At 10:55 AM 1/13/2004 -0500, Elizabeth Martinez wrote: >( How does a better video card improve GIS performance ?)
A GIS can spend a lot of its time drawing and redrawing images, some of which are quite detailed. A good example consists of head's-up digitizing, where the screen is constantly refreshed, perhaps once after each point is drawn--potentially several times a second. Any observable delay will slow down the entire work flow.
Several years ago, using AV 3.x (I think it was 3.0 or 3.1 at the time), I formally benchmarked 2D video performance on vector and raster redraws. For large-screen redraws of images, a medium-end card (Fire GL) could speed redraws by a factor of four, typically. At the time, that meant several seconds per redraw: a huge difference. I have to believe similar effects would be observed with other GISes.
Currently, cards overall have better capabilities, so the improvement in redraw speed obtainable from a video card upgrade may not be quite as dramatic, but it's definitely something to consider.
Note that 2D performance is not completely related to 3D performance: you have to ignore all the 3D hype and specs and look at a card's 2D benchmarks when comparing them.
The principle underlying these considerations is that a computing system is not a CPU alone. (Indeed, the CPU nowadays is itself a complex interacting network of quasi-independent processors and caches.) It is an interacting system of processors (CPU, video, disk, sound, and others), RAM, and peripheral devices sharing one or more buses and, perhaps, caches. Inferior performance of just one element can create a bottleneck that greatly impedes the entire system performance. For instance, you can probably get better performance for many GIS tasks from an old Pentium 3 with a great video card and disk array than with a multiple Xeon 3.2 GHz multi gigabyte RAM machine having an old crummy video card and slow disk installed (not that anyone would be that silly...).
Since the original question concerned buying a new machine, I will sign off with another principle: the high-end hardware you buy today is tomorrow's junk. (I have an entire room full of it going back only 11 years.) In one year it will be old: in three years it will practically be junk: in five years it will be an antique. So do not buy for your potential needs next year. If your software is not written for multiple processors, don't buy MP on the basis that your software will be multi-threaded "Real Soon Now" (RSN). If your GIS does not support 3D visualization now, don't buy a high-end 3D video card just because it will RSN. If your software is not compiled for 64 bit processing, don't buy a 64 bit chip because of the RSN promise.
Cheers, Bill Huber
_______________________________________________ gislist mailing list gislist@lists.geocomm.com http://lists.geocomm.com/mailman/listinfo/gislist
_________________________________ This list is brought to you by The GeoCommunity http://www.geocomm.com/
Get Access to the latest GIS & Geospatial Industry RFPs and bids http://www.geobids.com
|