBOs.. it ought not to be if I've got the code right!) Perhaps try with the patch I just sent in reply to KMK which moves a couple of glEnableClientState calls. I might have to work hard to keep those state changes to a minimum, as it certainly seems Nvidia think those calls can be expensive. (And others have reported weird performance drops due to glDisableClientState calls). I have compiz enabled on this computer and tried running the VBO local_customisation_no_pours branch with it disabled which gave an average frame rate of 89.4FPS. It seems that compiz doesn't affect things too much. No, compiz is surprisingly light on GPU usage. I think the BLITTER engines do most of the work when rendering non-translucent non-transformed windows. Might be wrong there though. Whatever though.. compiz should just amount to an extra buffer copy, perhaps with a blend operation. You might get a slight improvement without compiz, but it is typically minimal. I usually test with compiz off just to be sure though. -- Peter Clifton Electrical Engineering Division, Engineering Department, University of Cambridge, 9, JJ Thomson Avenue, Cambridge CB3 0FA Tel: +44 (0)7729 980173 - (No signal in the lab!) Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me) ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Thu, 2010-11-04 at 02:01 +, Peter Clifton wrote: Perhaps try with the patch I just sent in reply to KMK which moves a couple of glEnableClientState calls. Using VBOs that gives 85.4FPS, so effectively no difference. Using arrays it gives an average of 93.1FPS, literally no difference! Richard ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
Re: gEDA-user: PCB+GL Testers (please test)
On Thu, 2010-11-04 at 02:13 +, Richard Barlow wrote: On Thu, 2010-11-04 at 02:01 +, Peter Clifton wrote: Perhaps try with the patch I just sent in reply to KMK which moves a couple of glEnableClientState calls. Using VBOs that gives 85.4FPS, so effectively no difference. Using arrays it gives an average of 93.1FPS, literally no difference! Thanks for trying it. I'm embarassed to say your board manages about 9fps on my machine, with every trick I can throw at it! The board is an interesting test-case though, as I don't typically work with grouped layers. It might be something which triggers a performance hit. The solder layers render relatively fast on their own, (~30fps), as do the pads on their own (~60fps). But if I switch both on, I get ~5 fps. 1/30 = 33ms 1/60 = 17ms Assuming the overheads just sum, that is per 50ms frame with both enabled. That should give me approx 20fps, not 5! The more sub-composited layers involved, the more slow glClear calls are made. I'm not sure this accounts for the difference though, so I'll investigate. It has got me quite curious now! -- Peter Clifton Electrical Engineering Division, Engineering Department, University of Cambridge, 9, JJ Thomson Avenue, Cambridge CB3 0FA Tel: +44 (0)7729 980173 - (No signal in the lab!) Tel: +44 (0)1223 748328 - (Shared lab phone, ask for me) ___ geda-user mailing list geda-user@moria.seul.org http://www.seul.org/cgi-bin/mailman/listinfo/geda-user