On Mon, 21 Dec 2015 10:10:06 -0700 (MST) Mekko <serve@perdix.org> wrote: > If its performance is bound by disk seek-latency, wouldn't an SSD (a.k.a. > "flash drive") improve performance considerably? They're getting cheap these > days... Not a lot. An SSD has much better latency but the difference between RAM and even a top end SSD is several orders of magnitude (think information on your desk versus having to go into another room on a different floor to get it. In fact in most cases think versus having to get into the car and drive somewhere ot get it). More RAM will help if the model will fit better. What actually matters is what is called the "working set" - which is the pieces of information needed a lot. That's often smaller than the full memory needed - which is why swap works at all. I use 16GB with OpenSCAD and the box was purposely chosen so I could shove 16GB of RAM into it as some of my models are quite complex. I still have a few models that OpenSCAD can't do a good job on, for those I use other tools. For the rest OpenSCAD has the nice UI and is just easier to work with. Alan

My system has triple raided ssd on the c drive for speed. I do think profiling is in order.

Interesting. FWIW, here are some figures of merit for seek latencies: HDD is about 10ms SDD is about .1ms RAM (not cache) is about .0001ms If the workload is disk latency bound, a 100x improvement going from HDD to SSD might help a lot (though clearly not as much as the 100,000x jump going to RAM). Maybe the CPU has chip envy. Har har har. -- View this message in context: http://forum.openscad.org/Test-compile-big-openSCAD-model-tp15226p15268.html Sent from the OpenSCAD mailing list archive at Nabble.com.

Basically swapping on Windows doesn't work. It always brings the system to its knees. You need to have enough physical memory for all the programs or it becomes un-responsive for long periods. While it can schedule the CPU and share it between processes it seems unable to share the disk and the driver seems able to lock out everything else while shovelling 100's of megabytes in and out of the swap space. On 21 December 2015 at 17:43, david vanhorn <kc6ete@gmail.com> wrote: > My system has triple raided ssd on the c drive for speed. I do think > profiling is in order. > > _______________________________________________ > OpenSCAD mailing list > Discuss@lists.openscad.org > http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org > >

nophead wrote > Basically swapping on Windows doesn't work. It always brings the system to > its knees. You need to have enough physical memory for all the programs or > it becomes un-responsive for long periods. While it can schedule the CPU > and share it between processes it seems unable to share the disk and the > driver seems able to lock out everything else while shovelling 100's of > megabytes in and out of the swap space. > > On 21 December 2015 at 17:43, david vanhorn &lt; > kc6ete@ > &gt; wrote: > >> My system has triple raided ssd on the c drive for speed. I do think >> profiling is in order. >> >> _______________________________________________ >> OpenSCAD mailing list >> > Discuss@.openscad >> http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >> >> > > _______________________________________________ > OpenSCAD mailing list > Discuss@.openscad > http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org My observation is that as the page fault interrupts get served before anything else, when you have a chronically growing memory gobbler it starts to bottleneck on the page file, which also gets high I/O priority, thus delaying other I/O, such as system logs etc, thus clogging up everything. Prior to getting a SSD this would cripple my system to the extent that the mouse would lag and even a Ctrl-Alt-Del would take many seconds to get handled. Now with a SSD it is manageable. ---------------------------------------------- My Big Box results from yesterday: ... ECHO: "processed 37100" ECHO: "processed 37200" Rendering Polygon Mesh using CGAL... Geometries in cache: 74403 Geometry cache size in bytes: 54164656 CGAL Polyhedrons in cache: 53 CGAL cache size in bytes: 104371776 Total rendering time: 0 hours, 27 minutes, 59 seconds <<== Bug, was 10 hours Top level object is a 3D object: Simple: no Vertices: 1244484 Halfedges: 3943810 Edges: 1971905 Halffacets: 1384044 Facets: 692022 Volumes: 2341 WARNING: Object may not be a valid 2-manifold and may need repair! Rendering finished. Memory -------------------- Commit Size 21.745GB Peak WS 15.894GB <<== hence it fit on my 16GB system with little swapping Export -------------------- WARNING: Object may not be a valid 2-manifold and may need repair! See http://en.wikibooks.org/wiki/OpenSCAD_User_Manual/STL_Import_and_Export WARNING: Exported object may not be a valid 2-manifold and may need repair Error: Non-manifold triangle mesh created: 19 unconnected edges STL export finished. ---------------------- I tried to upload the zipped STL, but I think it didn't work. Let me know if anyone wants it. Netfabb Basic couldn't open it. Meshlab did: <http://forum.openscad.org/file/n15271/td15226_STL_Meshlab.jpg> <http://forum.openscad.org/file/n15271/td15226_STL_Meshlab_closeup.jpg> <http://forum.openscad.org/file/n15271/td15226_STL_Meshlab_non_manifold.jpg> ----- Newly minted Admin - PM me if you need anything, or if I've done something stupid... Unless specifically shown otherwise above, my contribution is in the Public Domain; to the extent possible under law, I have waived all copyright and related or neighbouring rights to this work. Obviously inclusion of works of previous authors is not included in the above. The TPP is no simple “trade agreement.” Fight it! http://www.ourfairdeal.org/ time is running out! -- View this message in context: http://forum.openscad.org/Test-compile-big-openSCAD-model-tp15226p15271.html Sent from the OpenSCAD mailing list archive at Nabble.com.